Applied Statistics

Applied
Statistics
(U.P. Unified & C.C.S. University Syllabus)
(for IIIrd year B.A./B.Sc. Statistics Students of All Colleges affiliated to
Universities in Uttar Pradesh)
By
Dr. Arun Kumar

Reader & Head
Dep't. of Statistics
M.M. (P.G.) College, Modinagar (U.P.)
Dr. Alka Chaudhary Dr. Qaim Akbar Rizvi

Reader Senior Lecturer
Dep't. of Statistics Dep't. of Statistics
Meerut College, Meerut (U.P.) Shia P.G. College, Lucknow (U.P.)
KRISHNA Prakashan Media (P) Ltd.

KRISHNA HOUSE, 11, Shivaji Road, Meerut-250 001 (U.P.), India
Jai Shri Radhey Shyam
Dedicated
to
Lord
Krishna
Authors & Publishers
A bout the Book
pplied Statistics is an up-to-date reference book for B.A/
B.Sc. (Statistics) IIIrd year Students of Uttar Pradesh.
The key features are:

1. Covers 100% syllabus of B.A/B.Sc. (U.P. Unified & C.C.S.
University).
2. A large number of solved problems have been listed chapter-
wise for better and easier understanding.
3. Numerous tables and diagrams are presented to rouse the
interest of students and also to present the content
effectively.
4. The book aims at:
(i) Accuracy
(ii) Clarity
(iii) Simplicity.
(vi)
Preface
his book on Applied Statistics has been specially
written to meet the requirements of unified syllabus of
B.A. and B.Sc. Part-III students of all the universities of Uttar
Pradesh. The subject matter has been discussed in such a
simple way that the students will find no difficulty to
understand it. The proofs of various theorems and examples
have been given with minute details. Each chapter of this book
contains complete theory and a fairly large number of solved
examples. Sufficient problems have also been selected from
various university examination paper.
To improve the scope of the book Objective type questions

viz. Multiple choice questions, True/False and Fill in the
blanks have also been added. As usual, recent question paper
of other universities have also been included in the last with
hints and solutions.
We have tried our best to keep the book free from misprints.
The authors shall be grateful to the readers who point out errors
and omissions which, inspite of all care, might have been there.
The authors will feel amply rewarded if the book serves the
purpose for which it is meant. Suggestions for the improvement
of the book are always welcome.
— Authors
(vii)
Acknowledgement
e are grateful to the participants of a numerous
execution development programmes for providing
the stimulus to write this book. We acknowledge our
indebtedness to a large number of books and articles which
we have consulted during the compilation of this book.
Special thanks to Mr. Satyendra Kumar Rastogi

(Managing Director) Krishna Prakashan Media (P) Ltd.,
for this support. We also thank Mr. Sugam Rastogi
(Executive Director), Mrs Kanupriya Rastogi (Director) for
an excellent job in editing the manuscript and entire K.P.M.
team for publishing and bringing out the book in this shape.
Last but not the least, we deeply express our thanks to the
readers. Suggestions and criticism would be highly appreciated
and acknowledged.
— Authors
(viii)
U.P. Unified Syllabus
Applied Statistics
B.A/B.Sc rdrd
B.Sc IIIIII Year
Year Paper-II
Paper-II (w.e.f.
(w.e.f. 2013-2014)
2014-2015)
Unit-I
Time series, its different components, illustrations, additive and multiplicative models,
determination of trend, growth curves, analysis of seasonal fluctuations, construction of
seasonal indices. Idea of Correlogram & periodogram.
Index number – its definition, application of index number, price relative and quantity or
volume relatives, link and chain relative, problem involved in computation of index
number, use of averages, simple aggregative and weighted average method. Laspeyre’s,
Paashe’s and Fisher’s index number, time and factor reversal tests of index numbers,
Unit-II
Educational Statistics: Scaling procedures – scaling of test items, test scores, rating of
qualitative answers and judgements. Test theory, linear models, parallel tests, true score,
reliability and validity of tests. Tetra-choric, bi-serial and point bi-serial correlation
coefficients.
Unit-III
Demographic methods : Sources of demographic data – census, register, ad-hoc survey,
hospital records, demographic profiles of Indian Censuses. Measurement of mortality,
crude death rates, age specific death rates, infant mortality rates, death rate by cause.
Measurement of fertility – crude birth rate, general fertility rate, age-specific birth rate,
total fertility rate, gross reproduction rate, net reproduction rate, standardized death
rates, age pyramid of sex composition , other measures of fertility. Logistic curve fitting
and its use in population projection. Complete life table, its main features and
construction. Official Statistics CSO, NSSO and census organizations their set-up and
functions.
Unit-IV
Control charts for variables and attributes, modified control charts, group control charts,
CUSUM charts, V mask. Sampling inspection by attributes – single and double sampling
plans. Producer’s and consumer’s risk, OC, ASN, ATI functions AOQL and LTPD of
sampling plans. Sampling inspection by variables – simple cases.
(ix)
(x)
C.C.S. University Syllabus
Applied Statistics
B.A/B.Sc IIIrd Year Paper-II (w.e.f. 2013-2014)
Unit-I
Time series its different components, illustrations, additive and multiplicative models,
determination of trend-graphic, semi-average, least square and moving average methods,
measures of seasonal variation-simple average, ratio to moving average, ration to trend,
link related method.
Unit-II
Index number its definition, application of index number, price relative and quantity or
volume relatives, link and chain relative, problem involved in computation of index
number, use of averages, simple aggregative and weighted average method. Laspeyre's,
Paashe's and Fisher's index number, time and factor reversal tests of index numbers,
consumer price index
Unit-III
Demographic methods : Sources of demographic data census, register, ad-hoc survey,
hospital records, demographic profiles of Indian Censuses. Measurement of mortality,
crude death rates, age specific death rates, infant mortality rates. Measurement of fertility
crude birth rate, general fertility rate, age-specific birth rate, total fertility rate, gross and
net reproduction rate. Standardized death rates. Complete life table, its main features
and construction (Abridged life table).
Unit-IV
Control charts for variables and attributes. Sampling inspection by attributes single and
double sampling plans. Producer's and consumer's risk, OC, ASN, ATI functions AOQL
and LTPD of sampling plans. Sampling inspection by variables simple cases.
(x)
(x)
Metrization Theorems and Paracompactness 1
B rief C ontents
Dedication .................................................................................................................................... (v)
About the Book.............................................................................................................................(vi)
Preface..............................................................................................................................(vii)
Acknowledgement..............................................................................................................(viii)
Syllabus (U.P. Unified & C.C.S. University Meerut)......................................................... (ix-x)
Brief Contents....................................................................................................................(xi)
Detailed Contents.........................................................................................................(xii-xvi)
* Sequence of contents is as per U.P. Unified Syllabus
Chapter 1: Time Series.........................................................................................(01-48)
Chapter 2: Index Numbers..................................................................................(49-98)
Chapter 3: Statistical Methods for Psychology & Education

Statistics.............................................................................................(99-114)
Chapter 4: Demographic Methods.................................................................(115-166)
Chapter 5: Statistical Quality Control........................................................(167-220)
(xi)
D etailed C ontents
Chapter 1: Time Series ................................................................................... (01-48)

1.1 Introduction 03
1.2 Definitions 03
1.3 Applications of Time Series Analysis 04
1.4 Components of a Time Series 05
1.4.1 Trend 06
1.4.2 Seasonal Variations 06
1.4.3 Cyclical Fluctuations 07
1.4.4 Irregular or Random Fluctuations 08
1.5 Analysis of Time Series 09
1.6 Measurement of Trend 09
1.6.1 By Inspection or Graphic Method 09
Ø Solved Examples 10
1.6.2 Method of Semi-Averages 11
1.6.3 Method of Moving Averages 12
1.6.4 Method of Curve Fitting 15
1.6.5 Some Other Growth Curves for Trend Measurement 25
1.6.6 Merits and Demerits of the Method of Least Squares 26
1.6.7 Conclusion 26
1.7 Measurement of Seasonal Variations 26
1.7.1 Method of Simple Averages 27
1.7.2 Ratio to Moving Average Method 29
1.7.3 Ratio to Trend Method 31
1.7.4 Link-Relative Method 33
1.8 Residual Method for Isolation of Cyclic Variations 35
Ø Exercise 37
Ø Objective Type Questions 41
Ø Answers 48
(xii)
12 Topology
Chapter 2 : Index Numbers.............................................................................(49-98)

2.1 Introduction 49
2.2 Characteristics of Index Numbers 49
2.3 Uses of Index Numbers 50
2.4 Points to be Considered in the Construction of Index Numbers 50
2.5 Types of Index Numbers 52
2.5.1 Price Index Numbers 52
2.5.2 Quantity Index Numbers 53
2.5.3 Value Index Numbers 53
2.5.4 Special Purpose Index Numbers 53

2.6 Methods of Constructing Index Numbers 55

2.6.1 Unweighted (Simple) Index Number 55
2.6.2 Weighted Index Numbers 61
2.7 Test for Index Numbers 65

2.7.1 Time Reversal Test 65
2.7.2 Factor Reversal Test 66
2.7.3 Circular Test 67
2.8 Quantity Index Number 71

2.9 Value Index Number 71
2.10 Chain Based Index 72
2.11 Relative Merits and Demerits of Chain Base an Fixed Base Method 72
2.12 Base Conversion 73
2.12.1 From Fixed Base to Chain Base 73
2.12.2 From Chain Base to Fixed Base 74
2.13 Cost of Living Index Numbers 76

2.14 Construction of Cost of Living Index Numbers 76
2.14.1 Method of Aggregates 77
2.14.2 Method of Weighted Relatives or Method of Family Budget 77

Ø Miscellaneous Examples 79
2.15 Limitation of Index Numbers 84

Ø Exercise 85
Ø Answers 97
(xiii)
Chapter 3 : Statistical Methods for Psychology

& Education Statistics................................................................(99-114)
3.1 Introduction 101
3.2 Scaling 101
3.3 Some Scaling Procedures 102
3.4 Linear Model of Test Theory 104
3.5 Reliability 105
3.6 Methods of Estimating Test Reliability 106
3.7 Validity 107
3.7.1 How to Estimate the Validity of A Test 108
3.7.2 Types of Validity 108
3.8 Parallel Tests 108
3.9 Point-Biserial and Biserial Correlation Coefficients 110
3.9.1 Tetra Choric r 111
3.9.2 Limitation 111
3.9.3 Tetra Choric r in Detail 111
Ø Exercise 113
Ø Answers 114
Chapter 4 : Demographic Methods........................................................... (115-166)

4.2 Uses of Vital Statistics 117
4.3 Collection of Vital Statistics 118
4.4 Rates 119
4.5 Fertility Measurements 119
4.6 Factors Affecting Fertility Variation 119
4.7 Crude Birth Rate 119
4.8 General Fertility Rate 120
4.9 Age-Specific Fertility Rate 120
4.10 Total Fertility Rate 121
4.11 Mortality Measurement 121
4.11.1 Crude Death Rate 121
4.11.2 Specific Death Rate 122
4.11.3 Infant Mortality Rate 123
4.11.4 Standardised Death Rate 123

4.12 Measurement of Population Growth 129
(xiv)
14 Topology
4.12.1 Crude Rate of Natsural Increase and Pearle's Vital Index 129
4.12.2 Demerit 129
4.12.3 Gross Reproduction Rate 129
4.12.4 Net Reproduction Rate 130
4.13 Stationary and Stable Populations 133
4.13.1 Difference between Stationary and Stable Population 133
4.14 Life Table or Mortality Table 133
4.15 Assumptions of a Life Table 134
4.15.1 Different Elements of a Life-Table 134
4.15.2 Additional Elements of a Life Table 135
4.16 Construction of a Life Table 135
4.17 Uses of a Life Table 136
4.18 Estimates of the Various Functions of the Life Table and their
Inter-Relationship 140
4.19 Statistical Organisation in India 141
4.20 Central Statistical Office (C.S.O.) 142
4.21 National Sample Survey Office (NSSO) 143
4.22 Governing Council 144
4.23 Working Groups (WG) 145
4.24 Socio-Economic Surveys 146
4.24.1 th
NSS 57 Round 146
4.24.2 NSS 58 th Round 147
4.24.3 NSS 59 th Round 147
4.24.4 NSS 60 th Round 147
4.24.5 NSS 61st Round 148
4.25 Sarvekshana, NSSO Bulletin 148
4.26 United Nations World Food Programme (UNWFP) 148
4.27 Agricultural Statistics 149
4.28 Price Data Collection 150
4.29 Urban Frame Survey 151
4.30 Industrial Statistics 151
4.31 Annual Survey of Industries 152
4.32 Labour Bureau 153
4.33 Army Statistical Organisation (ASO) 154
4.34 Census of India Organisation 154
4.35 Some Non-Government Statistical Organisations 155
4.36 Statistical Organisation in States 156
4.37 Statistical Organisation in U.P. and Uttaranchal 156
Ø Exercise 157
Ø Answers 165
(xv)
Chapter 5 : Statistical Quality Control........................................................(167-220)

5.2 What is Quality ? 170
5.3 How Quality is Measured ? 170
5.3.1 By Numerical Measurement 170
5.3.2 By Attributal Measurement 170
5.4 What is Quality Control ? 170

5.5 Statistical Quality Control (S.Q.C.) 171
5.6 Causes of Variation 171
5.6.1 Non-assignable Causes (Random Causes) 171
5.6.2 Assignable Causes (Preventable Causes) 171
5.7 Techniques of Statistical Quality Control 172

5.8 Control Charts and their Basis 173
5.9 Types of Control Charts 174
5.9.1 Control Charts for Variables 174
5.9.2 Control Charts for Attributes 189
5.10 Control Charts for Variables Versus Control Charts for Attributes 201
5.11 Advantages of Statistical Quality Control 201
5.12 Limitation of Statistical Quality Control 202
5.13 Sampling Inspection (Acceptance Sampling) 202
5.14 Producer's Risk 203
5.15 Consumer's Risk 203
5.16 Acceptance Quality Level (A.Q.L.) 203
5.17 Rejectable Quality Level (R.Q.L.) or Lot Tolerance Proportion Defective
(L.T.P.D.) 203
5.18 Average Outgoing Quality Limit (A.O.Q.L) 204
5.19 Average Sample Number (ASN) 204
5.20 Operating Characteristic (OC) 204
5.21 Single Sampling Plan 204
5.22 Double Sampling Plan 206
5.23 Single Sampling vs Double Sampling Plans 206
Ø Exercise 207
Ø Answers 218
mmm
(xvi)
UNIT
1. Time Series
2. Index Numbers
3
U nit-I
1
T ime S eries
1.1 Introduction
m ong the subjects of study in statistics, the classification of data in groups
A according to time of occurrence, place of occurrrence, cause of occurrence, etc. and
their cross classification is of unique importance. The technique of frequency table and
distribution function has been applicable to all these types of classification. The purpose
of time series is a little different.
It mainly concerns with the development of techniques for analysing the changes that
occur in a series of data, with the passing of time. Such an analysis differs from the
ordinary statistical analysis because of the difference in the data and the object. In most
statistical analysis, the sample, through which we study the population, is assumed to be
taken from a stationary population, while in time series each observation is, in general,
from a new population because with the passage of time several changes may take place
which have far reaching effect on the series under study.
1.2 Definitions
A time series is a series of values taken by a variable at different points of time. It gives the
measurements of a phenomenon over a period of time. Such data represent dynamic
variations and therefore, require specialised tools their handling. A series is denoted by:
t : t1, t2 ........ tn
ut : u1, u2 ...... un
4
where t1, t2 ,.... tn stand for different points of time. These points of times may be in
hours, days, weeks, months, years etc.
According to Kenny and Keeping, "A set of data depending on the time is called a time series".
According to Croxton and Cowden, "A time series consists of data arranged in chronologically".
1.3 Applications of Time Series Analysis

The abelity to model and perform decision modelling and analysis is an essential feature
of many real world applications: The following applications of time series may be
discussed:
1. In Financial Decision Making: Econometrics is vital in finance and in financial

time series analysis. A key element of financial planning and financial forecasting is
the abelity to construct models showing the interrelatedness of financial data.
Models showing correlation or causation between variable can be used to improve
financial decision making.
2. To Understand Past Behaviour: The analysis of a time series enables us to

understand the past behaviour or performance. We can know how the data have
changed over time to find out the probable reasons responsible for such changes. If
the past performance of a company has been poor, it can take corrective measures to
arrest the poor performance.
3. Helpful in Planning: A time series analysis helps directly in business planning. A

firm can know the long term trend in the sale of its products. It can find out at what
rate sales have been increasing over the years. This may help it in making sales
projections for the next few years and plan the procurement of raw material,
equipment and manpower accordingly.
4. To Understand Cyclical Variations: By time series analysis we can judge the

cyclical pattern in certain series of data. It helps in making generalizations in the
concerned business or industry.
5. To Analyse Meaningful Comparisons: A time series analysis enables one to make

meaningful comparisons in two or more series regarding the rate or type of growth.
For example, growth in consumption at the national level can be compared with
that in the national income over specified period. Such comparisons are of
considerable importance to business and industry.
The Examples of Times Series are as Under :

1. Rainfall in different years.
2. The annual production of steel in India over the last 10 years.
3. Monthly production of eggs.
4. Weekly circulation of bank notes.
5
5. The daily closing price of a share in Delhi stock exchange.

6. Hourly temperature recorded by meteorological department.
7. Daily sales and profits in a departmental store.
8. Growth of bacterial population per minute.
9. The weight of animal recorded at various stage of growth.
Requirements of a Time Series: The essential requirements of a time series are

1. The time gap between various values must be equal as far as possible.
2. It must consist of a homogeneous set of values.
3. Data must be available for a long period.
1.4 Components of a Time Series

By a casual observation one can notice the different types of observed time series. The
time series of a population of India shows a steady growth without much variation from
the main point of view. On the other hand the rainfall data shows a markable difference
from a steady growth or decrease. The rainfall may be 10 inches today, zero next day and
again 15 to 20 inches on the other day. In the case of population data if the population is
known for 2001, we can predict with more chance of success, the population of 2011 as
compared to the prediction of the amount of rainfall of next day or ten days after. All that
one can say, in the later case, is that the movement is very irregular. In certain other types
of data like production of ice cream, sugar or eggs, the values for the different months of
an year are more or less repeated year after year. In certain other time series particularly
in economic time series if the data is available over a long preiod of time, a certain types
of oscillation with a much larger period than one year is observed. This is attributed to
the so called business cycle.
After having seen the variety of observed time series it is natural to enquire, why there is
so much difference in time series ? A time series in its forward motion follows a certain
course that represents the net effect of the interaction of several forces pulling it up and
down. If these forces were in the state of equilibrium, then the series would remain
constant. But these forces are never in a state of equilibrium and hence the time series
also fluctuates. As usual with any scientific analysis the object of time series is to isolate
the various forces at work and, if possible, to study them separately. such an analysis is of
greate importance. Again, in order to predict what values the series will take for the next
value of t, a knowledge of various forces at work is very essential. Unless one knows the
various forces, one cannot decide whether or not the same forces will continue to operate.
6
Hence, the main problems in time series analysis are:

1. What are the forces or components, the net effect of whose interaction is expressed
by the movement of a time series, and
2. How they can be separated and studied independently ? The forces at work,
effecting a time series can be broadly classified into the following categories:
(i) Long-term components.
(ii) Short-term components.
It is shown below by a chart:

Let us discuss these components one-by-one,
Components of a Time series
Long Term Short Term
Secular Trend
Regular Irregular or
Random Fluctuations
Seasonal Cyclical
Variations Variations
1.4.1 Trend
Trend is the general tendency of the data, i.e., trend we mean smooth and regular
long-term movement of the data. It may be defined as the tendency of the data to
increase or decrease over a period of time. Some series may exhibit an upward or a
downward tendency or may remain constant. Some series after a period of growth
(decline) may reverse their courses and enter a period of decline (growth). It generally
happens with the time series of economic and business data. For example, an upward
tendency would be seen in population, prices of certain commodities, agricultural
production etc. While a downward tendency would be seen in the data of infant deaths,
price of electronic items etc.
The term "long period" of time series cannot be defined precisely. In some cases a period
as small as a week may be fairly long while in some cases it may be as long as 10 years or
20 years.
1.4.2 Seasonal Variations

Another important component in a time series is seasonal variations. They are those
periodic movements, where the period of movement is not longer than one year. They
refer ot the identical or almost identical patterns which a time series appears to follow
during the corresponding montsh of successive years. According to the name, weather
plays an important role in such type of variations. Following are the two causes
responsible for seasonal variations:
7
1. Those resulting from natural forces and

2. Those resulting from man-made conventions.
For example, the production of goods depend on season, e.g., sugar, eggs, etc. On the
other hand, in a departmental store up just before Christmas or Diwali or some other
festival, a circumstance related to man-made convention rather than to weather.
Seasonal variations are periodic and regular. They are definite and precise, and can be
predicted with greater degree of accuracy. For example, the number of market goers in
the first week of month is more and decrease day by day. Such variations are expected to
happen regularly, month after month and their effects or the magnitude can be
reasonable predicted. Similarly, there is a greater demand for umbrella in rainy season
than other reasons. Thus, we can infer that seasonal variations are the fluctuations set up
as a result of the changes in weather, climatic conditions, customs, habits of the people,
traditions and festivals taking place during a period of twelve months.
1.4.3 Cyclical Fluctuations

In additions to seasonal variations there is another oscillatory movement observable in
certain economic time series. it is generally of a period much larger than one year. It is
attributed to what is called the business cycle. Most of the economic and commercial
series, e.g., series relating to prices, production and wages etc., are effected by these
business cycles. Like seasonal variations, cyclical variations are also regular but the
period of business cycles is not regular as the period of seasonal variations. While
seasonal variations occur within a period of twelve months, cycles repeat at intervals
ranging from seven to ten years. Hence, the period of the cycle is rather uncertain. As
such cyclical variations cannot be predicted with accuracy.
Cyclical variations refer to long-term oscillations about a trend line. They may or may
not follow exactly similar patterns after equal intervals of time. An important example of
cyclical movements is the so called business cycles representing intervals of prospertity,
recession, depression and recovery. These phases are shown below:
Prosperity
D
Level of Economic Activity
ec
lin
e
Normal
A B
y
ver
o
Rec
Depression
5 10 15
Number of years
Fig. 1: Phases of business cycle
8
If there were no variations in a time series, a graph of the data over a period of time will be
a horizontal line. But when a time series is plotted, we get ups and downs in the form of
prosperity and depression. Fig. 1 shows that AB is the line representing normal situation.
Above AB we have the stage of boom and the stage of decline or recession. Below this line
we have depression in downswing and recovery in upswing.
The purpose of studying the cyclical variation is:

1. One can easily study the character of business fluctuations. Good policies can be
formulated at stablizing the level of business activity.
2. Businessman can take timely steps in maintaining business during booms and
depressions
3. A careful study of cyclical variations faclitates a businessman to face the recession
period and make them ready to reap the benefits during booms.
1.4.4 Irregular or Random Fluctuations

A part from all these regular components there is another factor which is invariably found
in all the time series. This is called irregular because of its upredictable nature. Such
variations include all type of fluctuations other than those which have been accounted
for the trend, the seasonal and the cyclical movements. The factors like earthquake, wars
flood, revolution, etc., are of unpredictable nature and influence a time series. They
reveal no definite pattern and that is why they are called irregular or random
fluctuations. Becuase of its random nature it is very difficult to isolate the effects of such
forces.
Uses of Time Series: The time series analyses is useful in every sphere. It is very
important in economics, business, state administration and planning, science,
astronomy, sociology, biology, research work etc., because of the following reasons:
1. It helps in understanding past behaviour and it will help in estimating the future
behaviour.
2. It helps in planning and forecasting. It is very essential in business and economics

with the help of time series we can prepare plans for the future; short term and long
term estimates are also possible.
3. Comparison between data of one period with that of another period is possible.
4. We can evaluate the progress in any field of economics and business activity. For
example, with the change in price level, we can know the change in the purchasing
power of money.
5. Seasonal, cyclical, secular trend of data is useful not only to economists but also to
the businessman.
9
1.5 Analysis of Time Series
The statistician's problem is to develop some methods to identify these components and
measure them separately. This description is generally mathematical. A time series is the
result of the combined effect of all the above four components. Using the classical
approach it is assumed that there is always a good relationship between these four
components and this is represented in the form of a multiplicative model. Symbolically,
U t = Tt ´ St ´ Ct ´ I t
where U t represents the time series value at the time 't', Tt represents the trend value
St , Ct and I t respectevely represent seasonal, cyclical and irregular components of a time
series at time t.
Another approach is to consider an additive model. Symobolically
U t = Tt + St + Ct + I t
where all the parameters are same as defined above. This model, however, is not generally
used since it is considered inappropriate for most economic data. However, if U t
represents the logarithm of the original variable, then this model may be used because
multiplicative decomposition of a time series is same as the additive decomposition of
the logarithmic values of the original time series. In practice, multiplicative model is used
in most of the economic series.
1.6 Measurement of Trend

The concept of trend values is very useful in the study of economic events. Trend values
indicate the trend of the given data and give us an idea of the different types of
fluctuations; they indicate whether the fluctuations are regular or irregular. We know
that if trend values are subtracted from the original series, the remainder is the short term
fluctuations. Following are the methods for measuring the trend:
1. By inspection or graphic method
2. Method of semi-averages
3. Method of moving averages
4. Method of curve fitting
1.6.1 By Inspection or Graphic Method

In this method, we plot all the values of the time series on a graph paper (time t an X -
axis and U t on Y -axis) and draw an off-hand curve which passes through as near to all
the points as possible. At the same times the smoothness of the curve is also maintained.
10
The resulting free hand curve will eliminate the short-term and irregular fluctuations and
shows the long-term tendency of the daya. This method is quite simple and flexible and
can be used for all types of trends, linear and non-linear and requires a minimum labour.
But the success of this method depends on the efficiency of the person who draws it.
Further, different persons may draw different trend lines from the same data. As such,
this method should be handled by experts only.
Solved Examples
Example 1: Following are the figures of sales for various years. Determine the trend line
by free-hand method:
Years 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Sales '000' 60 70 80 100 80 120 110 100 90 160 140 170

Units
Solution:
Y
180
170
Ut
160 ine
ndl
150 Tre
140
130
120
Sales (000 units)
110
100
90
80
70
60
50
40
30
20
10
X
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Years
Fig. 2: Trend by graphic method
11
1.6.2 Method of Semi-Averages

In this method, the entire data is divided in two parts and the values of each part is added
and averaged. These two averages are the two points corresponding to their middle years
on the graph. The line obtained by joining these two points is the required trend line and
may be extended downward and upward with a view to get intermediate values or to
predict future values.
When the number of years is an odd figure, the middle year is left out, i.e., the middle
year is not considered in such examples. This method is simple and trend line will not
vary from person to person as was in the case of graphic mehtod. However, this method
assumes that there is only a straight line relationship between these two points. This is
because through these two points we draw simply a straight line and not a curve. Also, the
effects of the cyclical fluctuations are not eliminated in this method.
Example 2: Determine straight line trend by method of semi-averages from the following data:
Years 1998 1999 2000 2001 2002 2003 2004

Export (Cr.) 19.5 22.5 20.1 24.9 30.8 28.7 27.5
Solution: We prepare the following table for finding semi-averages:
Year Export (in Cr.) Total export Semi-Average Average year
1998 19.5
1999 22.5 62.1 20.7 1999
2000 20.1
2001 24.9 Left out
2002 30.8
2003 28.7 87.0 29.0 2003
2004 27.5
Y
40
e
35 d lin
Tren
30
Export (in crores)
25 Original series
20
15
10
5
X
1998 1999 2000 2001 2002 2003 2004
Years
Fig. 3: Trend line fitted by semi-average method
12
Here the number of years is an odd figure and thus the figure corresponding to middle
year, i.e., 2001 is ignored. Now we plot the original series and trend line corresponding to
two points of semi-average.
1.6.3 Method of Moving Averages

This is an extension of the semi-average method. In this method, trend is obtianed by
means of a series of averages. By this method the effects of cyclical and erratic
fluctuations are ironed out. Moving average involves the calculation of average values for
different years by taking three, four or more yearly values for computing the average.
The process of moving average is very simple. It is calculated merely by using the
technique of simple arithmetic average. Moving average of period k is a series of
successive averages of k terms at a time, starting form Ist, 2nd, 3rd and so on until all the
terms of the time series are exhausted. if k is odd say (2 m + 1), then the moving average is
put against the mid-value of the period it covers, i.e., (m + 1). On the other hand, if k is
even say 2m, it is placed between two middle values of the period it covers. In case of k is
even, the moving value does not coincide with the original time period and it is called
uncentered moving average. In this case, the moving average of two moving averages is
taken and placed between two moving averages it is called the centered moving average.
The method will be discussed in due course.
The function of moving averages is to eliminate the fluctuations and given only the
general trend of the time series. All moving average periods will not give the desired
results. The central problem in this method is to select an appropriate period which will
eliminate all fluctuations that draw the series away from the trend. If the periodicity is
noticed in the cycle of the original series, the period of moving average must coincide
with the peirod of the cycle or its multiple. Further, if we have a more irregular series, the
period of moving average should be taken long so that the irregularity in one direction
may be reduced to minimum or cancle out by the irregularity in another direction.
1.6.3.1 Disadvantages of Moving Average Method

The moving average method has certain disadvantages which are discussed below :
1. Trend values for all the terms are not available. There is a tendency to cut corners
which results in the loss of data at ends.
2. It does not result in mathematical equation which may be used for forecasting. This
problem is removed in method of curve fitting.
3. The selection of period of moving average is a very difficult task.
4. It is sensitive if the data is irregular in nature.
5. If correct period is not chosen, correct picture of trend cannot be obtained.

13
Example 3: Find 3-yearly and 4-yearly moving average for determining trend values in
the following time series data:
Year 1995 1996 1997 1998 1999 2000 2001
Profit in (000) Rs. 12 14 16 15 13 14 18
Solution: Computation of 3-yearly moving average
Years Profit 3-Yearly moving totals 3-Yearly moving averages
(1) (2) (3) (4)
1995 12 — —
1996 14 42 14.00
1997 16 45 15.00
1998 15 44 14.67
1999 13 42 14.00
2000 14 45 15.00
2001 18 — —
Computation of 4-yearly moving average
Years Profit Sum of fours 4-Yearly moving 4-Yearly moving average

Average centered
(1) (2) (3) (4) (5)
1995 12 —
1996 14 —
57 14.25
14 .25 + 14 .50
1997 16 = 14 .38
58 14.50 2
14 .50 + 14 .50
1998 15 = 14 .50
14.50 2
58
14 .50 + 15 .00
1999 13 = 14 .75
60 15.00 2
2000 14 —
2001 18 —
14
In the above example, period of moving average is 4 (even) years, the moving averages
have to be centered as in Col. (5) so as to be related to specific years. Thus, the average
figure of 14.25 has been placed in between 1996-97. Two such averages have been added
up and divided by two so that the figure 14.38 has been obtianed as four yearly moving
average for 1997. This procedure is continued for the entire series.
Example 4: Find the trend values of a trading organisation by moving average method.
Year Annual sales Rs. in ('000') Year Annual sales Rs. in ('000')
1984 86 1994 84
1985 84 1995 96
1986 80 1996 92
1087 88 1997 104
1988 98 1998 116
1989 92 1999 112
1990 84 2000 102
1991 88 2001 114
1992 80 2002 108
1993 100 2003 126
(Use an appropriate period of moving average).
Solution:
First cycle ® 1984 – 1988 = 5 years
Second cycle ® 1989 – 1993 = 5 years
Third cycle ® 1994 – 1998 = 5 years
Fourth cycle ® 1999 – 2003 = 5 years
Thus, there is a peak after 5 years and period of moving average will be 5 years.
Computation of 5-yearly moving average
Year Annual sales (Rs. in 5-Yearly moving totals 5-Yearly moving

'000') average
(1) (2) (3) (4) = (3)/5
1984 86 — —
1985 84 — —
1986 80 346 69.2
1987 88 352 70.4

15
1988 98 352 70.4
1989 92 360 72.0

1990 84 352 70.4
1991 88 354 70.8
1992 80 346 69.2
1993 100 358 71.6
1994 84 362 72.4

1995 96 386 77.2
1996 92 402 80.4
1997 104 430 86.0
1998 116 436 87.2
1999 112 458 91.6
2000 102 462 92.4
2001 114 472 94.4
2002 108 — —
2003 126 — —
1.6.4 Method of Curve Fitting

This is the best and most objective method of determining trend. In this case an
appropriate mathematical function f (t) is first selected and then the constants involved
in the function are obtianed on the basis of data available. The selection of the
appropriate mathematical function is decided by plotting the original time series. An
examination of the plotted data often provides an adequate basis for deciding upon the
type of trend to use. The various types of curves that may be used to describe the given
data in practice are:
1. Y = a + bX Linear trend (Straight line)
2. Y = a - bX + cX 2 Quadratic trend (Second degree parabola)
3. Y = ab X Exponential trend (Exponential curve)
1.6.4.1 Fitting of Linear Trend

When a straight line trend is the 'best fit' of the original data, the mathematical device of
least squares may be used for trend-fitting. This method consists of finding the values of
the constants a and b in the straight line equation:
Y = a + bX
16
The constants 'a' and 'b' are also called the parameters of the straight line. 'a' represents
the value of the trend when X = 0, and parameter 'b' gives the slope of the line or the
amount by which the trend increases or decreases for each unit of time. The values of 'a'
and 'b' are obtained by the following two normal equations:
SY = Na + bSX ... (1)
SXY = aSX + bSX 2 ... (2)
Solving the above equations, the values of 'a' and 'b' are obtained as follows:
SXY - XSY
b= , a =Y = - bX
SX 2 - XSX
If a series consists of odd number of years (time), considerable arithmetic can be reduced
by taking the middle year of the time series as origin and deviations of various years are
calculated from that point. In this case, the constant 'a' and ' b' are evaluated by
SY SXY
a= ;b=
N SX 2
It is also called short-cut method.

Suitabelity of Least Squares Method: Trend equations are used for description of the
growth or decline of the time series and as an aid to the study of the long-term trend of
the economy. The method of fitting curves completely eliminates personal bias and trend
values for all the given periods can be obtained. This is, however, not possible with
moving average method. But the choice of the type of the curve is arbitrary and one can
not be sure whether a linear or parabolic curve will represent the trend best. The choice of
the trend equation may itself lead to a bias. It is, therefore, possible to get some idea of
the pattern of trend from the scatter diagram of the data.
Example 5: Below are given the figures of production ('000' quintals) of a sugar factory:
Year 1996 1997 1998 1999 2000 2001 2002
Production 80 90 92 83 94 99 92
Fit a straight line trend to the given figures:
(i) If 1996 is taken as origin.
(ii) If middle year (1999) of the series is taken as origin.
Solution: (i) If 1996 is taken as origin.

17
Calculation of trend values by the method of least squares
Years Production Deviation from XY X2 Trend values

1996 (X) Yt = 84 + 2 X
1996 80 0 0 0 84
1997 90 1 90 1 86
1998 92 2 184 4 88
1999 83 3 249 9 90
2000 94 4 376 16 92
2001 99 5 495 25 94
2002 92 6 552 36 96
N =7 SY = 630 SX = 21 SXY =1946 SX 2 = 91
The values of ' a ' and ' b ' are given by:
SXY - XSY 1946 - 3 ´ 630 1946 - 1890 56

b= = = = =2
SX 2 - XSX 93 - 3 ´ 21 91 - 63 28
21 630
a = Y - b X = 90 - 2 ´ 3 = 84 where X = = 3, Y = = 90
7 7
The fitted trend equation is :
Y t = 84 + 2 X
from this equation, the trend values for various values of X have been found and
shown in the last column of the above table. It is clear that their graph will give a
straight line.
(ii) If middle year, i.e., 1999 is taken as origin.

Calculation of trend values by the method of least squares.
Years Production Deviatioin XY X2 Trend values

(Y ) from 1999 Yt = 90 + 2 X
(X)
1996 80 –3 – 240 9 84
1997 90 –2 – 180 4 86
1998 92 –1 – 92 1 88
1999 83 0 0 0 90
18
2000 94 +1 + 94 1 92
2001 99 +2 + 198 4 94
2002 92 +3 + 276 9 96
N =7 SY = 630 SX = 0 SXY = 56 SX 2 = 28
SY 630 SXY 56
a= = = 90, b = = =2
N 7 SX 2 28
Thus, the fitted trend equation is
Y t = 90 + 2 X
The various trend values for different values of X are obtained and shown in the last
column of the above table.
It may be verified that trend values are same either by taking 1996 as origin or 1999 as
origin.
Example 6: Compute the trend values of the following data by the method of least
squares:
Years 1997 1998 1999 2000 2001 2002
Sales (in '000') Rs. 10 12 15 16 18 19
Solution: Here series consists of an even number of years. In this case the middle year
falls in between two years. Thus, slight modification is made, we transform the variable.
(Year - average of two middle values)

X=
1 / 2 (interval in years)
Calculation of trend by least square method
Year Sales ('000' X =2 X2 XY Trend values

Rs) (Y) (Year-1999.5) Yt = 15 + 0 .914 X
1997 10 –5 25 – 50 10.430
1998 12 –3 9 – 36 12.258
1999 15 –1 1 – 15 14.086
2000 16 +1 1 16 15.914
2001 18 +3 9 54 17.742
2002 19 +5 25 95 19.570
N =6 SY = 90 SX = 0 SX 2 = 70 SXY = 64
19
The values of ' a ' and ' b ' are obtained by using,
SY SXY
a= and b =
N SX 2
Thus, the fitted equation is
Y t = 15 + 0.914 X
The various trend values for different values of X are obtained and shown in the last
column of the above table.
Example 7: Below are given the figures of production (in quintals) of a rice mill:
Year 1993 1995 1996 1997 1998 1999 2000
Production (in 77 88 94 85 91 98 90
'000' quintals)
(i) Fit a straight line by the method of least squares and tabulate the trend values.
(ii) Eliminate the trend and obtain short term fluctuations.
(iii) What is monthly increase in the production of rice.
(iv) Find trend value for the year 2003 and 2005.
(v) Convert your annual trend equation into monthly trend equation
Solution: Computation of trend values by method of least squares
Years Production Deviation XY X2 Trend values Shrot-term
(Y ) from Yt = 90.01 fluctuations

1997 (X) + 7.11X Y - Yt
1993 77 –4 – 308 16 61.57 15.43
1995 88 –2 – 176 4 75.79 12.21
1996 94 –1 – 94 1 82.90 11.10
1997 85 0 0 0 90.01 – 5.01
1998 91 +1 91 1 97.12 – 6.12
1999 98 +2 196 4 104.23 – 6.23
2000 90 +3 450 9 111.34 – 21.34
N =7 SY = 623 SX = -1 SXY =159 SX 2 = 35
(i) For finding the values of ' a ' and ' b ', we solve the following normal equations:
SY = Na + bSX ... (1)
SXY = aSX + bSX 2 ... (2)

20
Substituting the values from the table, we get
623 = 7a – b ... (1)
159 = – a + 35 b ... (2)
Solving (1) and (2), we get

630 .11 1736
a= = 90 .01 and b = = 7 .11
7 244
The fitted trend equation is
Y t = 90 .01 + 7 .11X
The trend values are obtained by giving different values of X, which are shown in
the above table.
(ii) If we eliminate trend, we get short-term fluctuations, i.e., we obtain short-term

fluctuations as Y - Y t . These are obtained and shown in the last column of the above
table.
(iii) The monthly increase in the production of rice is obtained by dividing the annual
rate of increase, i.e., ' b ' by 12.
b 7 .11
\ Monthly increase = = = 0 .5925 thousand quintals = 592.5 quintals
12 12
(iv) For finding trend values for 2003 and 2005, we change them into X according to
the deviations, i.e.,
X = 2003 - 1997 = 6 and X = 2005 - 1997 = 8
Thus, trend equation for 2003 is
Y t = 90 .01 + 7 .11 ´ 6
= 90.01 + 42.66 = 132.67
Again trend value for 2005 is
Y t = 90 .01 = 7 .11 ´ 8
= 90.01 + 56.88 = 146.89

a b x 90.01 711
. x
(v) Monthly trend equation is Y t = + . = + . = 7 .5 + . 0453 x
12 12 12 12 12 12
Example 8: Find the linear trend equation from the following two normal equations
obtained from the origin 2002:
8 a - 12 b = 47.8
21
- 12 a + 60 b = - 67.4.
Solution: To determine the value of 'a' and 'b', we solve the given normal equations:
8a - 12 b = 478
. ... (1)
- 12 a + 60 b = - 674
. ... (2)
Multiplying equation (1) by 3 and equation (2) by 2, we get
24 a - 36 b = 143.4
- 24 a + 120 b = - 134.8
84b = 86
.
86
.
Þ b= = 0102
.
84
Putting the value of ' b ' in equation (1) we get
. = 8a - 12 ´ 0102
478 .
8a = 1.224 = 47 .8 or 8a = 47 .8 + 1.224
49.024
8a = 49.024 or a = = 6128
.
8
Thus, the required equation is
y = 6128
. + 0102
. x
1.6.4.2 Fitting of Quadratic Trend

It is noted that for a short period of time, a straight line may provide a reasonably good
description of the trend of a series, but that for longer periods, a curved line of some sort
may be called for. Apart from the two constants ' a' and ' b', a third constant ' c' is added to
the equation for the second degree curve, which has one bend. Because of the bend in the
second degree curve, the slope of the curve is continually changing. If a sufficient number
of X values are included, the second degree curve will have a positive slope in one portion
and a negative slope in another.
The second degree curve (parabola) is little more complicated than a straight line because
it involves the addition of cX 2 to the equation of straight line. Thus, the equation of
second degree curve is given by:
Y = a + bX + cX 2
22
Since there are three constants unknowns, the following three normal equations are
required:
SY = Na + bSX + cSX 2 ... (1)
SXY = aSX + bSX 2 = cSX 3 ... (2)
SX 2Y = aSX 2 = bSX 3 + cSX 4 ... (3)
The origin many be taken at the middle year or between the two middle years as taken in
the case of straight line.
Example 9: Fit a parabolic curve and find the trend values from the given data.
Year 1992 1993 1994 1995 1996 1997 1998
Values of Y 6 10 15 39 25 20 15
Solution: Let the equation of fitted parabolic curve is
Y = a + bX + cX 2
Table for Calculation of Parabolic trend values
Year Y Deviation XY X2 X 2Y X3 X4
form
middle year
1995, (X)
1992 6 –3 – 18 9 54 – 27 81
1993 10 –2 – 20 4 40 –8 16
1994 15 –1 – 15 1 15 –1 1
1995 39 0 0 0 0 0 0
1996 25 +1 25 1 25 1 1
1997 20 +2 40 4 80 8 16
1998 15 +3 45 9 135 27 81
N = 7 SY =130 SX = 0 SXY = 57 SX 2 = 28 SX 2Y = 349 SX 3 = 0 SX 4 = 196
Substituting the above values in the following normal equations:

23
SY = Na + bSX + cSX 2 ... (1)
SXY = aSX + bSX 2 = cSX 3 ... (2)
SX 2Y = aSX 2 = bSX 3 = cSX 4 ...(3)
130 = 7a + b × 0 + c × 28 ...(1)
57 = a ´ 0 + b ´ 28 + c ´ 0 ...(2)
349 = a ´ 28 + b ´ 0 + c ´196 ...(3)
or 130 = 7a + 28c ...(1)
57 = 28b ...(2)
349 = 28a + 196 c ...(3)
By equation (2)
57
b= = 2.04
28
For the values of ' a ' and ' c ', we multiply equation (1) by 4 and subtracting equation (3)
from (1), we get
520 = 28a + 112 c
349 = 28a + 196 c

– – –
171 = – 84 c
171
\ c=– = - 2.04
84
Putting the value of c in equation (1), we get
130 = 7a + 28 (– 2.04)
7a = 187 .12
18712
.
\ a= = 26.73
7
Thus the fitted parabolic trend curve is
Y t = 26.73 + 2.04 X - 2.04 X 2

24
From this equation we can get the trend values Y t for various values of X. Thus, the trend
values for different years are:
1992 where X = - 3 Y t = 2 .25
1993 where X = - 2 Y t =14.49
1994 where X = -1 Y t = 22.65
1995 where X = 0 Y t = 26.73
1996 where X = + 1 Y t = 26.73
1997 where X = + 2 Y t = 22.65
1998 where X = + 3 Y t =14.49
1.6.4.3 Fitting of Exponential Trend

Many economic and business data tend to gain a constant percentage each year . Their
progress is geometric. In such a case, a straight line given by the equation y = ab X is used.
This is the equation of an exponential curve. compound-interest-curve describes such a
trend best. It is not possible to fit the exponential curve directly to the Y values by least
squares; we can however, make a least square fit to the logarithms of the original data and
this results in minimising the squared deviations of the logarithms of the observed values
from the logarithmic trend values.
Putting the exponential equation in logarithmic form, we get

log Y = log a + X log b
which is a straight line in terms of X and log Y . Logarithms are substitutd for the actual
data when a curve is fitted by the method of least square. Like in straight line, the normal
equations are
S log Y = N log a + log bS X ...(1)
SX log Y = log a SX + log b SX 2 ...(2)
When middle year is taken as origin the above equations reduce to:
S log Y = N log a ...(1)

2
SX log Y = log b SX ...(2)
The values of ' a ' and ' b ' are

S log Y SX log Y
log a = , log b =
N SX 2
The values of ' a ' and ' b ' can be obtained by taking antilog.
Example 10: Find the trend values for the given data by the equation Y = ab X .
Year 1998 1999 2000 2001 2002
Profit (in lakhs) 10 35 25 40 70

25
Solution: Calculation of the ordinates of the logarithmic curve
Year Profit (in Year-200 X2 log Y X log Y Log of Trend

lakhs) 0 (X) trend ordinates
(Y) ordinates (in lakhs)
1998 10 –2 4 1.0000 – 2.0000 1.1282 13
1999 35 –1 1 1.5441 – 1.5441 1.3030 20
2000 25 0 0 1.3979 0 1.4778 30
2001 40 +1 1 1.6021 + 1.6021 1.6526 45
2002 70 +2 4 1.8451 + 3.6902 1.8274 67
N =5 SX = 0 SX 2 = 10 S log Y SX log Y
= 7 .3892 =17482
.
S log Y 7. 3892
log a = = = 14778
.
N 5
SX log Y .
17482
log b = = = 017482
.
2 10
SX
Thus, the fitted equation is
log Y = log a + X log b
log Y = 1.4778 + 0.17482X
Thus, we find log of trend values for different values of X, i.e.,
log Y = 14778
. + 017482
. (– 2)
= 14778
. - 0.34954 = 11282
.
which gives Y = Anti log (11282
. ) = 13
Similarly, other values of trend ordinates are calculated and shown in the last column of
the above table.
1.6.5 Some Other Growth Curves for Trend Measurement

In industry, it is seen that series generally follow a law of retardation in their growth. In
such cases an elongated 'S' curve is used for trend values. More commonly used is the
Gompertz Curve. A Gompertz curve is given by
Y = k ab X
log Y = log k + b X log a
The method of fitting of such type of curves is the same as that of exponential curve. The
other curve in such situations is Logistic Curve or Pearl-Reed Curve. The form of the
symmetric logistic curve can be given by
k
Y =
a + bX
1+ e
26
1.6.6 Merits and Demerits of the Method of Least Squares

1.6.6.1 Merits
1. Being a mathematical method, it is free from any bias.
2. The trend values for the extreme items are also available by this method while trend
values for the extreme items were not available in moving average method.
3. As the sum of the positive and the negative deviations of the plotted points from the
line of best fit is zero and the sum of their squares is the least, this method gives the
most satisfactory results.
1.6.6.2 Demerits
1. The choice of particular trend equation is largely subjective matter. There is no hard
and fast rule to decide the equation.
2. This method is time consuming and needs extra care while calculating trend values.
1.6.7 Conclusion
After discussing all the methods of measuring trend we find that graphic method is the
easiest and the method of least squares is the difficult one. However, method of least
squares gives the best trend values and that is why this method is more popular than
other mehtods.
1.7 Measurement of Seasonal Variations

Seasonal patterns are exhibited by most of the business phenomena and their study is
necessitated for the following reasons :
1. To isolate the seasonal variations i.e., to determine the effect of seasons an the size
of the variable and to eliminate them. In other words, to study as to what would be
the value of the variable if there were no seasonal swings.
2. The determination of the seasonal effects is of paramount importance in planning
(a) business efficiency (b) a production programme. For example, the head of a
departmental store would be interested to study the variation in the demands of
different articles for different months in order to plan his future stocks, to cater the
need of public due to seasonal swings. Moreover the isolation and elimination of
seasonal factor from the data is necessary to study the effect of cycles. Obviously,
for the study of seasonal variation,the data must be given for the parts of year, viz.,
month; quarter, week or a day. Different methods for measuring the seasonal
variations are:
(i) Method of Simple Averages (ii) Ratio to Moving Average Method
(iii) Ratio to Trend Method (iv) Link-relative Method.

27
1.7.1 Method of Simple Averages

When the data do not contain any trend or cyclical fluctuations, the simple average may
be found out for each month or quarters (if quarterly data is given) for a series of years.
This is the simplest of all the method of measuring seasonality and consists of the
following steps:
1. Find out the total of the values of similar months/quarters of various years.
2. Compute the average of each month or season for all the years.
3. Compute the grand mean of all the seasonal averages obtained earlier.
4. Seasonal indices for different seasons are obtained by expressing the seasonal
average as the percentage of the grand mean. Thus,
Mean of January
Index number of January = ´100
Grand mean
Mean of February
Index number of February = ´100
Grand mean
and so on
If quarterly data is given, the four seasonal index numbers for four quarters can be
calculated in the same manner.
This method is based on the assumption that the data do not contain any trend or cycle
component. This assumption is, however, not true as most of the series have trends.
Example 11: Compute the seasonal indices by the method of simple averages for the
following data of production of a commodity for the year 2001, 2002 and 2003:
Year Jan. Feb. Mar. April May June July Aug. Sept. Oct. Nov. Dec.
2001 12 11 10 14 15 15 16 13 11 10 12 15
2002 15 14 13 16 16 15 17 12 13 12 13 14
2003 16 15 14 16 15 17 16 13 10 10 11 15
Solution: Here the data is given in an arranged form according to years and months. We
prepare the following table:
Year Jan. Feb. Mar. April May June July Aug. Sept Oct. Nov. Dec.
.
2001 12 11 10 14 15 15 16 13 11 10 12 15
2002 15 14 13 16 16 15 17 12 13 12 13 14
2003 16 15 14 16 15 17 16 13 10 10 11 15
Monthly 43 40 37 46 46 47 49 38 34 32 36 44
Totals
Monthly 14.33 13.33 12.33 15.33 15.33 15.67 16.33 12.67 11.33 10.67 12.00 14.67
averages
28
14.33 + 13.33 + 12.33 + 15.33 + 15.33 + 15.67 + 16.33

+ 12 .67 + 11.33 + 10.67 + 12 .00 + 14 .67
Grand Mean =
12
164.02
= = 13.66
12
Thus,
14.36
Seasonal Index for Jan. = ´100 = 10512.
13.66
13.33
Seasonal Index for Feb. = ´100 = 97 .58
13.66
12 .33
Seasonal Index for March= ´100 = 90 .26
13.66
15 .33
Seasonal Index for April = ´100 = 112 .22
13.66
15 .33
Seasonal Index for May = ´100 = 112 .22
13.66
15 .67
Seasonal Index for June = ´100 = 114 .71
13.66
16 .33
Seasonal Index for July = ´100 = 119.55
13.66
12.67
Seasonal Index for Aug. = ´100 = 92.75
13.66
11.33
Seasonal Index for Sept. = ´100 = 82.94
13.66
10.67
Seasonal Index for Oct. = ´100 = 7811
.
13.66
12.00
Seasonal Index for Nov. = ´100 = 8784
.
13.66
14.67
Seasonal Index for Dec. = ´100 = 107.39
13.66
Example 12: Assuming that trend is absent, determine if there is any seasonality in the
data given below:
Years Questions
I II III IV
2000 3.7 4.1 3.3 3.5
2001 3.7 3.9 3.6 3.6
2002 4.0 4.1 3.3 3.1
2003 3.3 4.4 4.0 4.0
Solution: If trend is assumed to be absent, then the difference in averages of various

quarters (if any) shall be due to seasonal variations.
29
Calculation of Seasonal Variations
Years Quarters
I II III IV
2000 3.7 4.1 3.3 3.5
2001 3.7 3.9 3.6 3.6
2002 4.0 4.1 3.3 3.1
2003 3.3 4.4 4.0 4.0
Totals 14.7 16.5 14.2 14.2
Averages 3.675 4.125 3.550 3.550
3.675
Seasonal Index for I quarter = ´100 = 987
.
3.725
4125
.
Seasonal Index for II quarter = ´100 = 110.8
3.725
3 .550
Seasonal Index for III quarter = ´100 = 95 .3
3.725
3 .550
Seasonal Index for IV quarter = ´100 = 95 .3
3.725
3.675 + 4125
. + 3 .550 + 3 .550 14.9
Grand Mean = = = 3.725
4 4
1.7.2 Ratio to Moving Average Method

We know that a moving average eliminates periodic movements if the extent of moving
average is equal to the period of oscillatory movements sought to be eliminated. Thus, for
a monthly data, a 12 months moving average completely eliminates the seasonal
movements if they are of constant pattern and intensity. In this method the
multiplicative model U t = Tt ´ St ´ Ct ´ I t is used. The method consists of following steps:
1. First 12-months (if monthly data is given) moving averages are taken and are
further centred by taking 2-point moving average. These moving averages will also
eliminate some variations and will give the estimates of the combined effects of
trend and cyclical components.
2. The original data (except for 6 months in the beginning and six months in the end)
is expressed as percentage of these centered average values. Using multiplicative
model, these percentages would then express seasonal and some irregular components.
3. The unadjusted seasonal indices are then obtained by eliminating the random
effects by averaging these percentages. Either arithmetic mean or median or
modified mean can be used for averaging.
4. Grand mean of these 12-month means are obtained. Seasonal indices for different
months are obtained by expressing the seasonal averages as the percentage of the grand
mean. If grand mean is 100, the 12 monthly averages will be taken as seasonal indices.
30
This method does not completely utilise the data since in the case of 12-months moving
average seasonal indices cannot be obtained for first and last 6 months. But still among
the different methods it is most satisfactory, flexible and widely used method.
Example 13: The following are the wheat prices in Rs. per quintal:
(i) Estimate the trend values
(ii) Seasonal indices by ratio to moving average method.
Seasons Summer Monsoon Autumn Winter

Years
1997 75 60 54 59
1998 86 65 63 80
1999 90 72 66 85
2000 100 78 72 93
Solution: Table showing moving average (trend) and ratio to moving average
Years Sasons Wheat Sasonal Seasonal Moving Ratio of moving

t price Rs. totals totals average Tt average (Percentage)
Ut centered ut
´100
Tt
1997 Summer 75 — —
Monsoon 60 — —
248
Autumn 54 507 63.4 85.17
259
Winter 59 523 65.4 90.21
264
1998 Summer 86 537 67.1 128.17
273
Monsoon 65 567 70.9 91.68
294
Autumn 63 592 74.0 85.14
298
Winter 80 603 75.4 106.10
305
1999 Summer 90 613 76.6 117.49
308
Monsoon 72 621 77.7 92.66
313
Autumn 66 636 79.5 83.02
323
Winter 85 652 81.5 104.29
329
2000 Summer 100 664 83.0 120.48
335
Monsoon 78 678 84.7 92.09
343
Autumn 72 —
Winter 93 —
31
We arrange percentages of original data to moving average according to years and

seasons to obtain seasonal indices.
Calculation of seasonal indices
Seasons Summer Monsoon Autumn Winter

Years
1997 — — 85.7 90.21
1998 128.17 91.68 85.14 106.10
1999 117.49 92.66 83.02 104.29
2000 120.48 92.09 — —
Totals 366.14 276.43 253.33 300.60
Average 122.04 92.14 84.44 100.20
Adjusted Seasonal 122.04 9214

. 84.44 100 .20
´100 ´100 ´100 ´100
Indicies 99.70 99.70 99.70 99.70
=122.40 = 92.42 = 84.69 =100 .50
122.04 + 9214
. + 84.44 + 100 .20 39882
.
Grand Mean = = = 99.70
4 4
Check: Here, total = 122.40 + 92.42 + 84.69 + 100 . 50 = 400.01~
– 400
1.7.3 Ratio to Trend Method

This method is an improvement over the simple average method. This is based on the
assumption that the seasonal variations for any given period is a constant factor of the
trend. The following steps are taken:
1. First of all, the trend values are calculated by the method least squares by fitting a
straight line or any suitable curve to the given data.
2. The original data are expressed as the percentage of trend values. If the time series is
of multiplicative model, the percentage will contain seasonal, cyclical and irregular
components.
3. Arrange these percentages according to years and months for monthly data (or
according to years and quarters for quarterly data).
4. Find the monthly (or quarterly, for quarterly data) averages by considering
arithmetic mean or the median according to the data.
5. Find overall average of these monthly averages. If the overall average is 100, then
the monthly average will stand for the seasonal indices. However, if the overall
average differs from 100, then we express the monthly (or quarterly as the case be)
averages as percentages of the overall average to get the seasonal indices.
32
Example 14: Find seasonal indices by ratio-to-trend method from the following data:
Quarters
Years
I II III IV
2001 30 35 35 60
2002 30 44 50 60
2003 40 50 60 70
Solution: First, the trend values are obtained by fitting a straight line.
Computation of annual trend values
Year Annual totals Average of X X2 XY Trend value

annual totals Yt = 47 + 7.5X
(Y)
2001 160 40 –1 1 – 40 39.5
2002 184 46 0 0 0 47.0
2003 220 55 +1 1 55 54.5
N =3 SY =141 SX = 0 2 SXY =15
SX = 2
The values of ' a ' and ' b ' are found to be

SY 141 SXY 15
a= = = 47; b= = = 7 .5
N 3 2 2
SX
Thus, the fitted trend equation is Y t = 47 + 7 .5 X
From the equation it is clear that annual increment is b = 7 .5
b 7 .5
\ Quarterly increment = = = 188
.
4 4
Now, for computing quarterly trend from annual trend values given in the last column of
the above table we proceed as follows:
For obtaining trend values of quarter II, we subtract 1.88/2 = 0.94 from annual trend
values 39.5 and add 1.88/2 = 0.94 to annual trend value 39.5. That is,
Quarterly trend value for II quarter = 39.5 – 0.94 = 38.56
Quarterly trend value for III quarter = 39.5 + 0.94 = 40.44
Similarly, Quarterly trend value for I quarter = 38.56 – 1.88 = 36.68
Quarterly trend value for IV quarter = 40.44 + 1.88 = 42.32
Similarly, the quarterly trend values for other years are also calculated and summarised below:
Quarterly trend values
Year I II III IV
2001 36.68 38.56 40.44 42.32
2002 44.18 46.06 47.94 49.82

2003 51.68 53.56 55.44 57.32
33
Now, we find a table for ratio-to-trend values as given below:
Year I II III IV
2001 81.79 90.77 86.55 141.78
2002 67.97 95.53 104.30 120.43
2003 77.40 93.35 108.23 122.12
Total 227.16 279.65 299.08 384.33
Average 75.72 93.22 99.69 128.11
Seasonal 75.72 93 .22 99.69 12811

.
´100 ´100 ´100 ´100
99185
. 99185
. 99185
. 99185
.
= 76 .34 = 93.99 =100 .51 =12916
.
75.72 + 93 .22 + 99.69 + 12811

. 396.74
Grand Mean = = 99185
.
4 4
Check: Here total = 76 .34 + 93.99 + 100 .51 + 12916
. = 400
1.7.4 Link-Relative Method

This method is also known as Pearson's method and is based on averaging the link
relatives. A link relative is the value of one season expressed as a percentage of the
proceeding season. here the word 'seasons' refer to time period, it will be month for a
monthly data and quarter for a quarterly data. We may have link-relative (L.R) for any
season as follows:
Current season' s figure
´100
Previous season' s figure
The following steps are taken:

1. Change the given data into link-relatives by the formula given above. Obviously, L.R
for first season will not be available because previous season figure is not known.
2. Average the link-relatives for each season. Mean or median may be used for
averaging.
3. Convert the averate link relatives into chain-relatives (C.R) on the basis of first
season. Chain relative for any season is obtained by multiplying the link-relative of
that season by the chain relative of the preceeding season and dividing it by 100.
That is,
C.R. for a particular season
Average L.R of the particular season ´ C.R of preceeding season

=
100
Here C.R. for the first season is assumed to be 100.

34
4. Compute the new C.R. of the first season, say C, as

L.R of the first season ´ C.R of the last season
C=
100
This value of C will be different from 100 and will be corrected by a correction
factor ' d ' which is calculated as follows:
C -100
d=
Number of seasons
5. Find the adjusted C.R.'s for the second and other seasons by the following formula:
Adjusted C.R. for r-th season = original C.R of r-th season – (r -1)d
Here r =1, 2, 3 ...
6. Find the grand mean of the adjusted C.R's and express them as percentages of
overall average of adjusted 'C.R's to give the needed seasonal indices.
This method utilises the data more completely than moving average method. There is
only one link-relative less while a 12-months moving average results in the cut of
6-months at each end. This method is effective if the growth is of constant amount or at
constant rate.
Example 15: The data below gives the average quarterly prices of a commodity for five
years. Calculate the seasonal variation indices by the method of link relatives.
Year 1979 1980 1981 1982 1983

Quarter
I 30 35 31 31 34
II 26 28 29 31 36
III 22 22 28 25 26
IV 31 36 32 35 33
[C.C.S. Meerut 2004]

Solution: First, we will calculate link-relatives by the formula:
Current season' s figure
L.R = ´100
Previous season' s figure
26
Thus, L.R for II Quarter for 1979 = ´100 = 86.7
30
22
L.R. for III Quarter for 1979 = ´100 = 84.6
26
31
L.R. for IV Quarter for 1979 = ´100 = 140.9
22
Obviously, L.R for I quarter of first year (1979) will not be available because the value of
previous season's figure is not known. The calculated L.R's are presented in the following
table for the calculation of seasonal indices:
35
Computation of seasonal indices
Year Quarters
I II III IV
1979 — 86.7 84.6 140.9
1980 112.9 80.0 78.6 163.6
1981 86.1 93.5 96.6 114.3
1982 96.9 100.0 80.7 140.0
1983 97.1 105.9 72.2 126.9
Total or L.R.'s 393.0 466.1 412.7 685.7
Average L.R 393.0 4661
. 412.7 685.7
= 98 .25 = 93 .22 = 82 .54 = 137 . 14
4 5 5 5
Chain relative 100 100 ´ 93 .22 93 .22 ´ 82 .54 76.94 ´137 .14
100 100 100
= 93 .22 = 76.94 =105 .52
We compute new C.R (C) of the first quarter as

Average L.R of I quarter ´ C.R. of last quarter
C=
100
98 .25 ´105 .52
= = 103.67
100
\ Quarterly adjustment factor,
C - 100 103.67 - 100 3.67
d= = = = 0.92
No of seasons 4 4
Adjusted 100 93 .22 - 0.92 = 92 .3 76.94 - 2 ´ 0.92 105 .52 - 3 ´ 0.92

C.R = 751
. =102.76
Seasonal 100 ´100 92 .3 ´100 75 .1 ´100 102.76 ´100
= 99.74 = 8115
.
Indices 92 .54 92 .54 92 .54 92 .54
= 108.06 =11104
.
100 + 92 .3 + 751
. + 102.76 37016
.
Grand Mean = = = 92 .54
4 4
. + 99.74 + 8115
Check: Here total = 10806 . + 11104
. = 400
1.8 Residual Method for Isolation of Cyclic Variations

A crude method of measuring cyclic variations (Ct ) of a time series is provided by
Residual method. It consists in removing from the given time series the other three
components viz trend (Tt ), seasonal variation (St ) and irregular (I t ). First we remove the
seasonal variation from the data and then trend. In this way the values obtained contain
only cyclic and irregular variations. As a matter of fact, the two variations are almost
36
inseprable. The use of moving average method of suitable period averages (smoothens)
out the random component (I t ). Symbolically,
Ut
For multiplicative model, = Ct ´ I t
St ´ Tt
and for additive model, U t - Tt - St = Ct + I t
This method yields good results only if the trend and seasonal variations are perfectly
measured. The irregular variations can be identified by an examination of the periods
and corresponding observed values. If irregular variations are present, they may be
removed logically or applying the moving average method on Ct ´ I t values.
Harmonic Analysis: When the cyclical variations of series are of about the same duration
and amplitude, a more sophisticated method of determining the cyclical component is
used which is based on trignometrical ratios such as semi-cosine functions represented by a
Fourier series. Thus for a time series U t with period of oscilliation l, we have
2p Sin 2 p ü
U t = a0 + a1 Sin t + a2 . 2 t + ... + . ïï
l l
ý ....(i)
2p 2p ï
b1 cos t + b2 cos . 2t + ... ïþ
l l
n
where, a0 =
1
n åUt
t =1
n
ai =
2
n å U t sin æçè 2lp it ö÷ø,(i =1,2 ...)
t =1
n
bj =
2
n å U t cos æçè 2lp it ö÷ø, (i =1,2 ...)
t =1
where ' n' is the number of terms in a time series. For instance, if the period of scillation is
12 months and u1, u2 ... u12 is the series or average of series for a number of years, then
constants ai and bj are given by
12
a0 =
1
12 åUt
t =1
12
2p ö
ai =
2
12 å U t sin æçè 12 it ÷ , i =1, 2 ...6, l =12
ø
t =12
12
2p ö
bj =
2
12 å U t cos æçè12 it ÷, i =1, 2 ...5
ø
t =1
So far l in (i) is regarded as a known constant. Periodogram analysis provides an elegent

method of determining l.
37
E XERCISE
Long Answer Type Questions
1. Define a time series. Describe with examples the various components of a time
series. [C.C.S. Meerut 2004]
2. Define a time series. What is analysis of a time series ? Discuss the various
components of a time series. [C.C.S. Meerut 2001]
3. Describe in brief various components of a time series. Which of the components, is

mainly applicable in the following cases ?
(a) Business ups and downs after a regular interval of time.
(b) Downfall in production due to fire in a factory.
(c) Decrease in employment in a sugar factory during off season.
(d) Fall in death rate due to advance n medical science. [C.C.S. Meerut 2000]
4. Define a time series. What is analysis of a time series ? Describe the method of
moving averages for determining trend in a time series.
5. What is a time series ? Name the components of variation in a time series and
explain their salient feature with illustrations. [C.C.S. Meerut 2002]
6. What is meant by trend of a time series ? Describe the moving average method of
determining trend. Show, how this is related to the method of fitting a polynomial
by the principle of least square. [C.C.S. Meerut 2012]
7. Describe different method for determining trend in a time series. [C.C.S. Meerut 2002]
8. What is seasonal variation of a time series ? How do you determine seasonal indices ?
[C.C.S. Meerut 2002 (Old Pattern)]
9. What is the seasonal variation of a time series ? How do you determine seasonal
indices ?
10. Describe the moving average method of determining trend. Show, how this is
related to the method of fitting a polynomial by the principle of least square ?
11. What are the various methods of determining trend in a time series ? Describe the
moving average method in detail. [C.C.S. Meerut 2005, 06, 06 (Back)]
12. Which component of a time series is mainly applicable in the following examples ?
Given reasons.
(a) Downfall in production due to fire in a factory.
(b) A continuous increasing demands for synthetic fibre.
(c) Sales of new year greeting cards.
(d) Business ups and downs after a regular interval of time.
(e) Increase in hill station population during summer vacation.
38
13. (a) What are the various methods of determining seasonal variation in a time series?
Describe the method of link-relative in detail. [C.C.S. Meerut 2000 (Back)]
(b) Explain the link relative method of determining the seasonal variation in a
time series. [C.C.S. Meerut 2005, 08]
14. Calculate the trend values using the data given below by taking a 4-yearly moving
average
Year 1987 88 89 90 91 92 93 94 95
Sales 285 320 320 362 355 336 350 342 359
('000' Rs)
[C.C.S. Meerut 2004 (Back)]
15. Calculate trend values from the following data assuming 5-year and 7-year moving
average:
Years 1988 89 90 91 92 93 94 95
Values 110 104 98 105 109 120 115 110
Years 1996 97 98 99 2000 2001 2002 2003
Values 114 122 130 127 122 118 130 140
16. From the following data calculate the 4-yearly moving average and determine the
trend values. Plot the original data and trend on graph paper:
Years 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
Values 50.0 36.5 43.0 44.5 38.9 38.1 32.6 41.7 41.1 33.8
17. Calculate the trend values by the method of least squares from the data given below
and estimate the sales for the year 2001.
Years 1997 1998 1999 2000 2001
Sales of (V.CD 12 18 20 23 27
in '000')
18. Given below are the figures of the manufacture of cloth in lakh or yards. Calculate
the value of trend by the method of least squares:
Years 1999 2000 2001 2002 2003
Production (in lakh 5 7 6 8 10

of years)
39
19. Consider the following data:
Year 1995 1996 1997 1998 1999
Sales (Lakh Rs.) 45 56 78 46 75
(a) Fit a straight line to the above data by the method of least squares taking 1997
as origin and calculate the trend values.
(b) Plot the original and trend values on a graph paper.
(c) Estimate the trend value for the year 2000.
20. Fit a linear trend equation for the following data and estimate the expected profits
for the year 2006:
Year 1997 1998 1999 2000 2001 2002 2003
Profit ('000' Rs.) 60 72 75 65 80 86 5
21. Fit a second degree parabola and find the trend vaules for the following data:
Year 1996 1997 1998 1999 2000
Values 10 12 13 10 8
22. Find the seasonal indices by the method of simple averages from the data given
below:
Year Jan. Feb. Mar. April May June July Aug. Sept. Oct. Nov. Dec.
1998 20 20 19 26 25 30 28 29 26 29 26 30
1999 27 26 21 19 19 21 19 17 18 18 17 18
2000 17 18 15 17 18 19 17 19 21 22 21 26
2001 23 26 18 23 24 20 21 27 26 28 30 31
2002 29 28 29 25 22 23 25 26 30 36 35 40
23. Find the seasonal variation by the method of Ratio-to-Moving Average from the
following data:
Year I II III IV
1999 30 81 62 119
2000 33 104 86 171
2001 42 153 99 221
2002 56 172 129 235
2003 67 201 136 302

40
24. Calculate the seasonal indices by the "Ratio-to-Moving Average" method from the
following data:
Year and Quarter Y Year and Quarter Y
1998 I 75 2000 I 90
II 60 II 72
III 54 III 66
IV 59 IV 85
1997 I 86 2001 I 100
II 65 II 78
III 63 III 72
IV 80 IV 93
25. Calculate the seasonal variation indices by the ratio-to-trend method from the data
given below:
Quarters I II III IV
Year
1982 15 20 18 17
1983 17 26 25 22
1984 20 29 27 24
1985 27 38 34 31
1986 40 46 43 41
[C.C.S.Meerut 2005]
26. Apply link-relative method to the following data and calculate seasonal indices.
Quarters 1996 1997 1998 1999 2000
I 6.0 5.4 6.8 7.2 6.6
II 6.5 7.9 6.6 5.8 7.4
III 7.8 8.4 9.2 7.5 8.0
IV 8.7 7.3 6.4 8.5 7.1

41
Short Answer Type Questions

1. What is a time series ?
2. Explain different components of a time series. [C.C.S. Meerut 2003, 05, 06]
3. Which of the component of a time series is known as large-term component ?
4. What is the meaning of trend ?
5. Explain the irregular part of a time series with an example.
6. Write done the normal equations of a linear trend equation.
7. Write the additive and multiplicative model of a time series.
[C.C.S. Meerut 2004, 06 (Old), 12]
8. Explain any one method of determining the seasonal variations. [C.C.S. Meerut 2003]
9. Describe the method of moving average for determining the trend value of a time
series. [C.C.S. Meerut 2002, 04, 08]
10. What is meant by cyclical variation ? [C.C.S. Meerut 2002, 05]
11. Give the uses of time series. [C.C.S. Meerut 2002, 04, 06]
12. Which method is used to estimate the trend values for future ? [C.C.S. Meerut 2004]
13. Write the name of any two methods of determining seasonal variation. [C.C.S. Meerut 2004]
14. Give tow demerits of graphic method for trend.
15. Which component of a time series is mainly applicable in the sales of new year
greeting cards ? [C.C.S. Meerut 2005]
16. Describe the models of a time series. [C.C.S. Meerut 2005]
17. Describe with example the meaning of trend. [C.C.S. Meerut 2006]
18. Give an example of cyclical component. [C.C.S. Meerut 2006]
19. What is analysis of a time series ? Write both models of a time series.
20. Defferentiate between seasonal and cyclical fluctuations. [C.C.S. Meerut 2008, 12]
21. Explain time series with its applications. [C.C.S. Meerut 2008]
22. Discuss any one method of determining trend. [C.C.S. Meerut 2008]
Objective Type Questions

Multiple Choice Questions
1. The period of cyclic component is :
(a) Less than one year (b) More than one year
(c) More than five years (d) None of these [B.R.A. Agra 2012]
2. From which method secular trend is measured ?
(a) Ratio to moving average method (b) Ratio to trend method
(c) Link Relative method (d) Moving average method

[B.R.A. Agra 2004, 08, 10, 11]
42
3. In a time series 'depression' is related with :
(a) Trend (b) Seasonal variation
(c) Cyclic variation (d) Irregular variation [B.R.A. Agra 2010]
4. Random components in a time series are caused by :
(a) Strikes (b) Epidemices
(c) Flood (d) All of these [B.R.A. Agra 2008]
5. Periodic changes in a business time series are called :
(a) Seasonal variation (b) Cyclical variation
(c) Irregular variation (d) None of these

6. By method of semi-averages we can measure :
(a) Trend (b) Seasonal
(c) Cyclic (d) Random [B.R.A. Agra 2005]

7. From which method seasonal variation can be measured ?
(a) Like relative method (b) Method of last squares
(c) Method of moving averages (d) None of these [B.R.A. Agra 2005]
8. A Long term variation in a time series is attributed to:
(a) Secular trend (b) Seasonal variation
(c) Cyclic variation (d) Irregular variation [B.R.A. Agra 2006, 09]
9. The method of measurement of secular trend is :
(a) Method of moving average (b) Method of monthly (quaterly) averages
(c) Ratio to moving average method (d) Ratio to trend method [B.R.A. Agra 2006]
10. A time series has :
(a) Two components (b) Three components
(c) Four components (d) Five components [B.R.A. Agra 2009]
11. The method of minimum least squares is connected with the analysis of time series
for measuring :
(a) Seasonal variation (b) Long-period variation
(c) Cyclical variation (d) For all the above

12. In time series analysis both trends and seasonal indicies are studied because :
(a) The allow the elimination of the components from the series
(b) They describe past trends
(c) (a) and (b) both
(d) None of the above.

43
13. Time series is a set of data collected :
(a) Choronologically (b) Alphabetically
(c) Periodically (d) (a) and (c) both

14. The time series analysis help :
(a) To compare two or more series
(b) To make predections
(c) To know the behaviour of business
(d) All the above
15. The sales of woolen clothes in winter season is associated with the components of
time series is called:
(c) Irregular variation (d) None of these

16. The consistent increase in production of cereals constitutes the component of a
time series is called:
(c) Cyclical variation (d) All the above

17. Method of least square to fit the trend is applicable only if trend is :
(a) Linear (b) Quadratic
(c) (a) and (b) both (d) None of these

18. The overall tendency of rise or fall in the time series is called :
(c) Cyclical (d) Irregular

19. The component for short - term forecasting is:
(c) Cyclical (d) All the above

20. If annual data are given, the time series is representable by the additive model is :
(a) Y = T ´ S ´ C ´ I (b) Y = T + S + C + I
(c) Y = T + C + I (d) Y = S + C + I
21. In an additive model the component measurements are :
(a) Positive (b) Negative
(c) Absolute (d) None of these

22 A trend equation of the form Y = a + b x + c x2 is called :
(a) Linear (b) Parabolic

(c) Quadratic (d) (b) and (c) both
44
23. With a change in origin, the value of b in the equation y = mx + c :
(a) Remains the same (b) Also changes
(c) Is always positive (d) Is always negative

24. In a multiplicative model C, S and I have:
(a) Positive values (b) Negative values
(c) Either (a) or (b) (d) Indexing value

25 If trend is absent in a time series, the seasonal indicies are obtained by using :
(a) Method of simple averages (b) Ratio to trend method
(c) Ratio to moving average (d) Link relative method

26. Salient factors responsible for seasonal variation are :
(a) Customs and religion (b) Festivals
(c) Weather (d) All of these

27. The terms prosperity, depression, recession, boom are attached with :
(a) Secular trend (b) Cyclical variations
(c) Seasonal variations (d) None of these

28. Loss of working days due to fire in a factory is associated with :
(a) Trend (b) Seasonal component

(c) Cyclical component (d) Random component
29. Decline in morality rate due to advancement in science is associated to :

30. The first step in the analysis of time series is :
(a) To diseasenalize the series (b) To edit the series
(c) To find trend (d) None of these

31. In economic series, the model used is :
(a) Multiplicative (b) Additive
(c) Neither a nor b (d) Either a or b

32. Which of the following component of a time series can not be eliminated ?
(a) Systematic (b) Random
(c) Cyclical (d) Seasonal

33. Link relative means the ratio of a value :
(a) Succeeding value (b) Preceding value
(c) Both (a) and (b) (d) Neither (a) or (b)

45
34. Cyclical variations in a time series are measured by:
(a) Moving average method (b) Link relative method
(c) Residual method (d) Ratio to moving average method
35. A time series consists of:
(a) Long term fluctuations (b) Short term fluctuations
(c) Irregular variation (d) All of the above
36. A time series is affected by:
(a) Economic factors (b) Non-economic factors
(c) Both (a)and (b) (d)Neither (a) nor (b)
37. If the slope of the trend line is positive, it shows:
(a)Rising trend (b) Declining trend
(c) Stagnation (d) None of the above
38. For the given five values 15, 24, 18, 33, 42 the three year moving averages are:
(a) 19, 25, 31 (b) 19, 30, 31
(c) 19, 22, 33 (d) None of the above
39. Irregular variation in a time series are caused by:
(a) Strikes (b) Endemics
(c) Floods (d) All of the above
40. Cycles in a time series are regular in :
(a) Periodicity (b) Amplitude
(c) Both (a) and (b) (d) Neither (a) nor (b)
41. Moving average method is suitable for :
(a) Finding trend values (b) Projections
(c) Both (A) and (b) (d) Neither (a) nor (b)
42. If the origin in a trend equation is shifted forward by 5 years, x in the equation
y = a + bx will be replaced by:
(a) x - 5 (b) x + 5
(c) 5 x (d) None of the above
43. In a time -series 'Depression' is related with :
(a) Trend (b) Seasonal variation
(c) Cyclical variation (d) Irregular fluctuations

46
44. 'Prosperity' is related with:
(a) Seasonal variation (b) Cyclical fluctuations
(c) Trend (d) None of the above
45. In the time series 2, 0, –2, 0, 2, 0, –2, 0, 2 all the four yearly moving averages are:
(a) One (b) Zero
(c) Three (d) Two
46. The regular movements occurring in a period of more than one year are called as:
(a) Seasonal variations (b) Cyclical fluctuations
(c) Random fluctuations (d) None of the above
47. Which method is not used to measure secular trend ?
(a) Semi-average method (b) Moving average method
(c) Chain base method (d) Least square method
48. Loss of working days due to fire in a factory is associated to the:
49. A boom in business activity is associated to the:
(a) Cyclical component (b) Seasonal component
(c) Random component (d) Trend
50. Link relatives in a time series remove the influence of:
(a) The trend (b) Cyclical variation
(c) Seasonal variation (d) None of the above
51. Method of least squares for determining trend is used when:
(a) Trend is known
(b) Trend is curvilinear only
(c) The value Y is not a function of time-t
(d) None of the above
52. Value of b is the trend line Y = a + bX is :
(a) Always positive (b) Always negative
(c) Both positive or negative (d) None of the above
53. A method to measure seasonal variations is:
(a) Method of least squares (b) Semi-average method
(c) Link relative method (d) None of the above

47
54. A group of moving average consists of :
(a) 5 years period (b) 3 years period
(c) A period which forms a cycle (d) None of the above
55. Which of the following is not a mixed model of time series ?
(a) Vt = T + S + C + R (b) Vt = T ´ S + S ´ R
(c) Vt = T ´ S ´ S ´ R (d) Vt = T + S + C ´ R
56. A business cycle is composed of :

(a) Prosperity and depression (b) Recession and recovery
(c) Both (a) and (b) (d) None of the above
57. The sales of a departmental store on Christman are associated with the component
of a time series:
(c) Irregular variation (d) All the above
Fill in the Blank

1. Random fluctuations are also known as ........................ .
2. Secular trend is ........................ term fluctuations.
3. In linear trend equation y = mx + c, m stands for ...................... . [B.R.A. Agra 2007]
4. In linear trend equation y = a = bx, a stands for ........................ .
5. Time series is a set of values arranged in ........................ order (choronological).
6. Time series consists of ........................ components.
7. In a time series trend is a ........................ term component.
8. In a time series random component is a ........................ component.
9. In a time series boom and recession are associated to ....................... component.
10. In a time series, yearly data does not have ........................ .
True or False
1. Link relative method is used to measure secular trend. [B.R.A. Agra 2012]
2. Fire in a factory is associated to the random component. [B.R.A. Agra 2011]
3. Method of simple averages assumes that trend is absent in the data.
4. Period of moving average is determined by the perodicity of the time series.
5. Trend is a systematic part of the time series.
6. Cyclical component is of unsystematic nature.
7. Trend is a short term component of a time series.
8. When the term of moving average is even, the average does not concide with the
original period.
9. The linear trend equation y = a + bx obtained by the method of least square is
known as the line of best fit.
48
A NSWERS
1. (b) 2. (d) 3. (c) 4. (c)
5. (b) 6. (b) 7. (a) 8. (a)
9. (a) 10. (c) 11. (b) 12. (b)
13. (d) 14. (d) 15. (b) 16. (a)
17. (c) 18. (a) 19. (b) 20. (c)
21. (c) 22. (d) 23. (a) 24. (d)
25. (a) 26. (d) 27. (b) 28. (d)
29. (a) 30. (b) 31. (a) 32. (b)
33. (b) 34. (c) 35. (d) 36. (c)
37. (a) 38. (a) 39. (d) 40. (d)
41. (a) 42. (a) 43. (c) 44. (b)
45. (b) 46. (b) 47. (c) 48. (d)
49. (a) 50. (a) 51. (a) 52. (c)
53. (c) 54. (c) 55. (a) 56. (c)
57. (b)
Fill in the Blank

1. irregular 2. long 3. slope of the equation
4. intercept 5. chronological 6. four
7. long 8. systematic 9. cyclical
10. seasonality
True or False
1. False 2. True 3. True
4. True 5. True 6. False
7. False 8. True 9. True
❍❍❍
49
U nit-I
2
I ndex N umbers
2.1 Introduction
n index number is a statistical device for measuring the magnitude of relative
A changes in a series of data. It is an indicator of the changing tendencies in a variable
or variables. According to Croxton and Cowden "Index numbers are devices for measuring
differences in the magnitude of a group of related variables". That is, an index number is a
relative number or relative which expresses the relationship between two figures.
Actually index numbers are special type of averages in which the commodities are
expressed in different units, wheat is expressed in quintals, cloth in metres, kerosene in
litres and so on. Here the price pattern may be different for different commodities, for
some it may be upward while for some it may be downward. Our main aim is to get a
complete expression of this type of complex variation. Thus, an index number is a
number which summarises the movement of such heterogeneous but associated
elements. For comparative purposes index numbers are represented in the form of
percentages with some period taken as standard.
2.2 Characteristics of Index Numbers

1. Specialised Averages: We know that average is a single figure which represents a
group of figures. On the same lines, an index number is also an average which
represents a group of figures not necessarily given in same units. For example, the
wheat is given in quintals, rice is given in kg and milk is given in litres and so on.
50
2. Relative Measurement: The main feature of index number is that it is a relative

measure or comparative measurement of group of items. Infact, an index number
measures the relative changes of differences from time to time. For example, if the
index number of 2000 in comparison to 1994 is 160 then we can understand that
there is a 60% increase in prices of 2000 in comparison to 1994.
3. Measurement of Changes not Capable of Direct Measurement: Index numbers

are used to measure the changes in magnitude which are not capable of direct
measurement due to complex and mixed nature of field. For example, standard of
living, cost of living, changes in economic activities etc.
4. Expressed in Percentages: Index numbers are expressed in percentage to show the

extent of relative changes. However, sign of % is not used.
2.3 Uses of Index Numbers

Index number have become an important tool to measure the changes in economic and
business activities and thus it is sometimes called "Economic Barometer". The following
are the uses of index numbers:
1. Helps in Studying Trends: On the basis of index numbers, the trends in industrial
production, international trade, national income etc. can be studied. Also, the
index numbers are useful in future forecasting.
2. Helps in Policy Formation: Index numbers are used by any government

organisation or an individual business organisation for efficient planning and
decisions. For example, the employer uses cost of living index for declaring dearness
allowance for the workers.
3. Helps in Deflating Various Values: Index numbers are effectively used in

deflating national income on the basis of the price of the base year to enable us to
find out whether there is any change in the real income of the people.
2.4 Points to be Considered in the Construction of

Index Numbers
While constructing an index number, a number of problems are faced. The main
problems are discussed below:
1. Purpose of Index Numbers: The first major problem is that the object of the index
number must be very clear i.e., for what kind of change, we are going to construct an
index number. The main reason is that there is no single index number which can be
used for all purposes. That is why, Retail price index number is used for studying
51
the problems related to income and salary and General price index number is
used for studying the entire social circle.
2. Selection of Base Year: Index numbers are always constructed with reference to
some base period. The point of referecne is called base. While selecting a base
period, we should be careful that the base period should not be too distant in the
past and that period must be a normal period. By normal, we mean that it must be
free from natural calamities and abnormalities like flood, war, earthquake, boom
etc. Further, there are two methods of selecting a base period, first by fixed base
method and second by chain base method. If the base period or reference period is
kept fixed for all current periods of comparison, it is called fixed based period. In
chain base method, the change in the level of the phenomenon for any given period
is compared with the level of the phenomenon in the preceding period and not in
the base period.
3. Selection of Representative Items: There are a number of items which are sold
and purchased. But all the items are not taken into consideration, only those items
are taken which are relevant and serve the purpose of the index. Another point
which play an important role is the number of items. We know that small number
of items will not provide a representative index and a large number of items will not
be economically viable. Thus, a fairly large number of items may be taken.
4. Selection of Representative Prices: After the items are decided, the next step is to
collect the price quotations of these items. But it is a fact that the prices of a number
of items vary from place to place even shop to shop. And it is not possible to collect
price quotations from each place and each shop. In such situations, a selection must
be made of representative places and persons. These places must be well known for
trading. After the decision of well known place, a reliable person or an agency must
be appointed to collect price quotations.
Further, the price quotations in the market are divided into two parts, the retail
prices and wholesale prices. The choice would depend upon the purpose of the
index. If a consumer price index number is to be constructed, the retail prices will be
suitable not the wholesale prices.
5. Choice of a Suitable Average: To summarise a large amount of information, the

selection of a suitable average is necessary. This average plays an important role in
computing index numbers. We know that index numbers are special type of
averages and thus the choice of a suitable average is as important task. Arithmetic
mean, median and geometric mean are the commonly used averages in index
numbers each having its own limitations. For example, arithmetic mean is unduly
affected by the extreme items. Median also is not satisfactory average as it
completely ignores the extreme observations and it cannot measure relative changes
which are essentially required for the construction of index numbers. The geometric
mean is the most suitable average for construction of index numbers. It has the
52
merit of measuring relative changes and also not affected by the extreme
observations.
6. Selection of Proper Weights: In almost every collection of numerical data it may

be observed that the items are of varying importance. So weighing is very necessary
in the construction of index numbers. If weights are not assigned to these items, the
importance given to each commodity in constructing an index number, shall be a
matter of chance. The only solution to this problem is to assign suitable weights to
each item included in the Index. This makes an index number truly representative
of the population under study.
Weight may be either fixed or fluctuating. But as the relative importance of the
commodities keeps changing, it is always advisable that the weights should be
fluctuating from period to period.
7. Selection of Suitable Formula: The selection of a formula depends on the

information available and the accuracy desired, which again depends on the
purpose of constructing a particular Index number. There are various formulae for
calculating index number such as aggregate method or the average of relatives
methods, weighted aggregate index numbers, Fisher's ideal formula and so on.
2.5 Types of Index Numbers

In economic and business the index numbers are classified in four heads:
1. Price Index Numbers
2. Quantity Index Numbers
3. Value Index Numbers
4. Special Purpose Index Numbers
2.5.1 Price Index Numbers

When the relative change in the price is studied, it is called price index number. Such
index numbers are further classified as (i) wholesale price index numbers and (ii) retail or
cost of living price index numbers.
Price relative is the simplest example of an index number. It is defined as the ratio of the
price of a single commodity in the current year to its price in the base year. Thus, if p1 is
the current year price of a (single) commodity and p0 is the base year price of the
commodity, then
p
Price index number = 1 ´100
p0
53
2.5.2 Quantity Index Numbers

When the relative change in the quantity is studied, it is called quantity index number.
By quantity we mean the volume of quantity such as quantity of production,
consumption etc. Quantity relative is also the simplest example of an index number. It is
defined as the ratio of the quantity of a single commodity in the current year to its
quantity in the base year. Thus, if q1 is the quantity of current year and q0 is the base year
quantity, then
q
Quantity index number = 1 ´100
q0
2.5.3 Value Index Numbers

The value is defined as the product of price and quantity i.e.,
Value = Price × Quantity

These are intended to study the change in the total value in current year in comparison to
base year. Thus, if
Value of current period = p1q1
Value of base period = p0 q0
pq
Then, Value index number = 1 1 ´100
p0 q0
2.5.4 Special Purpose Index Numbers

If index numbers are prepared for some specific purposes then they are called special
purpose index numbers such as index numbers of national income, productivity, growth
rate etc.
Note: As discussed earlier, it is to be noted that, a price relative is itself a price index
number in case of a single commodity. Similarly, for the quantity or value of a single
commodity, the quantity or value relative is itself a quantity or value index number.
When there are more than one commodity, we have to combine these
price/quantity/value relatives to get one figure, called price/quantity/value index
number.
Thus, for getting index number in such cases (more than one commodity) various
methods are therefore to combine the price/quantity/value relatives. The easiest way is
to take arithmetic mean or geometric mean of these price/quantity/value relatives. The
various ways of combining these relatives give rise to the different formulae of index
numbers.
54
Solved Examples
Example 1: The prices of 4 commodities in the year 1990 and 1995 are shown in the
table. Find the price relatives taking 1990 as base year.
Commodity Price in 1990 Price in 1995

Rice 15 25
Milk 5 11
Sugar 11 17
Cement 92 136
Solution:
Commodity Price in 1990 Price in 1995 Price relative

p0 p1 p1
´100
p0
Rice 15 25 25
´100 = 166.67%
15
Milk 5 11 11
´100 = 220 .00%
5
Sugar 11 17 17
´100 = 154 .55%
11
Cement 92 136 136
´100 = 147 .82%
92
Example 2: Consider the data as production of rice of a country in millions of tons for
three years. Find the quantity relatives for the year 1996 and 1997 taking 1995 as the
base.
Years 1995 1996 1997

Production of rice 1090 988 1306
(Millions of Tons)
Solution: Based on the above data, quantity relatives for the year 1996 and 1997 taking
as 1995 as base are shown in the following table:
Years 1995 1996 1997
Production of rice 1090 988 1306

(Millions of Tons)
q 1090 988 1306
Quantity relative 1 ´100 ´100 ´100 ´100
q0 1090 1090 1090
= 100% = 90.64% = 119.82%
55
2.6 Methods of Constructing Index Numbers

The methods of calculating an index number are broadly classified into two categories,
one based on price relatives and the other on aggregative. As calculated above, a price
relative is simply the current price of a commodity divided by the base price. Like the
index itself, it is expressed as a percentage. Thus, the price relatives of different
commodities are averaged in some way to get index number. The other method is to
consider the aggregate of goods, e.g. the goods actually bought or sold in a given year and
compare their value at base year price. To study these methods in greater details, we may
represent them by the following chart.
Index Number
Unweighted (Simple) Weighted
Simple Average of Simple Average of

aggregate relatives aggregate relatives
2.6.1 Unweighted (Simple) Index Number

2.6.1.1 Simple Aggregate Method
This is the simplest method of construction of index number. In this method, we express
the total of commodity prices in the given year as a percentage of the total commodity
prices in the base year. Thus:
Sp1
Simple aggregative price index p01 = ´100
Sp0
where p01 = Index number of the current year

Sp1 = Sum of all commodity prices in current year
Sp0 = sum of all commodity prices in base year

Example 3: Obtain index numbers for 2000 taking 1998 as the base year by using simple
aggregate method.
Commodities Prices
1998 2000
A 100 140
B 80 120
C 160 180
D 220 240
E 40 40
56
Solution: Computation of Index Number (Simple Aggregate Method)
Commodities Prices
1998 (p0 ) 2000 (p1)
A 100 140
B 80 120
C 160 180
D 220 240
E 40 40
Total Sp0 = 600 Sp1 = 720
\ Price index number of 1998

Sp1 720
= P01 = ´100 = ´100 = 120
Sp0 600
Example 4: For the data given below, calculate the index number by taking:
(i) 1990 as the base year.
(ii) 1997 as the base year.
(iii) 1990 to 1992 as the base period.
Year Price of wheat (per kg) Year Price of wheat (per kg)
1990 4 1995 10
1991 5 1996 9
1992 6 1997 10
1993 7 1998 11
1994 8
Solution: (i) Index number taking 1990 as the base year.
Year Price of wheat (per kg) Index numbers (1990 = 100)
1990 4 100
1991 5 5
´100 = 125
4
57
1992 6 6
´100 = 150
4
1993 7 7
´100 = 175
4
1994 8 8
´100 = 200
4
1995 10 10
´100 = 250
4
1996 9 9
´100 = 225
4
1997 10 10
´100 = 250
4
1998 11 11
´100 = 275
4
(ii) Index numbers taking 1997 as the base year.
Year Price of wheat (per kg) Index numbers (1997 = 100)
1990 4 4
´100 = 40
10
1991 5 5
´100 = 50
10
1992 6 6
´100 = 60
10
1993 7 7
´100 = 70
10
1994 8 8
´100 = 80
10
1995 10 10
´100 = 100
10
1996 9 9
´100 = 90
10
1997 10 10
´100 = 100
10
1998 11 11
´100 = 110
10
58
(iii) Index numbers taking 1990 to 1992 as the base year. In this case we have to take an
average of prices in 1990, 1991, 1992:
4+5+6
Average = =5
3
Year Price of wheat (per kg) Index numbers (1990 to 1992)
1990 4 4
´100 = 80
5
1991 5 5
´100 = 100
5
1992 6 6
´100 = 120
5
1993 7 7
´100 = 140
5
1994 8 8
´100 = 160
5
1995 10 10
´100 = 200
5
1996 9 9
´100 = 180
5
1997 10 10
´100 = 200
5
1998 11 11
´100 = 220
5
2.6.1.2 Average of Price Relative Method

This method is based on price relatives for all the commodities included in the index
number. We know that price relative is a ratio of the price of a single commodity in a
given period to its price in base period and it is denoted by p1 / p0 . These price relatives
for different commodities are averaged by using any measure of central tendency (A.M.,
G.M. or median). If arithmetic mean is used, the formula for computing index number is:
å p01 ´100
p
P01 =
N
å p0 is the sum of price relatives of N commodities. If geometric mean is used, the

p1
Here
formula for computing index number is:

59
ì æ p öü
ïS log çç 1 ´100 ÷÷ ï
ï
P01 = Antilog í è p0 øï
ý
ï N ï
ï ï
î þ
ìS log P ü p1
= Antilog í ý, P = ´100
î N þ p0
Example 5: Construct index number for 2000 taking 1998 as the base year for the
following data using average of price relatives.
Commodities Wheat Rice Arhar Milk Clothing
Prices in (Rs) 1998 100 80 160 220 40
2000 140 120 180 240 40
Solution: Construction of price index numbers (Simple average of price relatives)
Commodities Price in 1998 Price in 2000 p

Price relative 1 ´100
p0
p0 p1
Wheat 100 140 140

´100 = 140.00
100
Rice 80 120 120

´100 = 150.00
80
Arhar 160 180 180

´100 = 112.50
160
Milk 220 240 240

´100 = 10910
.
220
Clothing 40 40 40
´100 = 100
40
N =5 æ p ö
å ççè p01 ´100 ÷÷ø = 61160
.
Using arithmetic mean of the price relatives, we get the following, price index number:
æ p ö
å ççè p01 ´100 ÷÷ø 61160
.
Price Index No. for 1998 = P01 = = = 122.32
N 5
60
Example 6: Using geometric mean, construct index numbers by the simple average of
relative method from the following data for 1998 and 1999 with 1997 as the base year.
Commodity Price (in Rs. per unit)
1997 1998 1999
A 100 120 150
B 40 45 60
C 30 35 45
D 10 12 15
E 20 22 23
Solution: Construction of price index number by using G.M. (base 1997)
Commodity Price (Rs) æ p ö log P1 p log P2

P1 = ç 1 ´100 ÷ P2 = 2 ´100
çp ÷ p0
1997 1998 1999 è 0 ø
p0 p1 p2
A 100 120 150 120 2.0792 150 2.1761

´100 ´100
100 100
= 120 = 150.0
B 40 45 60 45 2.0511 60 2.1761
´100 ´100
40 40
= 112.5 = 150.0
C 30 35 45 35 2.0671 45 2.1761
´100 ´100
30 30
= 116.7 = 150.0
D 10 12 15 12 2.0792 15 2.1761
´100 ´100
10 10
= 120.0 = 150.0
E 20 22 23 22 2.0414 23 2.0607
´100 ´100
20 20
= 110.0 = 115.0
N =5 SP1 = 579 .2 S log P1 SP2 = 715 Slog P2

= =
10.3180 10.7651
61
Using geometric mean of the price relatives, we get the following index numbers.
ìS log P1 ü ì10 .3180 ü
Price index for 1998 = Antilog í ý = A. L í ý
î N þ î 5 þ
= A.L. (2.0636) = 115.8
ìS log P2 ü ì10.7651ü
Price index for 1999 = Antilog í ý = A.L. í ý
î N þ î 5 þ
= A.L. (2.1530) = 142.2
2.6.2 Weighted Index Numbers

Weighted index numbers are those index numbers in which comparative or relative
importance is assigned to different items. As such these index numbers are considered
more rational and logical. These index numbers are also classified as:
1. Weighted Aggregate Index Numbers
2. Weighted Average of Relatives
2.6.2.1 Weighted Aggregate Index Numbers

In this method, appropriate weights are assigned to various items which reflect their
relative importance. A reasonable assumption is to consider the quantities consumed or
produced as weights. If W is the weight attached to commondity then a general weighted
price index can be formulated as under:
Sp W
Weighted Aggregative price Index = p01 = 1 ´100 ... (*)
Sp0 W
By using different type of weights, we have different formulae for index numbers. These
weights may be price or quantity of base year or current year or both. Following are the
different formulae for index number:
1. Laspeyre's Index Number
2. Paasche's Index Number
3. Marshall Edgeworth's Index Number
4. Fisher Ideal Index Number
5. Dorbish and Bowley's Index Number
6. Walsh Index Number
1. Laspeyre's Index Number: In this method quantities of the base year i.e., q0 is
taken as weights. The formula is:
Sp1 q0
P01 = ´100
Sp0 q0
62
2. Paasche's Index Number: In this method quantities of the current year i.e., q1 is
taken as weights. The formula is:
Sp1q1
P01 = ´100
Sp0 q1
3. Marshall Edgeworth's Index Number: This formula was suggested by Fisher and
was supported by Marshall Edgeworth. In this method average quantity of current
and base year is taken as weights. The formula is
S(q0 + q1) p1
P01 = ´100
S(q0 + q1) p0
4. Fisher's Ideal Index Number: In this method the geometric mean of Laspeyre's
and Paasche's formulae is taken. The formula is
é Sp q Sp q ù
P01 = ê 1 0 ´ 1 1 ú ´100
êë Sp0 q0 Sp0 q1 úû
5. Dorbish and Bowley's Index Number: In this method, the arithmetic mean of
Laspeyre's and Paasche's formulae is taken. The formula is
1 é Sp1 q0 Sp1q1 ù
P01 = ê + ú ´100
2 êë Sp0 q0 Sp0 q1 úû
6. Walsh Index Number: In this method, the geometric mean of the current and base
year quantities is taken as weight. The formula is
S q1q0 p1
p01 = ´100
S q1q0 p
0
Example 7: The prices per unit and the number of units consumed for three commodities
A, B and C for two time periods are given below:
Commodity Base year Current year
Price Quantity Price Quantity
A 1 6 3 5
B 3 5 8 5
C 4 8 10 6
Construct price index numbers using:

(i) Paasche's Index Number formula.
(ii) Laspeyre's Index Number formula. [C.C.S. Meerut 2002]
63
Solution: Construction of Price Index number
Commodities Base year Current year p0 q0 p0 q1 p1q0 p1q1
p0 q0 p1 q1
A 1 6 3 5 6 5 18 15
B 3 5 8 5 15 15 40 40
C 4 8 10 6 32 24 80 60
Total 53 44 138 115
Sp1q1 115
(i) Paasche's Formula: P01 = ´100 = ´100 = 261.36
Sp0 q1 44
Sp q 138
(ii) Laspeyre's formula: P01 = 1 0 = ´100 = 260.38
Sp0 q0 53
Example 8: Compute index numbers by Marshall-Edgeworth's method, Dorbish- Bowley

method and Fisher's ideal formula on the basis of data of example 7.
Solution: (i) Marshall-Edgeworth formula:
Sp1 (q0 + q1) é Sp q + Sp1q1 ù

P01 = ´100 = ê 1 0 ú ´100
Sp0 (q0 + q1) ëê Sp0 q0 + Sp0 q1 úû
é138 + 115 ù 253

=ê ú ´100 = ´100 = 260.82
ë 53 + 44 û 97
(ii) Dorbish-Bowley formula:

1 é Sp1q0 Sp q ù
P01 = ê + 1 1 ú ´100
2 êë Sp0 q0 Sp0 q1 úû
1 é138 115 ù 1
= ê + ú ´100 = [2.60 + 2.61] ´100
2 ë 53 44 û 2
1
= × 5.3 × 100 = 265.0
2
(iii) Fisher's Ideal formula:
é Sp q Sp q ù
P01 = ê 1 0 ´ 1 1 ú ´100
æ138 115 ö
= ç ´ ÷ ´100
è 53 44 ø
= 2.60 ´ 2.61 ´100 = 6.786 ´100
= 2.60 × 100 = 260.0

64
2.6.2.2 Weighted Average of Relatives Method

We have already discussed average of relatives method. This method is used to overcome
the limitations of the simple average of relatives method. In this method the price
p
relatives for the current year are calculated on the basis of the prices of base year 1 ´100
p0
These price relatives are assigned some weights W. The weights are multiplied by the
price relatives and these products are added up and are divided by the sum of the weights.
Symbolically:
æ p ö
å ççè p01 ´100 ÷÷ø W SPW
Weighted Index Number = P01 = =
SW SW
p1
where ´100 = P
p0
Similarly, the index number based on geometric mean of price relatives is given by:
æ SW log P ö
P01 = Antilog ç ÷
è SW ø
Example 9: An enquiry into family budgets of middle class family gave the following
information:
Items Food Rent Clothing Fuel Others
Expenditure % 30% 15% 20% 10% 25%
Price (Rs) in 1999 100 20 70 20 40
Price (Rs) in 2000 90 20 60 15 35
Compute the price index for 2000 by using

(i) weighted arithmetic mean of price relatives
(ii) weighted geometric mean of price relatives.
Solution: Computation of Weighted Index
Item Weights p0 p1 Price WP log P W log P

W 1999 2000 relatives
p
P = 1 ´100
p0
Food 30 100 90 90.0 2700.00 1.9542 58.626
Rent 15 20 20 100.0 1500.00 2.0000 30.000
Clothing 20 70 60 85.7 1714.00 1.9330 38.660
Fuel 10 20 15 75.0 750.00 1.8751 18.751

65
Others 25 40 35 87.5 2187.50 1.9420 48.550
SW =100 SWP SW log P

= 8851.50 = 194.587
(i) Index number based on weighted A.M. of price relatives:

SWP 8851.50
P01 = = = 88.515
SW 100
(ii) Index number based on weighted G.M. of price relatives:
é SW log P ù é194.587 ù
P01 = Antilog ê ú = Antilog ê ú
ë SW û ë 100 û
= Antilog (1.94587)
= 8826
.
2.7 Test for Index Numbers

Prof. Fisher has made a careful study of different formulae which are used to compute
index numbers. He suggested that a good formula should satisfy the following tests:
1. Time Reversal Test
2. Factor Reversal Test
3. Circular Test
2.7.1 Time Reversal Test

In words of Fisher "The formula for calculating an index number should be such that it will give
the same ratio as between one point of comparison and the other, no matter which of the two is taken
as base". That is, product of two index numbers must be equal to one. This follows that a
formula should permit interchange of two times (periods) without causing any change in
the results. As such, an index number should work backward as well as forward. In other
words, if the two periods, the base and the current (reference) periods are interchanged,
the product of the two index numbers should be unity, i.e., the two index numbers should
be reciprocal of each other. Since two index numbers should be reciprocal of each other,
their product should always be equal to 1. Hence: P01 = 1 / P10 or P01 ´ P10 = 1
Where : P01 = Price change in current year.
P10 = Price change in base year (based on current year)
Let us see whether Laspeyre's index number satisfy this property or not. For which
Laspeyre's formula is:
Sp1q0
P01 =
Sp0 q0
Sp0 q1
\ P10 =
Sp1q1
66
Sp1q0 Sp0 q1
Now P01 ´ P10 = ´ ¹1
Sp0 q0 Sp1q1
So, Laspeyre's formula does not satisfy the time reversal test. Similarly Paasche's index
formula does not satisfy this test. But if we consider, Fisher's formula
é Sp q Sp q ù
P01 = ê 1 0 ´ 1 1 ú
é Sp q Sp q ù
and thus P10 = ê 0 1 ´ 0 0 ú
êë Sp1q1 Sp1q0 úû
Now
é Sp q Sp q ù é Sp q Sp q ù
P01 ´ P10 = ê 1 0 ´ 1 1 ú ´ ê 0 1 ´ 0 0 ú
êë Sp0 q0 Sp0 q1 úû êë Sp1q1 Sp1q0 úû
=1
Thus, Fisher's index number satisfies time reversal test. It can be seen that
Marshall-Edgeworth's and Walsh's index numbers also satisfy time reversal tests.
2.7.2 Factor Reversal Test

In the words of Fisher "Just as a formula should permit interchange of two times without
giving inconsistent results, so it ought to permit interchanging the prices and quantities
without giving inconsistent results i.e., the two results multiplied together should give
the true value ratio."
This means that as a formula should permit interchange of two times, it should also
permit interchange of price and quantities without causing any change in the ratio of
their true value. In its simplest form we conclude that a change in price (denoted by P01)
is always followed by a change in quantity denoted by q01. Symbolically, the factor
reversal test is satisfied if
Sp q Value in the given year (1)
P01 ´ Q01 = 1 1 =
Sp0 q0 Value in the base year (0)
Let us see whether the Laspeyre's index satisfied factor reversal test or not. For which:
Sp1q0 Sq1 p0
P01 = and Q01 =
Sp0 q0 Sq0 p0
Thus,
Sp1q0 Sq1 p0 Sp1q1
P01 ´ Q01 = ´ ¹
Sp0 q0 Sq0 p0 Sp0 q0
Thus, Laspeyre's index does not satisfy factor reversal test. Let us try Fisher's index, we
have
67
é Sp q Sp q ù
P01 = ê 1 0 ´ 1 1 ú
é Sq p0 Sq1 p1 ù
and Q01 = ê 1 ´ ú
êë Sq0 p0 Sq0 p1 úû
é Sp q Sp q ù é Sq p0 Sq1 p1 ù
\ P01 ´ Q01 = ê 1 0 ´ 1 1 ú ´ ê 1 ´ ú
S
êë 0 0
p q S p0 1 úû
q êë Sq0 p0 Sq0 p1 ûú
Sp1 q1
=
Sp0 q0
Thus, Fisher's index also satisfies the factor reversal test. Since Fisher's index number
satisfies both the test, it is called Fisher's ideal index number.
2.7.3 Circular Test

It is often desired to shift the base. Hence the formula for the construction of an index
number should be such that it ensures that new index number which may be obtained by
the shifting process should be equal to the original index number. Thus:
P01 ´ P02 ´ P12 = 1
where P01 = Price change of the current year on the base year.
P02 = Price change of the current year on the second base year.
P12 = Price change of the base year on some other base.
This test of base shifting is called "Circular Test". It is a sort of extension of Time Reversal
Test.
Note: The most common formula i.e. Fisher's ideal formula fails to satisfy this test.
However, index numbers obtained by simple aggregate method or by fixed weight
aggregative method satisfy this test.
Example 10: Calculate Fisher's index number from the data given below and show that it
satisfies the time-reversal and factor reversal test:
A 12 10 15 12
B 15 5 20 5
C 34 7 20 9
D 5 16 5 14

68
Solution: Calculation of Fisher's ideal Index Number
Commodity Base year Current year p0 q0 p0 q1 p1q0 p1q1
p0 q0 p1 q1
A 12 10 15 12 120 144 150 180
B 15 5 20 5 75 75 100 100
C 34 7 20 9 238 306 140 180
D 5 16 5 14 80 70 80 70
513 595 470 530
é Sp q Sp q ù
Fisher's ideal index = P01 = ê 1 0 ´ 1 1 ú ´100
é 470 530 ù
= ê ´ ´100
ë 513 595 úû
é 249100 ù
= ê ´100
ë 305235 úû
= 0.9033 ´100 = 90.33

For time reversal test:
é Sp q Sp1 q1 ù é 470 530 ù

P01 = ê 1 0 ´ ú= ê ´
êë Sp0 q0 Sp0 q1 úû ë 513 595 úû
é Sp q Sp q ù é 595 513 ù
P10 = ê 0 1 ´ 0 0 ú = ê ´
According to test,
é 470 530 ù é 595 513 ù

P01 ´ P10 = ê ´ ´ ê ´ =1
ú
ë 513 595 û ë 530 470 úû
Thus, Fisher index satisfies time reversal test.

Factor Reversal Test: The test is satisfied if
Sp1 q1
P01 ´ Q01 =
Sp0 q0
é Sp q Sp q ù é 470 530 ù
Here, P01 = ê 1 0 ´ 1 1 ú = ê ´
é Sq p0 Sq1 p1 ù é 595 530 ù

Q01 = ê 1 ´ ú= ê ´
êë Sq0 p0 Sq0 p1 ûú ë 513 470 úû
69
é 470 530 ù é 595 530 ù 530

\ P01 ´ Q01 = ê ´ ´ ê ´ =
ú
ë 513 595 û ë 513 470 úû 513
Sp1 q1
=
Sp0 q0
Thus, Fisher's index satisfies Factor Reversal test.

Example 11: The following figures relate to the prices and quantities of certain
commodities. Construct an appropriate index number and show if it satisfies the time
reversal test.
Commodities 1995 1996
Wheat 30 50 32 50
Gram 25 40 30 35
Barley 18 50 16 55
Solution: Computation of index number by Fisher's ideal index number.
Commodities 1995 1996 p1q0 p0q0 p1q1 p0q1
p0 q0 p1 q1
Wheat 30 50 32 50 1600 1500 1600 1500
Gram 25 40 30 35 1200 1000 1050 875
Barley 18 50 16 55 800 900 880 990
Total 3600 3400 3530 3365
Using the formula

Sp1 q0 Sp1 q1
P01 = ´ ´100
Sp0 q0 Sp0 q1
æ 3600 3530 ö
= ç ´ ÷ ´100
è 3400 3365 ø
= 1111
. ´100 = 1054
. ´100 = 105.4
Time reversal test is satisfied if

P01 ´ P10 = 1
Sp1 q0 Sp1 q1 3600 3530
P01 = ´ = ´
Sp0 q0 Sp0 q1 3400 3365
Sp0 q1 Sp0 q0 3365 3400

p10 = ´ = ´
Sp1 q1 Sp1 q0 3530 3600
70
3600 3530 3365 3400

\ P01 ´ P10 = ´ ´ ´
3400 3365 3530 3600
= Ö1 =1
Thus, time reversal test is satisfied by Fisher's ideal index number.
Example 12: From the following table show that Fisher's index number satisfies the time
reversal tests.
A 6 50 10 60
B 2 100 2 120
C 4 60 6 60
[C.C.S. Meerut 2005, 06]
Solution: Computation of index number of Fisher's ideal index number.
Commodity Base year Current year p1q0 p0q0 p1q1 p0q1
p0 q0 p1 q1
A 6 50 10 60 500 300 600 360
B 2 100 2 120 200 200 240 240
C 4 60 6 60 360 240 360 240
Total 1060 740 1200 840
Time reversal test is satisfied by Fisher's formula if:

Sp1 q0 Sp1 q1 1060 1200
P01 = ´ = ´
Sp0 q0 Sp0 q1 740 840
Sp0 q1 Sp0 q0 840 740

and P10 = ´ = ´
Sp1 q1 Sp1 q0 1200 1060
1060 1200 840 740

\ P01 ´ P10 = ´ ´ ´ =1
740 840 1200 1060
Thus time reversal test is satisfied.
Factor reversal test is satisfied if
Sp1 q1
P01 ´ Q01 =
Sp0 q0
Sp1 q0 Sp1 q1 1060 1200

Here; P01 = ´ = ´
Sp0 q0 Sp0 q1 740 840
71
Sq1 p0 Sq1 p1 840 1200

Q01 = ´ = ´
Sq0 p0 Sq0 p1 740 1060
1060 1200 840 1200 1200

\ P01 ´ Q01 = ´ ´ ´ =
740 840 740 1060 740
Sp1 q1
=
Sp0 q0
Thus, fisher's index number satisfies factor reversal test.
2.8 Quantity Index Number

Instead of comparing the relative change in prices we may be interested to study the
relative change in quantity. In other words, if we study the changes in quantity of
production or consumption of certain commodities over a given period of time, we call
them quantity index number Q01. Accordingly the formulae may be listed as under :
Sq1
1. Q01 = ´100; Simple aggregative quantity index
Sq0
Sq1 / q0
2. Q01 = ´100; Simple average of relative method
N
Sq1 p0
3. Q01 = ´100; Laspeyre's quantity index using base year prices as
Sq0 p0
Sq1 p1
4. Q01 = ´100; Paasche's quantity index using given year prices as
Sq0 p1
Similarly, Fisher's, Marshall, Edgeworth's formula for quantity index numbers can also be
obtained.
2.9 Value Index Number

The value is defined as the product of price and quantity i.e.,
Value = Price × Quantity
If p0 and q0 denote the price and quantity of a commodity during a base period while p1
and q1 denote the price and quantity of the commodity during a given period, the total
values during these periods are given by p0 q0 and p1q1 respectively and we define.
p q æ p öæq ö
Value Relative = 1 1 = ç 1 ÷ ç 1 ÷ = price relative × quantity relative
p0 q0 çè p0 ÷ø çè q0 ÷ø
Just as we have obtained price index numbers and quantity index numbers, we can
convert the above formula for finding out the value index number also. The simplest such
index is:
72
Sp1 q1
Value index = ´100
Sp0 q0
As the values have not been weighted, the value index numbers are the simple aggregative
index numbers.
2.10 Chain Based Index

The index numbers we have considered so far are of the fixed base type, i.e., the base
period with which we compare the other time periods remains fixed with the progress of
time. We also know that with the passage of time new commodities enter the market and
old ones disappear, besides, the quality of the commodities may undergo a change. Also,
the relative importance of various commodities is dependent on the tastes and habits of
the consumer's changes. If an index number is needed for comparing successive time
periods, say 0, 1, ... x, it is not necessary to use a fixed base0. We use the previous period
as base for comparing any time period and construct what are called "Link Relatives".
There is no change in calculation, only the base period changes for each comparison and
in each case it is the previous period. The symbol used for such an index for comparing
the prices of period k with those of (k -1) is Pk -1, k . Thus we construct n link indices
- P01, P12 ,... Pn -1, n. By multiplying successive links i.e., by chaining we obtain the chain
indices as;
P01 = First link relative
P02 = P01 ´ P12
P03 = P01 ´ P12 ´ P23
...................................
...................................
P0 n = P01 ´ P12 ´ ... ´ I n -1, n
The practical advantage of a chain index is that the sample of commodities and/or the set
of weights may be kept quite up-to-date in any index number.
2.11 Relative Merits and Demerits of Chain Base an

Fixed Base Method
We have seen that fixed based index numbers become more and more inaccurate as the
distance between the base period and the current period increases. As the chain based
index numbers are based on a number of link indices, each of which is expected to be
quite accurate, it is claimed that chain base index numbers are more accurate than the
fixed base ones, so far as long term comparison is concerned.
73
Fixed base index numbers are generally easier to calculate and are more easily understood
by users of index numbers than chain base index numbers. A further disadvantage of the
chain-base method is that any defect or abnormality in the index for one year is
perpetuated in all subsequent years. The chain is only as strong as its weakest link. Also if
there is any bias in the formula employed, e.g., the A.M. of price relatives, the cumulative
effect of that bias may become serious as time goes on.
Example 13: Prepare chain base index numbers from the following data:
Year 1996 1997 1998 1999 2000
Price (Rs): 80 120 132 264 396
Solution:
Year Price (Rs) p ´100 Index number

Calculation 1
p0
1996 80 — 100
1997 120 120 ´100 150

80
1998 132 132 ´100 110

120
1999 264 264 ´100 200

132
2000 396 396 ´100 150

264
2.12 Base Conversion

In some specific situations, we also need to convert chain base index numbers to fixed
based index numbers. Base conversion may be of two types:
1. From fixed base to chain base and
2. From chain base to fixed base.
2.12.1 From Fixed Base to Chain Base

For such conversion the chain index for the first year will be 100 and for next years the
following formula is used:
Current year chain Base Index Number
Current year' s fixed base Index Number
= ´100
Previous year' s fixed base Index Number
74
Example 14: From the fixed base index numbers given below, prepare chain base index
numbers.
Year: 1998 1999 2000 2001 2002 2003
Index No.: 188 196 204 190 196 200
Solution:
Year Fixed base index no. Conversion Chain base index no.
1998 188 — 100
1999 196 196 ´100 104.26

188
2000 204 204 ´100 104.08

196
2001 190 190 ´100 93.14

204
2002 196 196 ´100 103.16

190
2003 200 200 ´100 102.04

196
For conversion we use the following formula

Current year' s fixed base index number
= ´100
Previous year' s fixed base index number
2.12.2 From Chain Base to Fixed Base

In this type of conversion, the fixed base index for the first year will be taken the same as
the chain base index. however, if it given specifically in the question that the first year is
to be taken as base, the index number for the first year will be 100. For calculating the
indices for other years the following formula will be used:
Current year fixed base index number
Current year chain index number ´ Previous year' s fixed base index number
=
100
Example 15: Construct fixed-base index numbers from the following chain base index
numbers:
Year: 1979 1980 1981 1982 1983 1984
Chain Index: 94 104 104 93 103 102
Solution: Conversion of chain base index number into fixed base index number.
75
Year Chain base index no. Conversion Fixed base index no.
1979 94 — 94
1980 104 94 ´104 97.76

100
1981 104 . ´104

9776 101.67
100
1982 93 . ´ 93
10167 94.55
100
1983 103 94.55 ´103 97.39

100
1984 102 97.39 ´102 99.34

100
Example 16: Construct fixed-base index numbers from the following chain base index
numbers:
Year 1980 1981 1982 1983 1984 1985
Chain Index Numbers 105 75 71 105 95 90
Solution: Conversion of chain base index number into fixed base index number.
Year Chain base index no. Conversion Fixed base index no.
1980 105 — 105
1981 75 105 ´ 75 78.75

100
1982 71 . ´ 71
7875 55.91
100
1983 105 55.91 ´105 58.71

100
1984 95 . ´ 95
5871 55.77
100
1985 90 55.77 ´ 90 50.20

100
76
2.13 Cost of Living Index Numbers

Amongst all the index numbers, the cost of living index numbers are perhaps the most
well-known indices. They are constructed to study the effect of changes in the prices of a
basket of goods and services on the purchasing power of a particular class of people
during current period as compared with some base period. These index numbers
represent the average change in prices, paid by ultimate consumer of particular goods and
services over a period of time. As such, they are also called "Consumer Price Index
Numbers."
The need to construct cost of living index numbers arises due to the fact that wholesale
general index numbers have failed to give an exact idea of the effects on different classes
of people in a society for any rise or fall in prices. The wholesale price index numbers
measure variations in general price level whereas cost of living index numbers throw light
on the effects on living conditions of different classes of people in a community for any
variation in prices.
2.14 Construction of Cost of Living Index Numbers

The following considerations are necessary in the construction of cost of living index
numbers.
1. First of all we must decide the class of people for which the index number is to be
constructed. As the consumption pattern differs from class to class, the class for
which index number is to be constructed must be, as far as possible, a homogeneous
group containing the people of more or less the same income group.
2. Next, we conduct a sample budget inquiry of the class of people selected on the
above considerations. That is, we select a sample of people which is the best
representative of the class of people and enquire about their budget i.e, income and
expenditures on various items made during a normal year. These items may be food,
clothing, fuel and lighting, rent and miscellaneous. Food can further be sub-divided
into wheat, rice, pulses, ghee, tea, milk, etc. This budget enquiry provides the
following informations.
(i) The nature, quality and quantity of commodities consumed by the people.
(ii) The retail prices of the different commodities.
(iii) The proportion of expenditure on each individual item of expenditure. For
example, proportion of expenditure on wheat to the food expenditure can be
obtained.
77
(iv) The proportion, which expenditure on each group bears to the total
expenditures, i.e., we may know the proportion of expenditure on food to the
total expenditure on all items.
The sample budget inquiry forms the basis of cost of living index numbers, since this
provides the information which are the basic needs for the construction of the cost of
living index numbers.
Note: 1. Here only those commodities must be included which are popularly used
among the class of people chosen for the purpose.
2. Retail price quotations should be obtained from the localitites in which the
selected class of people reside.
The construction of cost of living index numbers suggest the following two ways.
(i) Method of Aggregates
(ii) Method of weighted relatives or method of family budget.
2.14.1 Method of Aggregates

In this method the following steps are taken:
1. The total expenditure of base year is obtained by multiplying base year price (p0 ) by
the quanitty of base year (q0 ). The sum of p0 q0 i.e., Sp0 q0 is called total expenditure.
While calculating Sp0 q0 it should be carefully observed that units of quantity and
price are the same.
2. Next, the price of current year ( p1) is multiplied by quantity of base year (q0 ) and the
total expenditure on base of quantity i.e. Sp1q0 is obtained.
3. The following formula is used to calculate the weighted cost of living index number.
Total expenditure in current year
Cost of living index = ´100
Total expenditure in base year
Sp1 q0
= ´100
Sp0 q0
Note: This is nothing but Laspeyre's Index and is the most popular method of
constructing cost of living index number.
2.14.2 Method of Weighted Relatives or Method of Family Budget

In this method the price relatives (P) of current year is calculated by the following
p
formula = 1 ´100
p0
where p1 = Price of current year, p0 = Price of base year.

78
Then, the price relatives (P) so obtained is multiplied by weights (W ) of the commodity
and SPW is determined. If weights are clearly given in the question, then these weights
are multiplied by these price relatives. The following formula is used to calculate the
index number.
SPW
Cost of living index number =
SW
Note: Here the index numbers are not multiplied by 100 since the price relatives have
already been multiplied by 100.
Example 17: From the following data prepare, cost of living index number by aggregate
expenditure method:
Commodity Unit Quantity Price in base year, Price in current

consumed in base 1990 year, 2000
year
Wheat per quintal 2 quintal 400 800
Rice per quintal 1 quintal 600 900
Gram per quintal 2 quintal 1200 1800
Pulses per quintal 1 quintal 2000 2600
Sugar per kg 50 kg 13 16
Fuel pre quintal 5 quintal 80 90
Rent per room 1 room 600 800
Solution: Construction of cost of living index nos, by aggregate expenditure method.
Commodity Unit Quantity Price in Price Aggregate Aggregate

consumed base in expenditure expenditure
in base year current baseyear in current
year p0 year p0 q0 year
q0 p1 p1q0
Wheat per quintal 2 quintal 400 800 800 1600
Rice per quintal 1 quintal 600 900 600 900
Gram per quintal 2 quintal 1200 1800 2400 3600
Pulses per quintal 1 quintal 2000 2600 2000 2600
Sugar per kg 50 kg 13 16 650 800

79
Fuel per quintal 5 quintal 80 90 400 450
Rent per room 1 room 600 800 600 800
Total 7450 10750
Sp1 q0 7450
Cost of living index = ´100 = ´100 = 69.30
Sp0 q0 10750
Example 18: Construct the cost of living index no. by family budget expenditure method
from the data in example 17.
Solution: Construction of cost of living index numbers by family budget method.
Commodity Unit Quantity Base Current Price Weights PW

base year year year relatives p0 q0 = W
q0 price price P
p0 p1
Wheat Per quintal 2 quintal 400 800 200.00 800 160000
Rice Per quintal 1 quintal 600 900 150.00 600 90000
Gram Per quintal 2 quintal 1200 1800 150.00 2400 360000
Pulses Per quintal 1 quintal 2000 2600 130.00 2000 260000
Sugar per kg 50 kg 13 16 123.08 650 80002
Fuel Per quintal 5 quintal 80 90 112.5 400 45000
House Rent per room 1 room 600 800 133.33 600 79998
7450 1075000
p
where P = 1 ´100
p0
SPW 1075000
Weighted cost of living index no. = = = 144.30
SW 7450
Miscellaneous Examples
Example 19: A worker earned Rs. 3,000 per month in 2004. The cost of living index
increased by 70% between 2004 and 2007. How much extra income should be worker
have earned in 2007 so that he could buy the quantities as in 2004 ?
Solution: In 2004, when index was 100, worker's wages was ` 3,000 p.m.
In 2007, when index is 170, his wages should be
3,000
= ´170 = ` 5,100 p.m.
100
80
His extra income should be:

= 12 (` 5,100 – ` 3,000) [Q 1 Year = 12 months]
= 61,200 – 36,000
= ` 25,200 in the year 2007.
Thus, if the worker has earned ` 25,200 as extra income during the year 2007, he will be
able to buy the same quantities as in 2004.
Example 20: From the following data of cost of living index number of labourers, it is
proposed to find the D.A. to workers in such a way as to compensate fully the rise in prices
of food and clothing. What would be the D.A. expressed as a percentage of wages.
Items Indices Weights
Food 300 50
Clothing 350 10
Fuel 200 5
Rent 280 15
Miscellaneous 140 20
Solution: In this question increase in price of food and clothing is to be compensated.

Thus, the indices for fuel, rent and miscellaneous are to be taken equal to the base period,
i.e., 100.
Groups Index No. P Weights W PW
Food 300 50 15,000
Clothing 350 10 3,500
Fuel 100 5 500
Rent 100 15 1,500
Miscellaneous 100 20 2,000
SW =100 SPW = 22,500
SPW 22,500
Index No. = = = 225
SW 100
Thus, to compensate the increase in price of food and clothing (225 – 100) = 125, D.A.
should be fixed.
81
Example 21: In 2001, food was selling at Rs. 1,000 per quintal, cloth Rs. 75 per yard,
house rent Rs. 1,500 per house and other items at Rs. 100 per unit. By 2007, cost of food
increased by Rs. 200 per quintal, house rent by Rs. 500 per house, and other items
doubled in price. The cost of living index for 2007 (with 2001 as base) was 150. By how
much the cloth increased in price during 2001-2007?
Solution: Suppose the increase in price of cloth is x.
Index No. for 2005
Commodity 2001 2005
Price Index No. Price Index No.
Food 600 100 800 800

´100 = 133.33
600
Cloth 75 100 x x
´100 = 1.333 x
75
Rent 1,500 100 2,000 2,000

´100 = 133.33
1500
,
Miscellaneous 100 100 200 200

´100 = 200
100
466.67 + 1.333 x
The given index number for 2005 is 150. Hence, the sum of the index number of all
commodities will be 150 × 4 = 600. Thus,
466.67 + 1.333 x = 600
1.333 x = 600 – 466. 67
1.33 x = 133. 33
\ x =100.25
Therefore, the rise in the price of cloth was 100.25 – 75 = Rs. 25.25 per yard.
Example 22: The sub-group indices of the consumer price index number for urban employees
of an industrial centre for a particular year (with base 2000 = 100) were as follows:
Food 200
Clothing 130
Fuel and Lighting 120
House Rent 150
Miscellaneous 140
The weights are 60, 8, 7, 10 and 15 respectively. It is proposed to fix dearness allowance
in such a way so as to compensate fully the rise in the price of food and house rent.
What should be the dearness allowance expressed as a percentage of wages ?
82
Solution: As in question 20, the procedure applied is as follows:
Groups Index No. P Weights W PW
Food 200 60 12,000
Clothing 100 8 800
Fuel and Lighting 100 7 700
House Rent 150 10 1,500
SW =100 SPW =16,500
SPW 16,500
Index No. = = = 165
SW 100
Thus, to compensate the increase in price of food and house rent (165 – 100) = 65%,
D.A. should be fixed.
Example 23: The consumer price index over a certain period increased from 120 to 215
and the wages of a worker increased from Rs. 1,680 to Rs. 3,000. What is the gain or loss
to the worker ?
Solution:
Period Consumer Price Index Wages
I 210 1,680
II 215 3,000
Wage required in II period to maintain same standard of living

1680
, ´ 215
= = Rs. 3,010.
120
Wage for II period = Rs. 3,000. Thus, loss = Rs. 10.
Example 24: An enquiry into the budgets of middle class families of a certain city revealed
that on an average, the percentage expenses in the different groups were:
Food 45, rent 15, clothing 12, fuel and light 8, miscellaneous 20. The group indices for
current year as compared with a fixed base period were 410, 150, 343, 248 and 285
respectively. Calculate the cost of living index for the current year. Mr. X was getting Rs.
2,400 in the base period and Rs. 4,300 in the current year. State how much he ought to
have received as extra allowance to maintain his former standard of living.
83
Solution:
Commodity % Expenditure, W Index, P PW
Food 45 410 18,450
Rent 15 150 2,250
Clothing 12 343 4,116
Fuel & Light 8 248 1,984
SW =100 SPW = 32,500
SPW 32,500
Indix No. = = = 325
SW 100
2,400 ´ 325
For maintaining same standard of living Mr. X should earn = Rs. 7,800.
100
So he must receive an extra allowance of Rs. 3,500 (i.e., 7,800 – 4,300).
Example 25: Given the following data:
Commodities
A B
p0 1 1
q0 10 5
p1 2 x
q1 5 2
Where p and q respectively stand for price and quantity and subscripts stand for the time
period. Find x, if the ratio between Laspeyre's (L) and Paasche's index number is 28 : 27.
Solution:
Commodity p0 q0 p1 q1 p0q0 p1q0 p0q1 p1q1
A 1 10 2 5 10 20 5 10
B 1 5 x 2 5 5x 2 2x
15 20 + 5 x 7 10 + 2 x
Sp q 20 + 5 x
Laspeyre's Index No. : P01 = 1 0 =
Sp0 q0 15
84
Sp q 10 + 2 x
Passche's Index No. : P01 = 1 1 =
Sp0 q1 7
L 28
Given =
P 27
20 + 5 x 7 28
\ ´ =
15 10 + 2 x 27
140 + 35 x 28
=
140 + 30 x 27
3,780 + 945 x = 4,200 + 840 x

945 x - 840 x = 4,200 - 3,780
105 x = 420
Þ x =4
2.15 Limitation of Index Numbers

Although index numbers are very important tools for the relative measurement of
changes in different fields but in reality they also suffer from certain limitations. These
limitations are:
1. Index numbers are constructed on the basis of sample information, not on the facts
hence they may not be complete.
2. Index numbers just specify the direction of change but do not study about the
reality of facts.
3. Index numbers are constructed for quantitative changes. If changes are of

quantitative type, index numbers may not express them properly.
4. Index numbers indicate only average or approximate trend of changes. Hence, they
should be interpreted keeping this limitation into consideration.
5. Index numbers are not suitable for all purpose.

85
E XERCISE
1. Define an index number and mention its uses. Explain time reversal and factor
reversal tests. Show that Fisher's formula satisfies both these tests.
[C.C.S. Meerut 2003 O.C., 08]
2. What are the uses of index numbers? Describe the method of obtaining a cost of
living index.
3. Define the following index numbers and discuss their merits and demerits.
(i) Paasche's Index Number
(ii) Laspeyre's Index Number
(iii) Fisher's Ideal Index Number [C.C.S. Meerut 2002, 06 (Back)]
(iv) Cost of living Index number [C.C.S. Meerut 2008]
4. State and explain Fisher's ideal formula for price index numbers. Show how it
satisfies the time reversal and factor reversal tests. [C.C.S. Meerut 2002 (Back)]
5. What is an index number ? Explain the various problems in the construction of

index numbers.
6. What are the criteria of a good index number ? Describe in detail index number,
giving reasons why it is called ideal ? [C.C.S. Meerut 2000 (Back) 04]
7. Explain an index number, together with the properties it should have. Describe in
detail, an ideal index number, giving reasons why it is called ideal ?
8. (i) Define an index number and mention its uses. Explain various index number
formulae. [C.C.S. Meerut 2001]
(ii) Describe in detail; an ideal index number, giving reasons why it is called ideal.
9. (i) Define index number. Describe the problems involved in the construction of
index numbers. Explain chain base method in constructing an index numbers.
[Garhwal 2003]
(ii) Explain briefly the technique of satistical quality control charts.

10. Explain the tests of a good index number. Show how the Fisher's ideal index number
satisfies the time reversal and factor reversal tests. [Rohilkhand 2004]
11. (i) Explain in detail the construction of cost of living index numbers.
[C.C.S. Meerut 2004 (Back), 02 (Old), 03 (Old)]
(ii) What is an index number ? Describe any four important points that should be
considered in constructing index number. [C.C.S. Meerut 2005]
86
12. Write short notes on:

(i) Paasche's and Laspeyre's index number. [C.C.S. Meerut 2003]
(ii) Ideal index number
13. What is meant by index number? What are the critiria of a good index numbers?
Why is Fisher's price index number called ideal? Explain [C.C.S. Meerut 2012]
14. Construct an index number for 2000 (base-1990) for the following data using
weighted aggregative index number.
Commodity Weight Prices (`)
Base year (1990) Current year (2000)
A 30 4.25 5.20
B 40 2.95 3.75
C 15 2.15 1.95
D 15 8.85 8.10
15. Construct index numbers of price of 1999 from the following data by (i) Laspeyre's
method (ii) Paasche's method (iii) Marshall-Edgeworth method (iv) Fisher's
method.
Commodity Base year (1994) Current year (1999)
A 10 30 12 50
B 8 15 10 25
C 6 20 6 30
D 4 10 6 20
16. From the following data calculate price index numbers for 2000 with 1990 as base
by (i) Laspeyre's method (ii) Paasche's method (iii) Marshall-edgeworth method
(iv) Fisher's method
Commodity 1990 2000
Wheat 20 8 40 6
Gram 50 10 60 5
Rice 40 15 50 15
Pulses 20 20 20 25
87
17. Explain times reversal and factor reversal tests. Treating 1992 as the base year,
show with the help of the following data that both these tests are satisfied by
Fisher's ideal index number:
Article 1992 1995
A 20 3 30 3
B 8 15 8 12
C 12 2 12 4
D 16 8 24 6
E 12 4 12 12
18. Using the following data, construct Fisher's ideal index and show that it satisfies
factor reversal test and time reversal test.
Commodity Price Quantity
Base year Current year Base year Current year
A 6 10 50 56
B 2 2 100 120
C 4 6 60 60
D 10 12 30 24
E 8 12 40 36
19. Compute Laspeyre's, Paasche's and Fisher's price index numbers for 1995 using the
following data concerning three commodities.
Commodities
A B C
Quantity (kg) 1990 15 5 10
1995 12 4 5
Price (kg) 1990 15 20 4
1995 22 27 7
20. From the following data, which index do you consider appropriate to construct and
why ? Prepare that index.
88
Year Rice Wheat Jowar
Price Quantity Price Quantity Price Quantity
1994 9.3 100 6.4 11 5.1 5
2004 4.5 90 3.7 10 2.7 3
Hint : Compute Fisher's index.
21. Compute Fisher's ideal index number from the following data and check whether
the time reversal test is satisfied.
Price (`) Quantity Price (`) Quantity
P 2 7 6 6
Q 3 6 2 3
R 4 5 8 5
S 5 4 2 4

1. Describe the uses of index numbers [C.C.S. Meerut 2002, 04, (Back)]
2. What is cost of living index numbers. [C.C.S. Meerut 2004, 06 (Old course)]
3. In which price index numbers, base year quantities are used as weights.
4. Which index number is equal to A.M. of Paasche's and Laspeyre's index numbers.
5. Which index number is used to determine the grant of D.A. to the workers ?
[C.C.S. Meerut 2004, 04 (Back)]
6. What do you mean by base shifting ? [C.C.S. Meerut 2004, 04 (Back)]
7. Define Paasche's and Laspeyre's index numbers.[C.C.S. Meerut 2004 (Back), 06 (Old course)]
8. Which index number is equal to G.M. of paasche's and Laspeyre's index numbers ?
[C.C.S. Meerut 2004 (Back), 05]
9. In which price index number A.M. of base and current year quantities are used as
weights ? [C.C.S. Meerut 2006 (Old course)]
10. Explain time reversal and factor reversal test in index numbers.
[C.C.S. Meerut 2000, 02, 02 (Back), 06 (Back)]
11. What is an index number ? What are its main uses. [C.C.S. Meerut 2003]
12. Define chain base and fixed base index numbers.
13. Discuss price and quantity index numbers.

89
14. Define price, quantity and value relatives.
15. How will you choose a base year for constructing an index number.
16. Discuss aggregative type and average type index formula.
17. Distinguish between fixed base and chain base index numbers.
[C.C.S. Meerut 2005, 06 (Old course)]
18. Describe the construction of cost of living index number. [C.C.S. Meerut 2005]
19. Define Bowley's and marshall-Edgeworth's index numbers. [C.C.S. Meerut 2005]
20. Which index number is called an ideal index number. [C.C.S. Meerut 2005, 06 (Old)]
21. Describe cost of living index number. [C.C.S. Meerut 2012]
22. What is chain base index number ? [C.C.S. Meerut 2012]
23. Define Laspeyre's Paache's and Fisher Index numbers. Show that Fisher's formula is
an ideal formula [C.C.S. Meerut 2008]

1. In Paasche's price index number quantity of ... year is taken as weight.
(a) Base (b) Current
(c) Average of base and current (d) Product of base and current
[B.R.A. Agra 2012]
2. Fisher ideal index number formula satisfies:
(a) Time Reversal Test (b) Factor Reversal Test
(c) Both (a) and (b) (d) Neither (a) nor (b) [B.R.A. Agra 2006, 11]
3. Fisher Ideal Index number formula does not satisfy :
(a) Time Reversal Test (b) Factor Reversal Test
(c) Circular Test (d) None of these [B.R.A. Agra 2004, 10]
4. If Laspeyre's index number is 120 and Dorbish - Bowley's index number is 130.
Then Paasche's index number is:
(a) 140 (b) 120
(c) 130 (d) 125 [B.R.A. Agra 2010]
5. Index numbers measure :
(a) Relative changes (b) Absolute changes
(c) Both (a) and (b) (d) None of these [B.R.A. Agra 2009]
90
6. Laspeyre's Index number is :

Sp1 q0 Sp1 q1
(a) ´100 (b) ´100
Sp0 q0 Sp0 q1
Sp1 q1
(c) (d) None of these
Sp0 q0 [B.R.A. Agra 2008, 09]
7. In construction of an index number, the selection of base period is based on the

guidelines that:
(a) It should be a period of normal and stable economic conditions
(b) It should not be too distant from the given period
(c) (a) and (b) both
(d) None of these [B.R.A. Agra 2006, 08]
8. Fisher Ideal index number is the G.M. of:
(a) Marshall and Edgeworth price index number
(b) Walsh and Dorbish Bowley price index number
(c) Laspeyre and Paasches price index number
(d) None of these [B.R.A. Agra 2004]
9. Dorbish Bowley Price Index number is the A.M. of:
(a) Paasche's and Marshall Edgeworth index number
(b) Laspeyre and Marshall Edgeworth index number
(c) Laspeyre and Paasche index number
(d) None of the above [B.R.A. Agra 2005]
10. Fisher Ideal index number lies between:
(a) Laspeyre's and Paasche's index numbers
(b) Laspeyre's and Marshall Edgeworth index numbers
(c) Paasche's and Marshall Edgeworth index numbers
(d) None of these [B.R.A. Agra 2005]
11. Index number is a:
(a) Measure of relative change
(b) A percentage relative
(c) A special type of average
(d) All the above

91
12. The first and fore most step in the construction of Index number is :
(a) Choice of base period
(b) Choice of weights
(c) To delineate the purpose of index number
(d) All the above

13. Laspeyre's index number is also known as :
(a) Fixed base index (b) Given year method index
(c) Base year method index (d) None of these
14. Most preferred type of average for index number is :
(a) A.M. (b) G.M.
(c) H.M. (d) None of these

15. For comparing year to year changes in price level the suitable index to be used is:
(a) Fixed based index with average price as base
(b) Fixed based index
(c) Chain based index
(d) None of these

16. Symbolically Pon ´ Pno = 1 stands for :
(a) Unit test (b) Circular test
(c) Factor reversal test (d) Time reversal test

17. Symbolically P ´ Q Spn qn
on on = stands for:
Sp0 q0
(a) Unit list (b) Circular test
(c) Time reversal test (d) Factor reversal test

18. The weights used in Paasche's formula belongs to:
(a) A base period (b) Current period
(c) A given period (d) To any arbitrary chosen period

19. Weighted aggregative index formula using base year quantities as base is called:
(a) Laspeyres price index (b) Paasche's price index
(c) Bowley's price index (d) Fisher's price index

20. Weighted aggregative index formula using the average of base year and current
year's quantities as weights is called:
(a) Laspeyre's price index (b) Fisher's price index
(c) Marshall Edgeworth's index (d) Bowley's index

92
21. Weighted average of relatives if base year value is taken as weights gives:
(a) Fisher's index (b) Laspeyre's index
(c) Paasche's index (d) Bowley's index
22. Fixed base index numbers and chain indices are :
(a) The same (b) Different
(c) Always 100 (d) None of these

23. The time reversal test is satisfied if :
Spn qn
(a) Pon ´ Pno = 1 (b) Pon ´ Q on =
Sp0 q0
(c) Pon ´ Q on = 1 (d) Pon ´ Pon = 1

24. The suitable index number for the comparison of changes in price level every year is:
(a) Fixed base index number based on average prices
(b) Chain base index numbers
(c) Single year fixed base index numbers
(d) None of these

25. If Laspeyre's and Paasche's index numbers are 324 and 144 respectively, then
Fisher's ideal index number will be :
(a) 234 (b) 90
(c) 180 (d) 216

26. Index numbers are used for determining:
(a) Purchasing power (b) Deflation
(c) Splicing (d) None of these

27. Which of the following person is not related to the field of index number
(a) Prof. R.A. fisher (b) Prof. Irving fisher
(c) Marshall (d) Paasche

28. The price index number for 2010 to the base 2005 is 125 and for 2005 to the base
2010 is 80, The given indicies satisfy.
(a) Time reversal test (b) Factor reversal test
(c) Circular test (d) Unit test

29. A series of numerical figures which show the relative position is called
(a) A index number (b) Relative number
(c) Absolute number (d) None of these

93
30. Index number for the base period is always taken as:
(a) 50 (b) 100
(c) 0 (d) None of these
31. Price relative is equal to

Price in the given year Price in the base year
(a) ´100 (b) ´100
Price in the base year Price in the given year
(c) Price in the given year × 100 (d) Price in the base year × 100
32. Index number is equal to the :
(a) Sum of the price relatives (b) Average of the price relatives
(c) Product of price relatives (d) None of these
33. .............................. is an extension of time reversal test:
(a) Factor reversal test (b) Circular test
(c) Both (a) and (b) (d) None of these
34. The value at the base time period serves as the standard point of comparison:
(a) True (b) False
(c) Nothing can be said (d) Either (a) or (b)
35. The number of test of adequacy is :
(a) 2 (b) 3
(c) 4 (d) 5
36. We use price index numbers :
(a) To measure and compare prices (b) To measure prices
(c) To compare prices (d) None of these
37. Price relative is expressed in term of :

p p
(a) P = n (b) P = 0
p0 pn
p p
(c) P = n ´100 (d) P = 0 ´100
p0 pn
38. Cost of living index is expressed in terms of :

Spn q0 Spn qn
(a) ´100 (b)
Sp0 q0 Sp0 q0
Sp0 qn
(c) ´100 (d) None of these
Spn qn
94
39. Which of the following index satisfies the circular list ?
(a) Fisher's index (b) Paasche's index
(c) Laspeyer's index (d) None of these

40. In Paasche's price index formula:
(a) Current year quantities are taken as weights
(b) Base year quantities are taken as weights
(c) Base year prices are taken as weights
(d) Current year prices are taken as weights
41. Laspeyre's index is based on:
(a) Base year's quantities
(b) Current year's quantities
(c) Average of bases and current year's quantities
(d) None of the above
42. Fisher's formula for price index number is:
æ
(a) ç
å p1q0 ´ å p1q0 ö÷ ´100 æ
(b) ç
å p1q0 ´ å p0 q1 ö÷ ´100
ç
è å p0 q1 å p0 q1 ÷ø ç
è å p0 q0 å p0 q1 ÷ø
æ
(c) ç
å p1q0 ´ å p1q1 ö÷ ´100 æ
(d) ç
å p1q1 ´ å p0 q0 ö÷ ´100
ç
è å p0 q0 å p0 q1 ÷ø ç
è å p0 q0 å p1q0 ÷ø
43. Cost of living at two different cities can be compared with the help of:
(a) Consumer price index (b) Purchasing power of money
44. If the relation P01 ´ P10 = 1, it means that the price indices satisfy:
45. Factor reversal test was invented by:
(a) Walsh (b) A.L. Bowley
(c) Irving Fisher (d) J.I. Griffin
46. The first and foremost step in the construction of index number is :
(a) Choice of base period
(b) Choice of weights
(c) To delineate the purpose of index number
(d) All of the above

95
47. Fisher's ideal index number formula satisfies:
48. The index number which is most commonly used is:
(a) Diffusion index number (b) Price index number
(c) Value index number (d) None of the above
49. The base period for an index number should be:

(a) A year only (b) A normal period
(c) A period at distant past (d) None of the above
50. Fisher's ideal index number isth
(a) Arithemtic mean of Laspeyre's and Paasche's index numbers
(b) Geometric mean of Laspeyre's and Paasche's index numbers
(c) Both (a) and (b)
(d) Neither (a) and (b)
51. Laspeyre's index number is equal to:
1. p01 =
å p1q1 ´100 2. p01 =
å p1q0 ´100
å p0 q0 å p0 q0
3. p01 =
å p0 q0 ´100 4. p01 =
å p0 q1 ´100
å p1q1 å p0 q0
(a) 2 (b) 1
(c) 3 (d) 4
52. The formula to calculate an index number by 'family budget method' is:
(a)
å piq1 ´100 (b)
å p1 ´100
å p0 q1 å p0
(c)
å RW (d) None of these
åW
Fill in the Blank(s)
1. Marshall Edgeworth's formula does not satisfy ...................... reversal test.
[B.R.A. Agra 2000]
2. Fisher's ideal index number is the ...................... mean of ...................... and

...................... index numbers. [B.R.A. Agra 2007]
3. Index numbers are known as ...................... . [B.R.A. Agra 2007]

96
4. Index numbers are special type of ...................... .
5. Consumer price index is also known as ...................... .
6. If Pon × Pon = 1 it meas that ...................... test is satisfied.

å Pn qn
7. If Pon ´ Q on = , it means that ..................... .
å P0 q0
8. If a new series of index number is connected with an old one, it is called

....................... .
9. Fisher's ideal index number does not satisfy ...................... test.
10. The base period should preferably be ...................... .
True or False
1. Fisher Ideal index number does not satisfy circular test. [B.R.A. Agra 2012]
2. Fisher Ideal index number satisfy circular test [B.R.A. Agra 2011]
3. Index numbers are specialised type of averages
4. Index numbers are also known as economic barometers
5. Fisher's ideal index number satisfies time reversal test only
6. Time reversal property of relatives is P on × Q no = 1
7. Cost of living index numbers are based on retail prices of commodities.

Spn qn
8. P on × Q on = stands for time reversal test.
Sp0 q0
9. P on × P no = 1 stands for factor reversal test.
10. The weights used in Laspeyre's price index are denoted by q0 .
11. Weighted index numbers makes them less representative.
12. G.M. is the most appropriate average to be used for construction of an index
number.
13. Weighted average of relatives and weighted aggregative methods render the same
result.
14. "Like all statistical tools, index numbers must be used with great caution".
15. The value at the base period serves as the standard point of comparison.
16. Laspeyre's and Paache's method satisfy time reversal test.
17. There is no such thing as unweighted index numbers.

97
A NSWERS
14. P01 = 11127
. 15. (i) 118.96 (ii) 119.79
16. (i) 124.7 (ii) 121.77 (iii) 119.48 (iv) 119.29
(iii) 123.32 (iv) 123.33 18. 139.79
19. (i) 146.58 (ii) 145.36 (iii) 145.9 20. 49.13
21. 209.01

1. (b) 2. (c) 3. (c) 4. (a) 5. (a) 6. (a) 7. (c) 8. (c)
9. (c) 10. (a) 11. (d) 12. (d) 13. (b) 14. (b) 15. (c) 16. (d)
17. (d) 18. (c) 19. (a) 20. (c) 21. (b) 22. (a) 23. (a) 24. (b)
25. (d) 26. (b) 27. (a) 28. (a) 29. (a) 30 (b) 31. (a) 32. (b)
33. (b) 34. (a) 35. (c) 36. (a) 37. (c) 38. (a) 39. (d) 40. (a)
41. (a) 42. (c) 43. (a) 44. (a) 45. (c) 46. (d) 47. (c) 48. (b)
49. (b) 50. (b) 51. (a) 52. (a)
1. factor 2. G.M., Laspeyre's, Paache's 3. economic barometres
4. average 5. cost of living index 6. test is satisfied time reversal
7. factor reversal 8. backward splicing 9. circular
10. normal period

98
True or False
1. T 2. F 3. T 4. T 5. F
6. T 7. T 8. F 9. F 10. T
11. F 12. T 13. T 14. T 15. T
16. F 17. F
❍❍❍
UNIT
II
3. Statistical Methods for Psychology

& Education Statistics
101
U nit-2
3
S tatistical M ethods for P sychology
& E ducation S tatistics
3.1 Introduction
ome statistical methods have been developed to measure the psychological and
S educational characteristics.
A branch of psychology called psychometry deals with the measurement of psychological

trails or mental abilities like: intelligence, aptitude, interest, opinion, attitude or
scholastic (schools and education) achievements.
3.2 Scaling
Purpose of scaling: Unlike physical or biological characteristics, psychological
characteristics are rather abstract and hence can be measured only with abstract and
hence can be measured only with some degree of unreliability.
For the purpose of measurement of psychological and educational characteristics, one

has to develop a certain scale, which bears some similarity with a foot-rule used for
measuring or comparing lengths. As on a foot-rule, equal distances on a psychological
scale stands for empirically (relying on observation and experiment, not on theory) equal
differences in the psychological trait being measured. But the zero point of the
psychological scale, unlike that of the foot-rule, is arbitrary. However, distances from the
arbitrary zero are addictive. Here the most practical consideration is that the scales for
102
different tests should be comparable. In other words, a psychological scale and

‘interval-scale’ and not a ‘ratio-scale’ since there is no absolute zero point.
Most of the scaling procedures used in the psychological or educational data are based
upon the assumption of normal dist. For the characteristic or trail under consideration,
the zero point and the units of the scale are chosen arbitrarily, but the scale units should
be equal and remain stable throughout the scale.
Utility: ‘‘Its utility is that the scales for different tests are comparable. The scale units are
equal, meaningful, stable and can provide the comparability of the means, dispersions
and form of the dist.”
Besides this, the psychological and educational abilities like: I.Q., aptitude,
interest, opinion, attitude and scholastic achievements can also measured and compared
by the scaling.
3.3 Some Scaling Procedures

Scaling individual test items in term of difficulty (s- scaling): In this procedure, we
have a no. of items in a test administered to a large group of individual. The proportion of
individual success in each item is known and we are interested in arranging these items in
order of difficult, say from very simple to very difficult. We assume in the construction of
difficulty scale that the ability fig.3.1 which the group of items is measuring is normally
distributed with mean and S.D. we can arbitrarily take the origin at and write μ =0.
Let pi be the proportion of individuals

passing the ith item (i =1,2,.......,n). We
determine the point on x-axis for which
the area to the right of ordinate is pi. Let
Prob Density
the point be z i s thus z i s is the amount of

ability required for passing the item and P2
hence may be taken as a measure of
difficulty (di) for the ith item. Thus an
equal difference in d with mean are equal O d2
Ability
difference in ability required for passing
Fig. 3.1
the items.
1. Z score or s score and Z scaling

or s scaling): The mean expected in terms of S.D. are called s– scorce (x) in a test
has mean μ and standard deviation s
i.e., E ( X) = m
V ( X) = s (s scorce or Z-scorce)
103
x –m
Z= ...(1)
s
E ( x – m ) E( x) – E(m ) m – m
Z= = = =0
s s s
(x – m)
V (Z ) = V
s
1 1 s
= [V ( x – m )] = [var. x3 ] = 2 = 1
(s2 ) (s2 ) (s2 )
Hence the mean of a set of a scorce is always 0 and S.D. is unity.
2. Scaling of test-scores in several tests: In this method, the usual system of judging
the ability of an individual consists in adding the ‘raw scores’ of each individual in
different tests to get his ‘grand total’ or ‘composite –score’ and ranking them on the
basis of the grand totals. An individual with the highest total securing the first
position and so on.
This is not the valid procedure since the same raw score x on different tests may involve
different amounts of ability and hence may not be equivalent in different tests. Hence
the raw scores have to be scaled under some assumption regarding the distribution of the
trait which the test is measuring.
(i) Percentile scores or (percetile scaling): In this scaling, we assume that the
distribution of the trait under consideration in rectangular, under which we shall
have percentile differences equal throughout the scale. To determine the scale value
corresponding to a score x in a test, we have to find the percentile position of an
individual score x i.e., the percentage of individuals in the group having the score
equal to or less than x, which can be easily obtained from the score distribution
assuming that ‘score’ is a continuous variable. However, since most of the
psychological and educational data follow approximately the normal distribution,
the distribution of raw scores is rarely rectangular, so that the basic assumption
underlying the percentile scaling is wrong. Thus while using this scaling method one
should keep in mind it's limitations.
(ii) Z-scaling or s–scaling: Here, we assume that whatever differences there may be, in
the forms of the ‘raw- score’ distribution may be attributed to chance or to the
limitations of the test. As such, the distribution of the traits under consideration are
assumed to differ only in mean and S.D. Hence the scores in the different tests
should be expressed in terms of the scores in a hypothetical distribution of the same
form as the trait distribution with some arbitrarily Choosen mean and S.D. The
transformed scores are called “Z-scores”. To avoid negative standard-scores the
mean is generally taken to be 50 and S.D. to be 10. If a particular test has raw-score
mean and S.D. equal to μ and s respectively, then the Z-score corresponding to a
score x in that test is given by:
x – m Z – 50
=
s 10
104
x –m ö
or Z = 50 + 10 æç ÷ ...(2)
è s ø
(iii) T-scaling or T-score: In this method, we assume that the trait distribution of the
trait may be arbitrarily taken to be 50 and 10 respectively. To get the scaled score
corresponding to a raw-score x, first we find as in percentile-scaling the percentile
position (P) of an individual with score x and then find the point (T) on a normal
distribution with mean 50 and S.D. 10, below which the area is P/100.
æ T – 50 ö P
i.e., fç ÷= ...(3)
è 10 ø 100
where f (T) is the area under the curve of the normal variate from –¥ to T. The
scaled score obtained by this process is called ‘T-score’.
3.4 Linear Model of Test Theory

Suppose we are interested in getting the true measure of an individuals performance on a
test. Then by applying a measuring instrument, we will get the individuals raw score
(obtained score) or the test. Now, we consider a linear relationship, between the true
score of the 2 nd individual (t2 ) and his raw score (x2 ).
Let us consider, x i = ti + ei ; (i = 1, 2,.... n) ....(1)
where ( x i – ti = ei) is the error of measurement for the ith individual.
The obtained score (x) ¹ the unknown true score (t).
The difference score (x) ¹ error score (e) due to various factors.
In test theory, only the random errors e are considered. So we make the following
assumption for ei’s
m e = 0ü
ï
r te = 0 ý ...(2)
r eg eh = 0 ï
þ
where, m e = mean of error scores,
r eg eh = the correlation between error scores from different testing occasions (for two
parallel tests, g and h).
And r te = the error between true scores and error scores.
It is to be noted that under the given model, the estimates of m e , r te and r eg eh will tend
to zero if the no. of individuals (x) tends to infinity i.e. if x ® ¥.
105
Since, only random errors are considered, for a large no. of cases (for large x ) therefore
the +ve and –ve errors of all magnitudes (small and large) will cancel each other so that
the mean is zero.
1. There would be no correlation between true scores and error scores, (this is true for
both +ve and –ve scores) i.e. r te = 0.
2. Similarly the correlation r eg eh for two parallel tests g and h would be zero i.e.
r eg eh = 0.
Reliability of Test Score: A test is like measuring instrument one of the important
characteristics of any instrument is how reliable it measure. Reliability means
consistency in not technical language, if the instrument is reliable it should give
consistent result.
In modern test theory, every obtain score is thought of as being up to two parts. A
component which is called true score and a second component called error score. In
modern test theory,
Symmetrically Xt = X¥ + Xe
X t = obtained or Raw Score
X ¥ = true Score
X e = error Score.
3.5 Reliability
The term Reliability is defined as the reproducibility of the measurement when
re-measured under the identical conditions. The reliability of a test (a measuring
instrument) is given by the correlation between the raw scores of the given test and a
parallel test. Thus if g be the test and h any other test parallel to g, then the reliability of g
is measured by r xg xh and is denoted as r gg . Now for two parallel tests g and h,
we have r xg xh = s2tg / s2xg
or r gg = s2tg / s2xg
or r gg = 1 – s2tg / s2xg [Q s2t = s2x – s2e ] ...(1)
Thus reliability can also be defined as the ratio of true score variance defined as the ratio
of true score variance to raw score variance. Reliability lies between 0 and 1.
i.e. r gg = 1, when se = 0.
But se = 0, if all ei=0, since m e = 0
Thus a test is perfectly reliable (i.e., r gg = 1), if t2 = x2 for all i.

106
And r gg = 0, if st = 0 ( or se = s x ) and the test is said to be unreliable.
Hence for any test 0 £ r gg £ 1 .
3.6 Methods of Estimating Test Reliability

There are mainly four methods of estimating test reliability.
1. Parallel-test Method
2. Test-retest Method
3. Split-half Method
4. Kuder-Richard Method
1. Parallel-test Method: We know that the reliability is defined as the correlation

between raw scores and one of the two parallel tests. The reliability (r gg ) is
estimated by the correlation (r gg ) between the raw scores of the parallel tests
obtained from a sample and the two tests are administered to the same group with a
suitable time interval.
The ability measured should not change in the time interval between the
administrations of the tests. The parallel test reliability may also be obtained by
administering both the tests at the same time. In this case, the scores on the second
test may be influenced by familiarity with the material in the first test.
Thus parallel test method is the best method of estimating test reliability. But the
difficulty is to construct two parallel tests. So when only one test is available, we
cannot use this method.
2. Test-retest Method: In this method, the same test is administered twice after a
suitable time interval to eliminate familiarity with the test material etc. and then
the correlation between the test scores and retest scores is to find.
The difficulty with this method is that sometimes it is difficult to get the individual
again after an interval of time. In such a case, we cannot apply either the same test
twice or two parallel tests.
3. Split-half Method: In this method, one test is applied once and then the score is
divided into two equivalent halves, and the correlation between the scores on the
half-tests estimate the reliability of the half test. Then by the Spearman-Brown
formula, given by;
2r gg
r GG =
1 + r gg
where G is the composite test i.e G = ( g1 = g2 ). We may estimate the reliability of

the full test.
107
The test may be split into two parts in a no. of ways. The easiest way is to split on
the basis of odd and even numbered items.
4. Kuder-richard Son Method: Let us consider a test of length K which is made up of

K parallel items. Then the raw score variance is given by–
s2x = s ( x1 + x2 + ..........+ x k )
r xg xh = r gg
2
s2x – å s xg
and r gg = for all g and h since the items are all parallel.
(k – 1) å s2xg
Now, to obtain the reliability of the test of K parallel items form r gg we apply
Spearman-Brown Formula.
kr gg
r GG =
1 + (k – 1) r gg
This is called the Kuder- Richardson “ formula 20” for obtaining the reliability of a
test of K parallel items in terms of K, s x2 and s2xg .
If the mean of obtained scores on the test is m x , then

m x m x m 2x
s2x = –
k k2 k2
Then by formula (1), we get
ì k üì m x – m 2x / k ü
r GG = í ýí1 – ý
î k – 1þî s2x þ
This is called the KuderRicherdson “formula 21” for obtaining the reliability of a
test of K parallel items of equal difficulty in terms of K, s2x and m x .
The determination of reliability by Kuder-Richardson formulas is also known as the
method of “rational equivalence”.
3.7 Validity
A psychological test a measuring instrument should not only be ‘reliable’ but it should
also be 'valid’. By validity, we mean that the test should measure what it is supposed to
measure. For example, if we want to measure a trait ‘ A’ for a group of individuals with the
test, we must be sure, before using test that it is a suitable test for that purpose that is it
actually measures trait ‘A’ and also measure it reliably.
A test is valid for a particular trait for a particular group or for a particular situation. If we
use the same test for measuring different traits, then we must obtain its validity
separately for each case.
108
3.7.1 How to Estimate the Validity of A Test

To estimate the validity of a test we must know which trait we want to measure. This trait
is known as the “criterion variable’’ validity of a test is then estimated by computing the
“coefficient of validity” which determines the relationship between the scores obtained
on the test and the values of the criterion variable. It is difficult to get reliable measures
on the true criterion, because mostly we get only an approximate measure on the
criterion variable. Depending upon the situation, the criterion scores may be any one of
the following i.e. ratings by judges, scores on another valid test of the trait, measures on
later success etc.
3.7.2 Types of Validity

There are different types of validity.
1. Predictive Validity: These type of validity is used, when we use a test for selecting
applicants for a particular course or job and the criterion variable is the degree of
success at a later period i.e. after completion of the course or job. A test has a high
“predictive validity” if it can forecast efficiently later performance on a particular
measurable aspect of life and this is very useful in the selection of individuals for
different jobs and courses.
2. Concurrent Validity: Concurrent validity is obtained for tests for while the
criterion variable is also available at the same time as the results. Tests are
constructed for measuring a variable for which the result also may be obtained
without waiting. It is easier and save time and expenditure. It is used for diagnostic
tests.
3. Content Validity: In content validity of a test, we try to ascertain how for the test
covers the field of studying under investigation. Sometimes the tests constructed to
study the knowledge of the equals of certain specific areas of study, say verbal
ability, geometrical drawing ability etc. In this types of tests, we use the content
validity. However, it can not be expressed as a validity coefficient.
4. Construct Validity: Construct validity is useful when it is difficult to obtain
measurement on the criterion variable. This validity cannot be expressed in a single
measure as the correlation between test score and criterion scores. Some of the
common ways of establishing construct validity are following:
(i) Correlating different items or parts of the test. These correlations should be
high if the test is measuring a unitary variable.
(ii) Correlating different tests which measure same variable.
Construct validity is demonstrated by showing that the predictions expected on the
basis of the theory may be confirmed by the test.
3.8 Parallel Tests

Two tests are said to be ‘parallel’ when it makes no difference which one is used. If g and h
are two tests and if for the 2 nd individual tig ¹ tih , then we cannot say that it makes no
109
difference whether we use test g or h. So in order that g and h may be parallel tests, we
assume that,
1. tig = tih , [i = 1, 2,...., n] ....(1)

i.e. the true score of any individual should be the same on the two tests.
2. For the error scores on two parallel tests, we take
seg = seh ....(2)
i.e., the S.D. of errors on the two tests should be the same.
Thus the relations (1) and (2) define parallel tests in terms of unknown quantities.
These can be expressed in terms of the distribution of the raw scores using the relations as
follows:
From the linear relationship
x i = ti + ei and since m e = 0,
We get m t = m x , for any test.
From (1), we have m tg = m th , sth + stg = sth and stg sth = 1
Also, we have s2x = s2t + s2e for any test.
Therefore, we get
m x g = m x h and s x g = s x h ....(3)
For two parallel tests g and h.
i.e. the means and S.D.S. of raw scores on two parallel tests are equal.
If we have more than two parallel tests, then we will have to check one more condition to
say that the tests g, h and k are parallel.
cov( xg, xh) cov (tg, th) + cov (tg, eh) + cov(th, eg) + cov (eg, eh)
r xg xh = =
sxg sxh sxg sxh
ìQ g, h are parellel test &

cov(tg, th) ï
= íthe roaming cov terms are
sxg2 ï
îall zero and sxg = sxh
tg th stg sth
= r x g xh =
s2 xg
ìQ tg th =1 and
ï
s2 tg ï
ò
= í stg = sth Q gh
s2 xg ï
are parallel
ï
î
Thus for two parallel tests g and h,
s2 tg s2 th
rx g x h = = ....(4)
s2 xg s2 xh
110
Thus establishes the relation

i.e. rx g x h = rx g x k = r xh xk ....(5)
For a number of parallel tests.
Thus for three or more parallel tests the means, variances and the inter correlations of
raw scores are equal.
3.9 Point-Biserial and Biserial Correlation Coefficients

The point-biserial correlation coefficient is a statistic used to estimate the degree of
relationship between a naturally occurring dichotomous nominal scale and an interval
(or ratio) scale. For example, a researcher might want to investigate the degree of
relationship between gender being male or female .
Aside from the types of scales involved, the interpretation of the resulting coefficient is
very similar to that for the more commonly reported Pearson product-moment
correlation coefficient (sometimes referred to as Pearson r, or simply r). In brief, like the
Pearson r, the rpbi can range from 0 to +1.00 if the two scales are related positively (that
is, in the same direction) and from 0 to –1.00 & if the two scales are related negatively
(that is, in opposite directions). The higher the value of rpbi (positive or negative), the
stronger the relationship between the two variables.
In distinguishing the Point-Biserial from other correlation coefficients, We must first

point out that the point-biserial and biserial correlation coefficients are different. The
biserial correlation coefficient (or rbi) is appropriate when you are interested in the degree
of relationship between two interval (or ratio) scales but for some logical reason one of
the two is more sensibly interpreted as an artificially created dichotomous nominal scale.
To calculate the rpbi for each item use the following formula:
where;
rpbi = point-biserial correlation coefficient.
whole-test mean for students answering item correctly (i.e., those coded as 1's).
Mp =
whole-test mean for students answering item incorrectly (i.e., those coded as 0's).
Mq =
standard deviation for whole test.
St =
proportion of students answering correctly (i.e., those coded as 1's).
p=
proportion of students answering incorrectly (i.e., those coded as 0's).
q=
Point-Biserial Correlation Coefficient Used
As mentioned above, the point-biserial correlation coefficient can be used in any research
where you are interested in understanding the degree of relationship between a naturally
occurring nominal scale and an interval (or ratio) scale. For instance, We might be
interested in the degree of relationship between being male or female and language
aptitude as measured by scores on the Modern Language Aptitude Test (or MLAT;
111
Carroll & Sapon, 1958). The point-biserial correlation coefficient could help you explore
this or any other similar question.
A dichotomous variable is one that takes on one of only two possible values when
observed or measured. The value is most often a representation for a measured
variable (e.g., gender: male/female). If the dichotomous variable represents another
variable, that underlying variable may be either observed or unobserved. For example, a
dichotomous variable may be used to indicate whether a piece of legislation passed. The
dichotomous variable (pass/fail) is a representation of the actual, and observable, vote on
the legislation.
3.9.1 Tetra Choric r

If there are two variables one variable is continuous and is expressed as test scores and
other variable is in a two fold classification then biserial r & pt. biserial r provides an
adequate measure of relationship b / w product moment r & it becomes in area singly large
as the difference b / w p & q increase rbis gives an estimate of product moment r when
certain assumption have been met.
1. Continuity in the classified trait.
2. Normality of distribution underlying the classification.
3. A large N.
4. A split i.e. not too extreme, the closer to 0.5 is better.
3.9.2 Limitation
1. Biserial rbis can’t be used in a regression ea n.
2. It is not limited as in r to arrange of +1.
3. Rendering comparison with other coefficient of pt. Biserial- correlation difficulty.
When items are scared as if correct & zero. If incorrect i.e. right or wrong. The
assumption of normality in the distribution of right, wrong responses I unwanted then
point biserial r larger than biserial r is appropriate. Point biserial r assumes that the
variables which has been classified into two categories can be thought of as concentrated
at two distinct pt. along a graduated scale. Ex- of true classification are male-female,
living-dead, loyal-disloyal etc.
3.9.3 Tetra Choric r in Detail

If there are 2 variables one variable is continuous & is expressed as test scores & other
variable is in a two fold r provides an adequate measure of relationship between the two.
When both variables are in a two fold classification then these method are not applicable.
Then we have 2×2 or 4 table from which a variety of product moment corr. called tetra
choric r is useful when we wish to measure the association between two characters or
attributes neither of which is measurable in scores but both of which are being separated
into two categories. If we want to find the corr. b /w school attendance & current
employment persons might I st be classified into those who have graduated from high
school & those who have not & classified a II nd time into those who are presently
employed & those who are unemployed. Suppose we want to discover the relationship
112
b / w intelligence & social maturity & intelligenc, children might be classified as above or
below aug. In intelligence as social mature & socially immature. Tetra choric r assumes
that the two variables under study are essentially continuous & would be normally
distributed if it were not possible to obtain score or exact measure & thus be able to class
by both variables into frequency distribution.
Let the 2 characters be A & B.

Table 3.1: Total frequencies & cell frequencies of 2×2 table are:
B Class/A Class A1 B2 Marginal Totals
B1 a b a+b
B2 c d c+d
Marginal Total a+c b+d N
If the table is assumed to be derived through a double classification of a bi-variate normal

distribution with p.d.f.
1 é –1 ù
f ( x, y) = exp ê exp{ x2 – 2 pxy + y2 }ú
2
2 x 1 – p2 ë 2(1 – p ) û
Then f can be obtained in this measurement is the desired measure of association & is
called Tetra choric r. Thus the mode of measuring association assumes that the 2
characters under study are essentially continuous & would be normally distributed if it
were possible to measure them. Here, we assume the variables x & y corresponding to A
classification & B classification resp.
Let the value of x that forms the boundary b / w the classes A1 & A2 be h & be the value of
y that forms the boundary b / w B1. Since marginal distribution of x & y are univariate
normal with mean 0 & variance 1 i.e.
1 ì x2 ü
g( x ) = exp í ý i –¥< x <¥
2À î2 þ
1 y2
h( y) = e– i –¥ < y<¥
2À 2
¥ a+ c ¥ a+ c
Now
òk g ( x)dx =
n
&
òk h ( y)dy =
n
having by scaling these equation we get the values of we are then to solve for P the
¥ ¥ a
equation
ò ò
k n
f ( x, y) dx dy = .
n
This equation can be expressed in the form
å
¥ a
TiTi1 = – (*)
i=0 n
The function Ti are called Tetra choric function of h & Ti are Tetra choric function of k.
113
E XERCISE
1. Discuss different types of Scaling giving its purpose.
2. Explain Scaling, Reliability, Validity.
3. Define point-biserial and biserial correlation coefficients.
4. Define parallel test and state their properties.
5. Explain Tetra choric and biserial correlation coefficient.
6. Explain s – Score and T – Score.
7. What do you mean by Reliability of test ? Give various methods of finding

Reliability?
8. Define validity and explain the different type of concept of validity.
9. What is meant by scaling? What are raw score and Z – score?
10. Explain briefly concept of validity of scores in educational experiment.
11. What do you understand by test reliability? Define parallel test and discuss it.

1. What is ment by the length of the test.
2. Define reliability of test.
3. What is Rulon method of estimating reliability.
4. Give comparison between reliability and validity.
5. What is ment by the validity for a test infinite length.
6. Define test of validity.

1. Reliability ranges lies between
(a) 0 to 1 (b) -1 to 1
(c) - ¥ to 1 (d) - ¥ to ¥
2. Every obtain score is though of as being made up of two parts component called
(a) True Score & Raw Score (b) Error Score & Raw Score
(c) True Score & Error Score (d) Z – Score & Raw Score
114
3. Assumption of Error Component is distributed normally with

(a) mean 0 var 1 (b) mean 1 var 1
(c) mean 0 var ¥ (d) mean 0 var s square
4. Method of rational equivalence is also known as
(a) Lenth test (b) Kuder – Richardson Method
(c) Parrlel Method (d) Split Method
5. A test gave mean 60 & s.d. 12 complete the table for raw score
Raw score : 84,78,72 Z – Score : – , – , 1
(a) 3, 1, 1 (b) 2, 1, 2
(c) 1, 2, 2 (d) 2, 1.5, 1
Fill in the Blank

1. What are the important factor affecting reliability of test Error
Component are...................with the true value.

2. Reliblity of any set of measurments is define as the part of the................... .
3. Standard Drviation of Error Score is also known as................... .
4. Means Varience of the raw score on two Parallel test are................... .
5. A given test has a reliability coefficient of 0.8 S.D. of 20.What is max. Correlation
................... .
6. Reliability coefficient of a test 50 items is .60 . How much...................the test
length to raise the self Correlation to .90
A NSWERS
1. (a) 2. (c) 3. (a) 4. (b) 5. (d)
Fill in the Blank
1. uncorrelated 2. varience 3. standard Error
4. equal 5. .89 6. 6
❍❍❍
UNIT
III
4. Demographic Methods
117
U nit-III
4
D emographic M ethods
4.1 Introduction
emographic data or vital statistics are the numerical records of vital events. By vital
D events we mean the events of human life as birth, death, sickness, marriage,
migration divorce, adoption, separation etc. In short, all the events which have to do with
an individual's entrance into or departure from life together with the changes in civil
status which may occur to him during his lifetime. In other words, we shall be concerned
here by the different activities of the population. The word population here covers only
the population of human beings, of course, in certain very special cases, the population of
animals etc. can also be dealt with.
4.2 Uses of Vital Statistics

Vital statistics are extremely useful and their significance can be judged from the
following angles:
1. Use of Individual: Record of births, deaths, divorce and marriage are a paramount
use to the individual. The basic registration document or a certified copy thereof
has legal significance to the person concerned.
2. Use of Operating Agencies: Records of births, deaths and marriages are useful to
governmental agencies for a variety of administrative purposes. For example, the
control programmes for infectious diseases with in the family and within the
community, often depend on the death registration report for their initiation.
118
3. Study of Population Trend: Vital statistics reveal the changing pattern of the
population of a country expressed in terms of the number of births, deaths and
marriages etc. Moreover, these informations are indispensable for planning and
evalution of the schemes of health, family planning and other civil amenities. The
division of the population of different regions by birth and death rates enables us to
form some idea about the population trend of the region or countries and general
standard of living.
4. Use in Public Administration: The vital statistics is useful in planning and

evaluation of economic and social development programme. The statistics of infant
mortality gives us information about the prematurity of birth and birth defects.
5. Use in Actuarial Science and Life Insurance: It is the actuarial science and life
insurance scheme that involves the greatest and most useful application of vital
statistics as it concerns itself with all possible factors contributing to deaths in
various age groups. Further, life table guides in general calculation of risk at various
age groups.
6. Accuracy of Census Figures: Vital statistics can be used to checkup the accuracy
of the information provided by the census.
4.3 Collection of Vital Statistics

The vital statistics is collected by one of the three ways:
1. By Vital Registration: In many countries, there is a system of registering the

occurrence of every vital event under legal requirement. The registration of births
and deaths is maintained in almost all the towns, cities and villages by the
municipal boards, nagar nigam and gram panchyat. Whenever a child is born, the
matter has to be reproted to the proper authorities, together with the age of mother,
religion of parents etc. Similarly, every death occurring in the community gets
automatically recorded, because the disposal of the body requires death certificate
from the authorities. Although, the registration method is simple and effective it
suffers from the problem that all the births and deaths are not registered. This is
because the law has not been inforced strictly, particutary, in rural India.
2. By Demographic Surveys: Some times goverment conducts the surveys for getting
the information on vital events. For specially conducted population surveys, data
may be obtained from the hospital records. National Sample Survey organisations,
central statistical organisation etc. conduct this type of surveys.
3. By Census: In the census (complete enumeration) a survey is conducted by the

government at a certain interval of time say 10 years or 15 years. In India census is
conducted after every 10 years and particulars are collected regarding age, sex and
some social, economic, ethnic or familial characteristics of the individuals.
119
However, these information pertain to the census years only (once in ten years).
The data for the years other than census years are not available.
Here, we shall concern ourselves with birth death-the two most important vital events.
4.4 Rates
We always express the results of vital events in terms of rates. These rates are the relative
figures of the vital events as compared to the population in general. The general
definition of a rate is given below:
Rate of a vital event
Number of cases of the vital event

=
Total number of person exposed to the risk of occurrence of the event
According to the above definition, a rate is a proper fraction. For the sake of
understanding, the fraction is generally multiplied by a constant, which for most rates is
1000. Vital statistics rates are thus generally expressed "per thousand of population".
4.5 Fertility Measurements

In demography, the word fertility is used in relation to the actual production of children
or occurrence of births (live births). Fertility measures the rate at which a population
adds to itself by births and is normally assessed by relating the numbre of births to the
size of some section of the population, such as the number of married couples or the
number of women of child bearing age, an appropriate yardstick of potential fertility.
4.6 Factors Affecting Fertility Variation

The number of children produced by a group of women in a given year will depend upon
their ages, whether they are married and how many children they have already born. It
will also depend upon the economic condition, housing conditions and the educational
facilities available. It may also depend upon where they live e.g. whether in a urban or a
rural environment.
4.7 Crude Birth Rate

This is the simplest type of measure to give an idea about the fertility. It is usually
calculated by relating total live births in a year to the total population of all ages and
expressing it as a rate per 1000, i.e.,
120
B
C.B.R. = ´1000
P
B = Number of live births of a given region of a given period
P = Total population of the given region of a given period.
Here, the total population is not the proper population at risk so far as births are
concerned since it contains males and also females out side the child bearing span (an age
interval of 15 to 49 year is assumed and it is supposed that females are capable of bearing
a child. This age interval i.e., 19 to 49 is called child bearing span). Again, the risk varies
in females from one group to another, a woman of 25 is certainly under a greater risk than
a woman of 40.
4.8 General Fertility Rate

The crude birth rate is not a detailed measure of fertility, as the fertility differs from one
age group to another. Moreover, the C.B.R. is normally affected by sex ratio, marriage
ratio, migration and family planning. etc.
By relating the, total number of live births to the number of females in the child bearing
ages, the general fertility rate is obtained, Thus, G.F.R. is given by
B
G.F.R. = w ´1000
2
å fpx
w1
B = Number of live births in the given region during the given period.
fpx = Number of females of age x (last birthday) in the given region during the given period.
w1w2 = Lower and upper limits of the female reproductive period as 15 – 49.
The G.F.R. shows how much the woman of child bearing span have added to the existing
population through births. It takes into account the sex composition of the population
and also the age composition to some extent. Yet, it is calculated without proper regard
to the age composition of the female population in child bearing ages. As such, two
populations may show quite different G.F.R. although, they may have the same fertility
in each one year age-group.
4.9 Age-Specific Fertility Rate

In order to overcome the draw-back of G.F.R. and get a better idea of the fertility
situation prevailing in a community, it is necessary to compute the fertility rates for
different age groups of child bearing age separately. The fertility rate computed on the
121
basis of specification with respect to age is called age specific fertility rate and for age x to
x + n is
B
A.S.F.R. = n ix = n x
fn Px
n B x = Number of live births to women of age x to x + n in a given region during a given period.
fn Px = Number of woman of age x to x + n in the given region of the given period.
In the case of annual age-specific fertility i.e., n =1, A.S.F.R. (n ix ) reduces to
B
A.S.F.R. (Annual) = ix = x ´1000
fpx
4.10 Total Fertility Rate

The age-specific fertility rate cannot be used to compare the fertility rates of two
communities at the same period (or of the same community at different period). To be
practically useful, age specific fertility rates have, therefore, to be combined in a single
quantity. For this purpose, a standardised fertility rate may be employed. A much
simpler method is to add up the annual age-specific rates and taken the sum, called the
total fertility rate, thus
w2
T.F.R. = å ix
w1
When only quinquenennial, instead of annual fality rates are available, an approximate
value of the T.F.R. is given by
T.F.R. = 5 å 5 ix
4.11 Mortality Measurement

The purpose of measuring mortality is to draw inferences aout the likelihood of deaths
within a specific population during a specified period of time. The following are the main
rates for measuring mortality.
4.11.1 Crude Death Rate

The simplest type of rate used in the measurement of mortality is crude death rate
(C.D.R.). The crude death rate is the number of deaths due to all causes within a year per
thousand of the population. It is given by
Number of deaths (due to all causes) of the given region during the given period
C.D.R. = ´1000
Total population of the given region during the given period
122
D
= ´1000
P
It is most widely used rate due to the fact that it is relatively easy to compute. However, it
has some serious draw backs. In using the C.D.R. we ignore the fact that the chance of
dying is not the same for the young and the old males and females. Further, it may also
vary with respect to race, occupation or locality of dwelling. Because of this, C.D.R. is not
suitable as an index of relative mortality in different places unless the population of the
places compared have substantially identical in age and sex composition. Thus, a
population with higher number of old people will show a higher C.D.R.
4.11.2 Specific Death Rate

The C.D.R. normally varies with age, specifically in the group by (0-5) and (60 - above),
this rate is quite high and perhaps it is lowest in the age group of (25 - 40). Moreover, the
death rates for males and females are not same. Thus, we define specific death rates.
The S.D.R. is computed particularly for a specified section of the people. Usually, the
specific death rate is calculated for specific age or sex. Age specific death rate is defined as:
D
A.S.D.R. = nmx = n x ´1000
n Px
where n D x = Number of deaths between ages x and x + n -1 among the residents in a
community during a period.
n Px = Number of persons in the same age group in that community during that period.
The annual age specific death rate (for n = 1) is given by

Dx
Annual A.S.D.R. = mx = ´1000
Px
Again let m m
n Px and n D x denote the number of males aged x to x + n -1 and the number of
deaths occurring to such males, then age specific death rate for males is given by
m
D
A.S.D.R. for males = nm x ´1000
n Px
This is a death specific for both age and sex. Similarly, age specific death rate for female is
calculated by the following formula:
f
D
A.S.D.R. for females = n x ´1000
f
n Px
For general purposes, death rate specific for age and sex is one of the most widely used
types of death rate. It also supplies one of the essential components required for
constructing life table (mortality table) and net reproduction rate, to be discussed later on.
123
4.11.3 Infant Mortality Rate

In demography, the infants are defined as the population belonging to the age group
(0-1) years. It is a matter of common practice that the mortality behaviour under one
year of age is erratic and needs a separate consideration. Keeping in view of this aspect,
the infant mortality rate is separately defined as
Total number of deaths of children under one year of age during a period
I.M.R. = ´1000
Total number of live births during the same period
4.11.4 Standardised Death Rate

Suppose we want to compare the death rate of two populations say A and B. Now, if we
compare their C.D.R., we cannot justify our judgement because C.D.R. takes into
account all the sections and age group of the population with equal weight age. On the
other hand, if we compare A.S.D.R. of two populations A and B, then we will have to
compare the death rate in different age groups. Also one series may have higher S.D.R.
than the other for some of the segments but lower S.D.R's for others segments. In such
cases, we will not be able to give a general statement of the form "mortality is higher (or
lower) in A than B". For this purpose we use standard death rate.
Specific death rates are standardised in a number of ways. We shall consider here only
two methods:
1. Direct Method 2. Indirect Method
4.11.4.1 Direct Standardised Death Rate

To overcome the difficulty mentioned above, we introduce a 3 rd population known as
standard population and on the basis of this population we calculate the S.D.R. (Specific
Death Rate) of the two population A and B. We denote the standard population by S.
The selection of standard population is not a difficult job, for example, comparing Assam
and West Bengal in respect of mortality, we may take the population of whole of India or
the Eastern India as standard. Thus if PxS is the number of persons in the same age group
x in the standard population, then the standardised death rates for the region A and B are
given respectively by:
(St.D.R.)A =
å mxA PxS ´1000
å PxS
D xA
where, mxA =
PxA
and (St.D.R.)B =
å mBx PxS ´1000
å PxS
DB
where, mB
x =
x
PxB
124
Solved Examples
Example 1: For the following data find out which of the city is more healthy.
Age Group Standard Population Death Rate
City A City B
0-5 5000 25 30
5 - 15 8000 15 20
15 - 55 10000 10 15
55 and above 6000 60 70
[C.C.S., Meerut 2004]
Solution: Table showing calculation
Age Group Death Rate Standard m xA PxS m xB PxS

City A, m xA City B, m xB Population PxS
0-5 25 30 5000 125000 150000

5 - 15 15 20 8000 120000 160000
15 - 55 10 15 10000 100000 150000
55 & above 60 70 6000 360000 420000
Total 29000 705000 880000
å mxA PxS = 705000 = 24.31

(St.D.R.) A =
å PxS 29000
(St.D.R.)B =
å mBx PxS = 880000 = 30.35
å PxS 29000
Since the standard death rate for city A is less as compared to the city B, we infer that city
A is more healthy.
Example 2: Compare the mortality conditions in the two countries.
Age Group Death rate per 1000 Standard population
Country A Country B (in, 000)
0-4 4.348 18.870 119.9

5-14 0.465 0.759 206.9
15-24 0.767 1.385 183.2
25-34 1.075 2.048 147.9
125
35-44 1.882 3.326 120.5

45-54 4.669 7.006 93.9
55-64 12.477 18.111 70.8
65-74 34.060 45.795 40.5
75 and more 116.433 124.258 16.4
Solution: Table for calculation
Age Group Death rate per 1000 Standard m xA PxS m xB PxS

Country A Country B Population
m xA m xB PxS
0-4 4.384 18.870 119.9 525.64 2262.51

5-14 0.465 0.759 206.9 96.21 157.04
15-24 0.767 1.385 183.2 140.51 253.73
25-34 1.075 2.048 147.9 158.99 302.90
35-44 1.882 3.326 120.5 226.78 400.78
45-54 4.669 7.006 93.9 438.42 657.86
55-64 12.477 18.111 70.8 883.37 1282.26
65-74 34.060 45.795 40.5 1379.43 1854.70
75 and more 116.433 124.258 16.4 1909.50 2037.83
1000 5758.85 9209.61
SmxA PxS 575885

.
(St.D.R.) A = = = 5.7589
SPxS 1000
S mB S
x Px 9209.61
(St.D.R.)B = = = 9. 2096
SPxS 1000
Mortality conditions in country A are better to that of country B, i.e., country A is more
healthy.
Example 3: Compute standard death rate for two countries from the following data:
Age-Group Country A Country B Standard

Population
(years) Population Deaths Population Deaths
PxS
PxA D xA PxB D xB
0-15 10000 250 15000 450 12000
15-30 20000 450 12000 480 15000

126
30-50 15000 300 30000 600 25000
50 & above 30000 150 20000 250 35000
Solution: First of all we calculate death rate for each age group of country A and country
B by the formula:
D xA DB
mxA = ´1000 and mB
x =
x ´1000
PxA PxB
Table for calculation of mxA and mxB
Age-Group Country A D xA Country B D xB

´ 1000 ´ 1000
(years) PxA PxB
Population Deaths Population Deaths
PxA D xA = m xA PxB D xB = m xB
0-15 10000 250 25.0 15000 450 30.0
15-30 20000 450 22.5 12000 480 40.0
30-50 15000 300 20.0 30000 600 20.0
50 & above 30000 150 5.0 20000 250 12.5
Table for Calculation of S.D.R.
Age Group Death Rates Standard m xB PxS

m xA PxS
Population
mxA m xB
PxS
0-15 25.0 30.0 12000 300000 360000
15-30 22.5 40.0 15000 337500 600000
30-50 20.0 20.0 25000 500000 500000
50 & above 5.0 12.5 35000 175000 437000
87000 1312500 1897000
St. D.R. of Country A =

å mxA PxS = 1312500 =15.09
å PxS 87000
St. D.R. of Country B =
å mBx PxS = 1897000 = 2181
.
å PxS 87000
127
4.11.4.2 Indirect Standardised Death Rate

For computing the standardised death rates by direct method it is necessary to know the
number of persons and age specific death rates for different age groups. In some cases, we
may have population classified by age but the S.D.R s , for each age-groups may not be
known. Only the total number of deaths are known and hence the C.D.R. may be
obtained.
In such a case, the S.D.R. in calculated in the following steps.

Step 1: We find the C.D.R. of population A, B and population C as:
å mxA PxA
C.D.R. of population A = C A = x
å PxA
å mBx PxB
C.D.R. of population B = CB = x
å PxB
å mCx PxC
C.D.R. of population C = CC = x
å PxC
Step 2: We find the index death rate of population A by applying A.S.D.R. of
population C on population A and summing it over all the age groups and
dividing it by the total number of population A. Thus,
å mCx PxA
Index death rate for population A = I A = x
å PxA
å mCx PxB
Index death rate for population B = I B = x
å PxB
Step 3: Now, we find adjustment factor by dividing the C.D.R. of standard population
by the index death rate of population A and B. Thus,
Adjustment Factor for A = Cc / I A
Adjustment Factor for B = Cc / I B
Step 4: St. D.R. for A and B are obtained by multiplying adjustment factor by C.D.R. of
A and B separately. Thus,
St. D.R. for population A = Adjustment factor for A ´ C A

128
St. D.R. for population B = Adjustment factor for B ´ CB .
The following formulae are obtained.
St. D.R. of population A =

å mxA PxA . å mCx PxC
å mCx PxA . å PxC
St.D.R. of population B =
å mBx PxB . å mCx PxC
å mCx PxB . å PxC
Example 4: Find the standardised death rate by direct and indirect method for the
following data:
Standard Population Population A

Age
Group Population Specific Death Population Specific Death Rate

Rate
0-5 8000 50 12000 48
5-15 10000 15 13000 14
15-50 27000 10 15000 9
50 and above 5000 60 10000 59
Solution: Computation of STDR by Direct and Indirect methods.
Age Standard Population Population A m xA PxS mS

x Px
A
PxS mS
x PxS mS
x PxA m xA PxA m xA
0-5 8000 50 400000 12000 48 576000 384000 600000
5-15 10000 15 150000 13000 14 182000 140000 195000
15-50 27000 10 270000 15000 9 135000 243000 150000
50 & above 5000 60 300000 10000 59 590000 295000 600000
Total 50000 1120000 50000 1483000 1062000 1545000
SmxA PxS 1062000

Direct Method: (STDR) A = = = 21.24
SPxS 50000
mxA PxS 1483000

Indirect Method: (CDR) A = = = 29. 66
PxA 50000
129
^ S mS S
x Px PxA 1120000 50000
Adjustment factor: C = ´ = ´
SPxS mS
x Px
A 50000 1545000
= 0.7249
\ (STDR) A = C$ ´ (CDR) A
= 0.7249 × 29.66 = 21.5005
4.12 Measurement of Population Growth

When measures of mortality and fertility are obtained, a question that naturally arises is
whether the tendency of the given population , as indicated by these measures, is to
increases, to decrease or to remain stable. It is, therefore, necessary to devise measures of
population growth on the assumption that current mortality and fertility will also
continue to prevail in future. In the following sections, we shall study some measures of
population growth under the assumption that in future also it is subject to the current
fertility and mortality rates.
4.12.1 Crude Rate of Natural Increase and Pearle's Vital Index

The simplest measure of population growth is the crude rate of natural increase, which is
obtained by subtracting the C.D.R. from the C.B.R. The C.B.R. give the proportion by
which the population increases through births, while the C.D.R. represents the
proportion by which it decreases through deaths. The difference of the two, therefore,
shows the net gain (or loss) in the population size through births and deaths taken
together.
An alternative measure of the same type is the ratio of the total number of births to the
total number of deaths (sometimes multiplied by 100). This is called Pearle's Vital
Index. Actually, it is the ratio of the C.B.R. to the C.D.R.
4.12.2 Demerit
Since C.B.R. and C.D.R. are both crude measure of birth and death and therefore,
considered unsuitable as indices of population growth.
4.12.3 Gross Reproduction Rate

In order to have a better idea about the rate of population growth, it is first of all
necessary to take into account the age-sex composition of the population.
Since our aim is to measure population growth, it is appropriate to consider female births
alone, because it is mainly through females that a population increases. The gross
reproduction rate (G.R.R.) is defined as the sum of age-specific fertility rates calculated
130
from female births for each year of reproductive period (child bearing span 15-49).
Symbolically, age specific fertility rate is given by
f
Bx
fi = ´100
x f
Px
where f B x is the number of female births to women of age x during the given period in
the given community.
Summing the above rates for all ages in the reproductive period (15-49 say w1 to w2 ), the
gross reproduction rate (G.R.R.) is obtained, Thus,
w2
G.R.R = å fi x
w1
On the other hand if the fertility rates are for quinquenial age groups, then the gross
reproduction rate is given by
w2 f5
å
Bx
G.R.R. = 5 ´1000
f5
w1 Px
4.12.4 Net Reproduction Rate

The drawback of the G.R.R. is that it does not take into account the fact that some of the
females who are assumed to begin life together may die before reaching age 15, some may
die between ages 15 and 16 and so on. In other words, the G.R.R. takes into account
current fertility not current mortality.
To take into account the factor of mortality in measuring population growth, we

construct a life table for females on the basis of the observed age-specific death rates for
females, fi . The values in Lx column of the table ( f Lx in this case) give the mean size of
x
the cohort of f l0 females in the age interval x to x + 1 for varying x.
Hence, fi f Lx
x
gives the number of female children that would be born to the cohort at age x1. b. d. The
sum of these values i.e.,
w2
å fi f Lx
x
w1
131
is the total number of female children that are expected to be born to f l0 females during
their life time, our new measure of population growth is
w2 f w2
å å fi
1 f Lx
fi Lx =
f x f x
l0 w1
l0 w1
and is called the net reproduction rate.
In common parlance,
N.R.R. =
å (No. of female births ´ % survival rate)
100
f f
Bx Lx
where, fi = ´1000 and
x f f
Px l0
is called survival factor or proportion of female survivors to that age or age group.
The experience gives some idea about the values of N.R.R. and their importance as
follows:
1. Normally N.R.R. varies from 0 to 5
2. N.R.R. can not exceed G.R.R. as N.R.R. takes into account the mortality.
3. N.R.R. = G.R.R. if all newly borned female children survive till their maximum
child bearing span.
4. If N.R.R. = 1, the female population will exactly replace itself into new generation
and population remains constant.
5. If N.R.R. < 1, this will result into the reduction in the number of mothers and will
thus cause reduction in population.
6. If N.R.R. > 1, there will be a greater number of mothers in the next generation
which will tend to increase the population.
Example 5: Compute (i) G.F.R., (ii) S.F.R., (iii) T.F.R. and (iv) the gross reproduction
rate from the data given below:
Age group of child: 15-19 20-24 25-29 30-34 35-39 40-44 45-49
bearing females
Number of women: 16.0 16.4 15.8 15.2 14.8 15.0 14.5
(in thousands)
Total births: 260 2244 1894 1320 916 280 145

132
Assume that the proportion of female births is 46.2 percent.
Solution: Table for calculation.
Age Number of women Number of births Age-specific fertility rate

f n Bx n Bx ´1000
(years) ('000') n Px n ix = f
n Px
15-19 16.0 260 16
20-24 16.4 2244 137
25-29 15.8 1894 120
30-34 15.2 1320 87
35-39 14.8 916 62
40-44 15.0 280 19
45-49 14.5 145 10
Total 107.7 7059 451
G.F.R. =
Total number of live births
´1000 =
å n B x ´1000
w2
å fnPx
å f
n Px
w1
7059
= ´1000 = 65.54 per thousand
107700
w2
T.F.R. = 5 × n ix = 5 ´ S 451 = 2255 per thousand
w1
In this case, we cannot compute G.R.R. by the formula used in G.R.R. since we are not
given the groupwise break-up of female births. However, since it is given that the
proportion of female births is 46.2 percent, we can obtain the value of G.R.R. by
approximate formula i.e.,
Number of female births

G.R.R. = ´T . F. R .
Total births
46. 2
= ´ 2255 = 1041.81 per thousand.
100
133
4.13 Stationary and Stable Populations

A population is said to be stationary if:
1. The number of births and number of deaths are exactly equal and distributed
uniformly throughout the year.
2. The population is closed against migration.
3. The age and sex composition is fixed.
Thus, population with constant size (because number of births and deaths are equal) and
constant age and / sex composition over time is called stationary.
On the other hand a population is said to be stable if:
1. The age and sex distribution is fixed
2. If constant mortality and fertility rates are experienced at each age and
3. The population is closed against migration. This is to be noted that in a stable

population mortality and fertility rates are constant but they need not be equal.
4.13.1 Difference between Stationary and Stable Population

If the mortality and fertility rates happen to be exactly equal, than the stable population
is said to be stationary one because the population size remain fixed.
4.14 Life Table or Mortality Table

The life table gives the life history of a hypothetical group (called cohort) as it is gradually
diminished by deaths. The record begins at the birth of each member and continues until
all have died. It gives the age distribution of mortality in tabular form and is a powerful
tool for measuring the probability of life and death of various age sectors. It contains the
answers of the questions like: the expectation of life at some particular point of time, the
expected number of persons attaining a particular age, average longevity of life etc.
Life tables are generally constructed for various sections of the people which generally
have different patterns of mortality. Thus, there are life tables constructed for different
races, occupational groups and sex. It is to be noted that life tables constructed for
different races, occupational groups and sex. It is to be noted that life table is based on a
stationary population. There are two type of life tables:
(i) Complete life table and (ii) Abridged life table.
A life table in which the age interval is a year throughout the table and the life table
functions such as l x , d x , px , q x , Lx , etc. are given for all integral value of x, is called a
134
complete life table. On the other hand in abridged life table the values of these functions
are given either for some integral values of x which may be usually 5 years or 10 years. It is
to be noted that abridged life tables are obtained through the condensation of the
complete life table and not by omitting some of its facts.
4.15 Assumptions of a Life Table

The following assumptions are taken at the time of construction of a life table:
1. The cohort generally consists of the members of one sex only. Because mortality
pattern is different in males and females.
2. The cohort is closed for migration. That is, there is no change is allowed except the
losses due to deaths.
3. The deaths are evenly distributed throughout the period ( x, x + 1) for each x.
4. The cohort originates from some standard number of births say 1000, 100000 or
100000. This is called the radix of the table.
5. Individuals die at each age according to a pre-determined schedules which is fixed

and does not change.
6. The cohort diminishes gradually only by deaths.
4.15.1 Different Elements of a Life-Table

A typical life table has the following elements.
(1) (2) (3) (4) (5) (6) (7) (8)
x lx dx qx px Lx Tx e0x
sometimes three more columns are used in a life table; these are denoted by mx ,m x and e x .
Let us explain these elements one by one.
x : x =1, 2, 3,... i. e. the symbol x represents the integral values of exact age in years.
lx : l x is the number of persons living at any specified age x in any year out of an
assumed number of births called the cohort or radix of the life table and is denoted
by l0
d x : d x is the number of persons among the l x persons (attaining a precise age x) who die
before reaching the age ( x + 1). Thus, d x = l x - l x + 1 is the number of persons dying
between exact age x and x + 1.
q x : q x is the probability that a person of exact age x will die before reaching age x + 1. It
follows that
135
d
qx = x
lx
p x : px is the probability that a person of precise age x will survive till his next birth day.
L x : Lx is the number of years lived, in the aggregate, by the cohort of l0 persons between
ages x and x + 1.
1
Thus,
ò
Lx = l x + t dt
0
an approximate value of which is given by
lx + lx + 1 1
= lx - d
2 2 x
T x : Tx is the total number of years lived by the cohort after attaining an exact age x. i.e.,
Tx is the total future life time of the l x persons who reach age x. Thus,
Tx = Lx + Lx + 1 + Lx + 2 + ...
e0x : It is the complete expectation of life at age x. It is given e0x = Tx / l x
Thus e0x is the average number of years lived after age x by each of the l x persons who
attain that age. e0 , the expectation of life at age 0 is the average age at death.
4.15.2 Additional Elements of a Life Table

The other elements life mx ,m x and e x are obtained by the relations
2 qx
mx = , the central death rate
2 - qx
1
m x = mx - , force of mortality
2
1
e x = e0x - = the average number of years lived by the persons of the given age x.
2
4.16 Construction of a Life Table

From the discussion of different elements of a life table it is clear that a complete life table
constructed if q x is computed for every x from 0 to limiting age and the radix of the life
table, that is l0 , is known, which is the number of persons exposed to the risk of death.
Starting with the radix l0 and q x , (x = 0,1, 2,...) we have
d0 = l0 q0 Þ l1 = l0 - d0
136
d1 = l1q1 Þ l2 = l1 - d1
and so on. from these values of l x ( x = 0,1, 2 ...) the columns Lx , Tx and e0x can be
completed as follows:
lx + lx + 1
å Li,
T
Lx = , Tx = e0x = x
2 lx
i= x
A model life table is shown below :
Life Table
x lx dx qx Lx Tx e0x
0 100000 15322 0.15322 88509 4188830 41.89
1 84678 2552 0.03014 82404 4100321 48.42
2 82126 1950 0.02374 80404 4017917 48.92
3 80176 1473 0.01837 78086 3937513 49.11
... ... ... ... ... ... ...
... ... ... ... ... ... ...
... ... ... ... ... ... ...
96 27 17 0.62330 19 28 1.04
97 10 07 0.70399 7 9 0.90
98 3 2 0.79364 2 2 0.67
99 1 1 0.89276 1 — —
4.17 Uses of a Life Table

Although the basic objective of a life-table is to present a clear picture of the mortality
experienced by a given population group it may be put to other important uses.
1. The most important use of life table is for actuarial work. Life tables form the basis
for determining the rates of premium necessary for various amount of life
assurances.
2. Life table is used for the measurement of growth in the computation of net
reproduction rate (Lx element of life table) and in population projection.
3. Life table is used for analysing the effects of mortality on the age and sex
composition of a population.
137
4. Life table for two or more different group of population may be used for the relative
comparison of various measures of mortality such as death rate, expectation of life
at various ages etc.
5. Life tables may be used by private/public sector organisations for determining the
ratio of retirement benefits for its employees.
6. It is used for the appraisal of the accuracy of causes enumeration and vital
registration data.
7. Life table is used for the calculation of probability of surviving and dying.
8. Life tables have also been used in preparation of population projections by age and
sex. That is, in estimating what the size of the population will be at some future
date.
Example 6: Complete the following Life-Table:
x 11 12 13 14
lx 12000 8000 6000 4000

and L14 = 3000
Solution: Here years are 11, 12, 13 and 14.
(i) Using the relation d x = l x - l x + 1, we have
d11 = l11 - l12 = 12000 - 8000 = 4000
d12 = l12 - l13 = 8000 - 6000 = 2000
d13 = l13 - l14 = 6000 - 4000 = 2000
where suffix denotes the year.

d
(ii) Using the relation q x = x = 1 - px , we have
lx
d 4000
q11 = 11 = = 0.333 \ p11 = 1 - q11 = 0.6667
l11 12000
d 2000
q12 = 12 = = 0.2500 \ p12 = 1 - q12 = 0.7500
l12 8000
d 2000
q13 = 13 = = 0.3333 \ p13 = 1 - q13 = 0.6667
l13 6000
138
lx + lx + 1
(iii) Using the relation Lx = , we have
2
l +l 12000 + 8000
L11 = 11 12 = = 10000
2 2
l +l 8000 + 6000
L12 = 12 13 = = 7000
2 2
l +l 6000 + 4000
L13 = 13 14 = = 5000
2 2
(iv) Using the relation Tx = Lx + Lx + 1 + ..., we have
T11 = L11 + L12 + L13 + L14 = 10000 + 7000 + 5000 + 3000
= 25000 (as L14 = 3000 given)
T12 = L12 + L13 + L14 = 7000 + 5000 + 3000 = 15000
T13 = L13 + L14 = 5000 + 3000 = 8000
T14 = 3000 (already given)

T
(v) Using relation, e0x = x we have
lx
0 T 25000
e11 = 11 = = 2.0833
l11 12000
0 T 15000
e12 = 12 = = 1.8750
l12 8000
0 T 8000
de13 = 13 = = 1.3333
l13 6000
0 T 3000
e14 = 14 = = 0.7500
l14 4000
Finally, the following table is formed.
Age in lx dx px qx Lx Tx e0
x
years X
11 12000 4000 0.3333 0.6667 10000 25000 2.0833
12 8000 2000 0.2500 0.7500 7000 15000 1.8750
13 6000 2000 0.3333 0.6667 5000 8000 1.3333
14 4000 — — — 3000 3000 0.7500

139
Example 7: Complete the following life table marked with question marks.
Age x lx dx px qx Lx Tx e0
x
30 762270 ? ? ? ? 27296732 ?
31 758580 — — — — ? ?
(C.C.S. Meerut 2000 (Back))
Solution: Taking age 30 as x and age 31 as x + 1, in the usual notations, we get
d30 = l30 - l31 = 762270 - 758580 = 3690

d 3690
q30 = 30 = = 0.00484
l30 762270
p30 = 1 - q30 = 1 - 0.00484 = 0.99516

l +l 762270 + 758580
L30 = 30 31 = = 760425
2 2
T31 = T30 - L30 = 27296732 - 760425 = 26536307
0 T 27296732
e30 = 30 = = 35.8098
l30 762270
0 T 26536307
e31 = 31 = = 34.98
l31 758580
Finally, the following table is formed:
Age x lx dx px qx Lx Tx e0
x
30 762270 3690 0.99516 0.00484 760425 27296732 35.8098
31 758580 — — — — 26536307 34.9800
Example 8: Given the following table for lx , the number of rabbits living at age x, complete
the life table for rabbits.
x: 0 1 2 3 4 5 6
lx : 100 90 80 75 60 30 0
Solution: The complete life table for rabbits is given below:
Age x lx d x = lx - lx + 1 q = d x lx + lx + 1 T = L + L T
x Lx =
x x x + 1 ... e0x = x
lx 2 lx
0 100 10 0.10 95 385 3.85
1 90 10 0.11 85 290 3.22

140
2 80 5 0.06 77.5 205 2.56
3 75 15 0.20 67.5 127.5 1.70
4 60 30 0.50 45 60 1.0
5 30 30 1.00 15 15 0.5
6 0 — — — — —
4.18 Estimates of the Various Functions of the Life Table

and their Inter-Relationship
We have already discussed the various functions (elements) of life table and their
relations. Let us discuss some of their relations in detail.
1. If d x is available but l x is not known, then we can use Lx as its estimate and define
central mortality rate mx as
d dx d /l qx 2 qx
mx = x = = x x = =
1 - qx 2 - qx
Lx l - d1 d
1 x 1
x 1-
2 x 2 lx 2
Also, if q x is to be computed then solving
2 qx
mx = for q x , we get
2 - qx
2 mx
qx =
2 + mx
2. So far, we considered l x will only the integral values of x, but since deaths occur at all
ages and at every fraction of time of the year, l x is a continuous function of x. At any
age x, the rate of decrease in l x is given by the expression.
lx - lx + t lx + t - lx dl x
lim = = - lim =-
t ®0 t t ®0 t dx
dl x
where is the differential coefficient of l x with respect to x. The force of
dx
mortality at age x is defined as the ratio of instantaneous rate of decrease in l x to
the value of l x . It is denoted by m x and is given by
1 dl x d
mx =- = - (log l x )
l x dx dx
141
1
3. To prove Tx = l x + l x + 1 + l x + 2 + ...
2
¥
By definition Tx = å Lx + t (as Tx = Lx + Lx + 1 + Lx + 2 ....)
t=0
¥ l
x + t + lx + t + 1 lx = lx + 1
= å 2
(as Lx =
2
)
t =0
¥
=
1
l +
2 x å lx + t
t =1
Abridged Life Table: In complete life table age interval is a year throughout and the
various function (like px , q x etc.) are obtained for every year of age. In abridged life table
the age interval is more than a year, say, 5 years. There are two types of abridged life
tables. In one form of abridgement the functions are tabulated for some integral values of
x which are at some distant apart, usually 5 years or 10 years. In the other type of
abridgement, the values of the functions are stated for 5 year or 10 years age-groups and
hence this type of life-table is obtained through a condensation of complete life-table.
The principle method used for construction of abridged life-table are

(i) George King's Method (1914)
(ii) T.N.E. Greville Method (1946)
(iii) L.T. Reed and M. Menell Method (1939)
George king's method uses the first kind of abridgement i.e. the various functions of life
table are evaluated for some integral value of x usually 5 years or 10 years. While the
method of Greville, Reed and Menell are meant for second type of abridgment in which
various functions of life-table are evaluated for 5 year or 10 year age-groups.
4.19 Statistical Organisation in India

The present statistical system in India consists of the Central Statistical Organisation,
National Sample Survey office the various statistical units in the different ministries of
the centre, the State Statistical Bureau, and the other statistical units attached to the
state ministries and office. Besides several institutions, like Indian Statistical Institute
(I.S.I), Indian Agricultural Statistics Research Institute (I.A.S.R.I.), Institute of
Economic Growth, the National Council of Applied Economic Research, Institute of
Research in Medical Statistics (I.R.M.S.), Institute of Applied Manpower Research and
several other are contributing to the statistical wealth of the country. Let us discuss some
of them.
142
4.20 Central Statistical Office (C.S.O)

Central statistical office was established in May 1951 by central government under
cabinet secretariat, with the objective of creating co-ordination of large variety of
statistical information, collected at the centre and state levels. It performs the following
functions :
1. Co-ordinate the available data
2. Interpretation of concepts and laying down standards
3. Advise on statistical matters
4. Computation of certain measures to aid in planning
5. Initiation of conferences and training
6. Publication of statistics
7. Compilation of national income estimates
8. Provide the Indian data to united nations statistical offices and other statistical
institutes.
9. Plan and co-ordinate the conduct of annual survey of industries.
This orgainisation does many other things concerning statistical matter, needed from
time to time. Each of the above functions is looked after by separate division with its
head-quarter at Parliament Street, New Delhi, excepts the industrial statistics wing at
Calcutta. Let us discuss the above functions in brief.
The C.S.O. examines various schemes running at state level and make specific
suggestions regarding the concepts and definitions of the survey. Now, C.S.O. has a
network all over the country. It is a coordinating agency for the statistics collected by
different ministries. It also co-ordinates National Sample Survey Organisation (NSSO).
The planning cell of C.S.O. established in 1955 undertakes studies on planning and
keeps in touch with Indian Statistical Institute Calcutta and the planning division of the
government of India. This cell makes a study about growth models, economic indicators,
plan progress etc. and prepares Annual Statistical Reports on the progress of the national
plan. The national income unit studies the different income aspects and makes national
income estimates for the centre, as well as for the states, in collaboration with various
statistical agencies at the state level. A National Income Unit (NIU) exclusively collects
data related to national income and publish it in the form of a white paper by an
industrial intelligence unit with regard to industries in India.
The C.S.O. supplies authentic data to foreign officials like united nations statistical
officers international agencies. It initiates conferences in the country and sends delegates
to the international conferences. It organises training courses for senior and junior
officers in collaboration with Indian Statistical Institute and the International Statistical
Education Centre.
143
The work of C.S.O. in developing concepts and techniques for population, production
and economic census and the improvement of statistical publication is valuable. Its
exhibition hall or graphical division makes pictorial presentation of statistical data of
national significance. The C.S.O. has close technical relationship with the State
Statistical Bureau and to help this organisation, there is a standing committee of
departmental statisticians. Every year joint conferences of central and state statisticians
is organized besides periodical meetings.
As a matter of fact, statistical information is collected an compiled in a vast field of

economic activity–like civil aviation, motor vehicles, inland trade, taxation, revenue,
agriculture, forest, fisheries, livestock, etc. This compiled information is published in one
and another form in the journals and bulletins. The following publications are released
by C.S.O. regularly :
(i) Monthly Abstract of Statistics
(ii) Weekly Bulletin of Statistics
(iii) Statistical Abstract of India (Annual)
(iv) Estimates of National Income (Annual)
(v) Monthly statistics of the production of selected industries
Other publications issued from time to time include the following :

(a) Reports of joint conferences
(b) Reports of working parties
(c) Key to current official statistics
(d) Statistical system of India.
4.21 National Sample Survey Office (NSSO)

The National Sample Survey Office (NSSO) is mandated to conduct nation-wide surveys
on various socio-economic aspects, Annual Survey of Industries (ASI), follow-up surveys
of Economic Census and supervision of area enumeration and crop estimation surveys of
state agencies under the Improvement of Crop Statistics (ICS) scheme, Urban Frame
Survey with a view to prepare a frame of compact units of urban areas as the first stage
sampling units (FSUs), organizing methodological studies and pilot surveys on
important subjects and collection of price data from rural and urban sectors. The NSSO
functions under the overall direction of a Governing Council (GC) with requisite
autonomy in the matter of NSS data collection, processing and publication. In addition
to the non-official Professional Chairman, the GC is composed of five academicians, five
data users from Central and State Governments and senior officers of the Ministry.
Director General and Chief Executive Officer (DG and CEO) is responsible for
coordinating and supervising all activities of the organization. He is Member-Secretary
144
of the GC and is assisted by one Additional Director General and 4 Deputy Director
Generals.
The NSSO has 4 Divisions viz.
(i) Survey Design and Research Division (SDRD); (ii) Field Operations Division (FOD);
(iii) Data Processing Division (DPD); and (iv) Coordination and Publication Division
(CPD). (i) The SDRD is headed by a Deputy Director General and locate at Kolkata is
responsible for planning of the survey, formation of sample designing of enquiry
schedules and preparation of reports based on survey results. (ii) The FOD is headed by
an Additional Director General, with headquarter at Delhi and a network of six Zonal
Offices at Bangalore, Guwahati, Jaipur, Kolkata, Lucknow and Nagpur, 48 Regional
Offices and 117 Sub-Regional offices spread throughout the country. The Division is
responsible for collection of primary data for the surveys undertaken. (iii) The DPD
headed by a Deputy Directory General with its headquarter at Kolkata has six Data
Processing Centres (Ahmedabad, Bangalore, Delhi, Giridih, Kolkata and Nagpur) and is
entrusted with the responsibilities of sample selection, software development, processing
and tabulating the data collected through surveys. (iv) The CPD headed by a Deputy
Director General, is the secretariat of the GC of the NSSO. Apart from coordinating the
activities of different Division, it brings out bi-annual Sarvekshana and quarterly NSSO
Bulletin.
4.22 Governing Council

The 86 th meeting of the GC was held of 28 th March 2003 where in it was decided that;
(i) The data on mortality should not be collected in the NSS 60 th Round, which is only
of six months duration, Morbidity and health care would be covered in this round, (ii)
Constitution of a Working Group (WG) under the Chairmanship of prof. N.
Bhattacharya to look into the technical details of the NSS 60th Round survey; (iii) The
Pilot Study on Non-Profit Institutions Serving Households (NPISH) would be taken up
during the survey period of NSS 60th Round and de-linked from the regular Survey of
the NSS 60 th Round (a Working Group under the chairmanship of Prof. S.N. Mishra
would look into the technical details of the Pilot Study); (iv) The field work for the study
on Food Security under UN World Food programme would be of six months duration
instead of three months Proposed earlier; (v) A pilot survey on suitability of appropriate
reference period for non-food items might be taken up; (vi) The estimation procedures
for NSS 57 th and 58 th Round were approved; (vii) The NSSO might work out the
details for data collection on current daily status though the consumer expenditure
schedule to be canvassed in the annual surveys from the 60 th NSS Round to meet the
data requirements of the Planning Commission for annual estimates of employment.
The 87 th meeting of the GC held on 7-8 th August 2003, approved (a) The proposal to
145
conduct the field work for Pilot Study on food Security under UN World Food
Programme survey during September 2003–February 2004, (b) Survey methodology for
the NSS 60 th Round, (c) Tabulation Plan along with the Sample Design and Estimation
Procedure for the NSS 59 th Round, and (d) Constitution of Working Group for the NSS
61st st Round (July 2004-June 2005) under the Chairmanship of prof. S.D. Tendulkar.
The 88 th meeting of the GC held during 25-26 February 2004 approved the sample
design and schedule of enquiry for the NSS 61st Round survey and for the Pilot study on
NPISH. The GC also approved the tabulation plan and estimation procedure for the
study on UN World Food Programme. Further, the proposal for taking up the Family
Living Survey (FLS) by the NSSO was approved in the meeting. The GC decided that the
FLS be conducted during the survey period of NSS 62 nd Round (July 2005 – June 2006),
which will be devoted to the survey on unorganized manufacturing. On the proposal of
the UNICEF to associated the NSSO only for data collection work for a baseline survey
in 41 districts, the GC desired, that the NSSO should take up the entire work up to
release of data and should also get joint authorship. The GC authorized the DG and
CEO, NSSO to constitute a Committee for finalizing the survey methodology of this
baseline survey.
4.23 Working Groups (WG)

The fourth meeting of the Working Group on 59 th Round of NSS was held on 4 th June
2003 under the chairmanship of Prof. G.S. Bhalla to discuss the Tabulation plan and
Estimation Procedure for the NSS 59 th Round
To formulate the sample design, schedules of enquiry and other technical details of the
survey relating to the 60 th Round of NSS, a Working Group was set up under the
chairmanship of prof. Nikhilesh Bhattacharya. The Working Group met during 5-6 June
2003. Considering the request of the Planning Commission for collection of
employment-unemployment data including current daily status for monitoring the
employment-unemployment situation in the country, the WG decided that a separate
schedule for employment-unemployment would be canvassed in the NSS 60 th Round,
so as not to make the consumer expenditure schedule too long and unwieldy. The WG
decided the draft schedule and structure on morbidityand health care. The second
meeting of the WG on NSS 60 th round held during 2-3 July 2003 approved the sample
design and various schedules of enquiry.
A WG set up under the chairmanship of Prof. S.N. Mishra, to look into the technical
details of the Pilot Study on NPISH met on 4 th July 2003 and recommended forming
of two separate sub-groups : (i) under the chairmanship of Shri M.Neelakantan, Addle,
DG, FOD and (ii) under the Chairmanship of Dr. Vaskar Saha, Addle. DG, NAD. The
representatives from RBI, FOD, SDRD and CPD were included. The first sub-group
146
would explore the range of frame-based operation and the second to look into the
coverage an definitional aspects on NPISH. The WG in its second meeting held on 10
November 2003 decided that the RBI should make available the list of NPIs maintaining
accounts with the branches of various banks in respect of 6 states where the pilot study
would be carried out. This list should be utilized for updating the draft sample design.
The WG recommended that the field work for the Pilot study could be carried out during
April-September 2004. The sample design and schedules for the Pilot Study were
finalized in the third meeting held during 6-7 February 2004.
The first meeting of the WG on NSS 61st Round, held on 20-21 October 2003,
discussed the draft sample design and schedules of enquiry on household consumer
expenditure are employment. The issue relating to the abridgement of consumer
expenditure schedule was also discussed. The WG constituted a sub-group under the
chairmanship of Dr. Pranab Sen to work out a code list of vocational skills and formal
vocational training and also to formulate a few questions relating to the informal sector
for data collection. The WG continued the discussions in the second meeting held on
20-21 November 2003. The WG deliberated on the Employment-Unemployment
schedule and sampling design in the third meeting held on 5 January 2004. The survey
instruments including the sample design and schedules of enquiry where finalized in the
fourth meeting held during 19-20 February 2004.
4.24 Socio-Economic Surveys
4.24.1 NSS 57 th Round

NSS 57 th Round (July 2001-June 2002) was devoted to collection of data on economic
and operational characteristics of unorganized enterprises in service sector (excluding
trade and finance). Reports brought out by the NSSO on the basis of the recent survey
result are being sold on magnetic media as well as in the form of hard copies the nominal
prices to institutions/individuals. Validated unit level data (household/enterprise) are
also available to the public for further research and analysis. Unit level data pertaining to
this round has been released in September 2003. The following reports pertaining to this
round have been released :
Report No. Title of the Report
481 Household Consumer Expenditure and employment-Unemploy-

ment Situation in India, 2001-02
482 Unorganized Service Sector in India-2001-02 : Salient Features
483 Unorganized Service Sector in India-2001-02: Characteristics or

Enterprises
147
The NSSO organized a National Seminar on the results of NSS 57 th Round survey
results at Jaipur on 24 March 2004. Thirteen papers were presented in the three
technical sessions of this seminar by the authors from research institutions and
government organizations.
4.24.2 NSS 58th Round

The 58 th Round of NSS (July–December, 2002) was earmarked for collection of data on
disability, housing condition, village facilities and slum particulars besides household
consumer expenditure and employment-unemployment. A press note on 'Condition of
Urban Slums' based on the NSS 58th Round Survey was released on 8 March 2004. The
following reports have been released and two reports on Housing are expected to be
released shortly :
Report No. Title of the Report
484 Household Consumer Expenditure and employment-Unemployment

Situation in India.
485 Disabled Persons in India
486 Condition of Urban Slums
487 Report on Village Facilities

The NSS 59 th Round (January–December 2003) was devoted to collection of data on
land and livestock holdings; debt and investment and household consumer expenditure
along with some key characteristics of employment-unemployment. The Situation
Assessment Survey (SAS) of farmers was also conducted during this round on Behalf of
Union Ministry of Agriculture and Cooperation. This programme was carried out jointly
by the NSSO and the State Directorates of Economics and Statistics. All the States and
Union Territories except Andaman and Nicobar Islands, Dadra and Nagar Haveli and
Lakshadweep participated at least on equal matching basis. The total sample size
(central) was 10576 villages and urban blocks. Each sample village and urban block was
visited twice during this round. Team approach was followed in this round also.

The 60 th Round of NSS (January-June 2004) is devoted to Morbidity and Health Care,
Employment and Unemployment and Household Consumer Expenditure. On the
subject of Morbidity and Health Care, the Survey covers information on Morbidity and
utilization of Health Care Services provided by the public and private sectors, together
with the expenditure incurred by the households for availing these services. Besides,
collection of comprehensive information relating to Morbidity, information on
Maternity and Child Care and the problem of aged persons are also collected from a
148
sample (central) of 7612 villages and urban blocks. At the instance of Planning
Commission, a separate enquiry on employment-unemployment situation is also
undertaken during the 60 th Round of survey in order to obtain an assessment of the
latest employment scenario in the country. In addition, as usual the annual survey of
household consumer expenditure is also undertaken during this period.
4.24.5 NSS 61st Round

The seventh quinquennial survey on consumer expenditure, employment and
unemployment (NSS 61st Round) will be conducted during July 2004 – June 2005.
Sampling design and survey instruments required for conducting this round survey were
finalized.
4.25 Sarvekshana, NSSO Bulletin

'Sarvekshana' has been converted into a technical journal of the NSSO and one issue of
the journal (Issue No. 86) has been brought out during the year. The 86th issue
contained the technical paper on magnitude of the Women Work Force in India and
integrated summary of the NSS 55 th Round (July 1999 – June 2000) Consumer
Expenditure Survey Results. NSSO Bulletin is the main medium to keep all the officers
and staff of the NSSO informed about the developments and activities taking place in
the NSSO. Two issues were brought out during the year.
4.26 United Nations World Food Programme (UNWFP)

The United Nations World Food Programme (UNWFP) has started India Country
Programme (2003-08) and is confined to 36 food insecure districts in Chhattisgarh,
Gujarat, Jharkhand, Madhya Pradesh, Orissa, Rajasthan and Uttaranchal. Under the
WEP programme the beneficiary households will receive various types of food assistance
entitlements. The NSSO on behalf of the UNWFP has taken up a pilot survey on food
security and related aspects for enabling UNWFP to develop suitable methodology for
evaluating their approach as well as for monitoring key human indicators. The field work
of the pilot survey was conducted from 1st September 2003 to 29 th February 2004 in 6
selected most food insecure districts in the country namely:
1. Sarguja in Chhattisgarh,
2. Surendra Nagar in Gujarat,
3. Jhabua in Madhya Pradesh,
4. Koraput in Orissa,
5. Banswara in Rajasthan and
6. Chamoli in Uttaranchal.
149
The primary objective of this pilot survey was to collect detailed information from the
rural households in the selected districts so that district level estimated of some specific
human development indicators of food security namely – percentage expenditure on
food items, nutritional intake, access to different food assistance schemes, seasonal
migration, indebtedness, enrolment in elementary schools etc. are worked out.
4.27 Agricultural Statistics

For Agricultural Statistics, the NSSO has the responsibility of guiding the states for
developing suitable technique for obtaining reliable and timely estimates of area and
yield and ensuring adoption of uniform concepts, definition and procedures in the Crop
Estimation Surveys (CES). It also keeps a constant watch on the quality of Crop
Statistics through the ICS scheme, which envisages locating deficiencies in the State
system of collection of Crop statistics and suggests suitable remedial measures. Under
the scheme, sample checks on the primary field work relating to area enumeration and
area aggregation in approximately 5000 villages and supervision of about 15000 crop
cutting experiments are undertaken by the FOD in each Agricultural Year. The States are
also participating in the programme of sample check on matching basis. In addition,
Sample check on aggregation of area figures above the village level i.e. at Tehsil, District
and State level are also undertaken since 1980-81 in 19 Land Record States spread over
36 Regions of the NSSO. The work relating to Sample Check on Area Enumeration was
carried out in 1270, 4949, 4649 and 3841 villages during Early Kharif, Kharif, Rabiand
Summer season of 2002-2003 respectively. Sample Check on Aggregation of Area was
carried out in 628, 4142, 4149 and 3347 villages during these seasons respectively.
Progress for Sample Check on Aggregation of Area is slightly less due to non-completion
of work by the State primary workers. Further, 15654 Crop Cutting Experiments were
supervised as planned during 2002-2003. For Agricultural Year 2003-2004 starting in
July, 2003, Sample Check of Area Enumeration was completed in 1268 villages during
Early Kharif, 4949 villages during Kharif and 3898 villages during Rabi and 176 villages
during summer season upto March 2004 as against 1270, 4949 and 4055 villages
planned for the Agricultural Year 2003-2004. Work relating to Aggregation of Area was
completed in 626 villages during Early Kahrif, 3502 villages during Kharif, 1888 villages
during Rabi and 160 villages during Summer up to March 2004 as against 628, 4285,
4399 and 3651 villages planned for the Agricultural Year 2003-2004. For Sample Check
on Crop Cutting experiment 1160, 8080, 2689 and 176 experiments were supervised
upto March 2004 during Early Kharif, Kharif, Rabi and Summer season as against 1200,
8342, 5210 and 980 experiments planned respectively for the Agricultural Year
2003-2004. The data collected through the supervision of crop cutting experiments at
harvest stage are also utilized to computer the advance estimates of yield rate of specified
crops for the States covered under the ICS scheme during the Agricultural Year
2002-2003. Estimates of yield rate for as many as 170 crops pertaining to these States
150
where forwarded to the Union Ministry of Agriculture as per the predecided time
schedule. Finding under the ICS scheme are brought out in the form of reports on the
Status of Estimation of Crop Production in States as well as in all-India. The Statewise
reports are brought out for Kharif and Rabi/Summer season of each Agricultural Year. A
consolidated report on the 'Review of Crop Statistics System in India through ICS' is also
prepared annually. Statewise status reports (Final) for 2000-2001 (Kharif an Rabi) have
been issued for all the nineteen States and for 2001-2002 (Kharif season) in respect of
fourteen States have been sent to the DES and other agencies for further necessary
action. "Consolidated Results of Crop Estimation surveys on Principal Crops" –a report
for the year 2000-2001 has been brought out.
High level Coordination Committees (HLCCs) have been formed in the States to
provided a link for proper coordination between the Centre and States forming a basis
for the improvement in quality and maintaining timeliness of collection of Agricultural
Statistics. During the year 2003, meeting of HLCC where held in Andhra Pradesh,
Haryana, Karnataka, Madhya Pradesh, Rajasthan, West Bengal, Gujarat and Bihar.
District Level Committees (DLCs) are also unfunctioning with the task of examining the
reasons at the field level, for large variations or no variations at all in the entries recorded
by the Patwaris and as observed by Central/State supervisors during sample check on
area enumeration work under the ICS. The functioning of the committee goes a long way
in effecting improvement in the State system of crop statistics. The response noticed
under this scheme during the reference year is quite encouraging.
Two Pilot surveys where conducted during the year 2002-2003 as detailed below :
1. Pilot Study for Working out a Correction Factor Based on Data Pertaining to
ICS : In pursuance of the recommendations of the National Statistical Commission
(NSC) on Agricultural Statistics, a Pilot surveys was conducted in two districts each
of Uttar Pradesh, and Tamil Nadu for formulating Correction Factor based on ICS
data for calibration of estimates of area at State level.
2. Pilot Study Under RKBY : Under Rashtriya Krishi Bima Yojana a pilot study has
been undertaken in one district each of 5 States of Uttar Pradesh, West Bengal,
Tamil Nadu, Madhya Pradesh and Maharastra for estimating crop yield at
Panchayat level, using Farmers Appraisal Survey data. The report on the pilot study
has been sent to the Union Ministry of Agriculture and Cooperation for necessary
action at their end. Based on the conclusions from the study the methodology of
Small Area Estimation under GCES will be evolved.
4.28 Price Data Collection

1. Urban Retail Price Collection : The FOD has been collecting the retail price data
on weekly basis from 59 urban centres for compilation of Consumer Price Index
Number in respect of Urban Non-Manual Employees (UNME) by the CSO. Apart
151
from the collection of price data, house rest data of rented dwellings and off take
data from various sources of purchases are also collected at the interval of six
months for compilation of this index. Details are available in Chapter-III.
2. Rural Retail Price Collection : The FOD is also responsible for collecting rural
retail price data on monthly basis from 603 sample villages and also daily wage rates
of 18 major agricultural and non-agricultural occupations for Labour Bureau,
Ministry of Labour. Prices are used for compiling Consumer Price Index numbers
for Agricultural/Rural Labourers by the Labour Bureau. The DPD process the rural
price data and brings out quarterly bulletin on 'Prices and Wages in Rural India'.
4.29 Urban Frame Survey

The Urban Frame Survey (UFS) is one of the most important surveys conducted by the
FOD regularly. For conducting Socio-Economic enquiries in the urban areas, an updated
frame of First Stage units is a pre-requisite. The UFS is therefore carried out to meet the
need of an updated frame. Under this programme, every town/city/urban agglomerations
and out growths are updated over a period of five years which is known as phase. The
UFS blocks are formed (lime enumeration blocks in Census Operations) taking into
account the increase/decrease in the population of cities/town/urban agglomerations and
also in newly declared towns. Each town/city is divided into compact UFS blocks with
clear cut identifiable boundaries and permanent land marks having a population content
of 600 to 800 and 120 to 160 number of households. The current phase 2002-2007 is
going on w.e.f. July 2002 covering a total 4465 towns as per census 2001 comprising
about 406790 blocks. Out of the allotment of phase 2002-2007, 1008 towns comprising
91491 blocks have been updated by the end of March 2004. Apart from the primary use
as the First Stage Sampling Frame in Socio Economic Survey Rounds, the maps of UFS
blocks are also being provided free of cost to various Government Departments and @
` 100.00 per UFS blocks map to other private user agencies and research institutions. An
amount of ` 14,87304 has been realized so for on account of sale of UFS block maps.
4.30 Industrial Statistics

The Industrial Statistics come under four categories viz.
1. Annual Survey of Industries
2. Index of Industrial Production
3. Standardization of Industrial Classifications and
4. Energy Statistics.
The details of Industrial Production and Standardisation of Industrial Classifications

have been indicated in Chapter-II.
152
4.31 Annual Survey of Industries

The Annual Survey of Industries (ASI) provides statistical information to assess and
evaluate the change in the growth, composition and structure of the organized
manufacturing sector comprising activities relating to manufacturing processes, repair
services, generation, transmission and distribution of electricity, gas water supply and
cold storage. The survey is conducted annually under the statutory provisions of the
Collection of Statistics Act, 1953. The ASI extends to the entire country except the
States of Arunachal Pradesh, Mizoram, Sikkim and Union Territory of Lakshadweep. It
covers all factories registered under sections 2m (i) and 2m (ii) of the Factories Act, 1948.
The survey also covers bidi and cigar manufacturing establishments registered under the
Bidi and Cigar Workers (Conditions of Employment) Act, 1966. All the electricity
undertakings engaged in the generation, transmission and distribution of electricity
registered with the Central Electricity Authority were covered under ASI irrespective of
their employment size. The electricity units, however, have been excluded from ASI
1998-99 onwards as the relevant information of such electricity undertaking are
collected, compiled and disseminated by the Central Electricity Authority (CEA).
Certain services and activities like cold storage, water supply, repair of motor vehicles
and of other consumer durables like watches. etc., are covered under the survey. Defence
establishment, oil storage and distribution depots, restaurants, hotels, cafe and computer
services and also the technical training institutes are excluded from the purview of the
survey. The data collected through ASI relate to capital, employment and emoluments,
consumption of fuel and lubricants, raw-material and other input/output, value added,
labour turnover, absenteeism, labour cost, construction of houses by employers for their
employees and other characteristics of factories/industrial establishments. Field work is
carried out by the Field Operations Division, NSSO. The CSO processes the data and
publish the results.
All ASI works are undertaken in the CSO, Industrial Statistics Wing at Kolkata from
1993-94 and onwards. The works in IS Wing has been geared up in order to :
1. Develop in-house computer software for data processing for generation of all tables
and their publications with effect from ASI 1993-94;
2. Maintaining the timeliness for release of detailed ASI data annually within six
months of data collection; and
3. Print ASI detailed reports with the help of in-house printing facility.
In recent past many changes have been made for improvements in the collection,
processing and dissemination of ASI data. Over the last few decades there has been a
spurt in the growth of the number of registered factories and consequently in the number
of units from whom data are collected and analysed annually. The collection and
compilation of data from about 60000 units in a very lengthy schedule have not only led
153
to delay in the availability of results but also involve non-sampling errors. The process of
changing the scenario started from 1997-98 which involved modification of sampling
design as well as the schedule. According to the new sample design about 43200 units
where surveyed through ASI 2000-2001 and about 44610 units where surveyed through
ASI 2001-2002. This has certainly reduced the data collection and processing time. For
ASI 2002-2003, the sample selection is over and 44,595 units have been selected for the
purpose of estimating the state and industry level estimates with a better precision. The
entire processing of data of ASI 1995-96 onwards is being done at CSO (IS Wing),
Kolkata. Necessary hardware and software are available to introduce modern
client-Server architecture and networking. Under the current plan scheme, the client
server architecture with RDBMS as oracle engine has been introduced very recently to
streamline the large data processing to release the reliable data within a very short time
span.
An action plan has been drawn to reduce the time lag in releasing the ASI results. From
ASI 1998-99, release of provisional result has been discontinued. Detailed final results in
tabular form (in two volumes) and Quick Estimates are being released. From ASI
1999-2000, decision has been taken to have a separate short schedule for all the units
selected and on the basis of these schedules, CSO (IS Wing) has released state wise Quick
Estimates on some important parameters at 2 digit level of NIC-98. The ASI results are
available to users on electronic media on payment basis. CSO (IS Wing) provides
necessary technical support to State Government and FOD, specially in respect of
computerization of ASI frame, fully computerized sample selection, data processing as
well as analytical works on ASI data. The liaison work is also undertaken with State
Governments and other organization/agencies in India and abroad. Some State DESs
process ASI data on the basis of ASI returns received from NSSO (FOD). State DESs
have been advised to modified their action plan in ASI in the light of the new sampling
design with reduced sample size and to avoid the duplication of work in ASI at the State
level. The CSO (IS Wing) provides the database as well as tables needed by State
Governments for generating district/micro level estimates.
4.32 Labour Bureau

It comes under the ministry of labour, employment and realization, government of India.
Labour bureau is mainly responsible for labour statistics. It publishes a number of
journals, books and adhoc reports. Some main publications are :
1. Pocket book of labour statistics (Annual)
2. Indian labour journal (Monthly)
3. Indian labour statistics (Annual)
4. Trade union in India (Biennial)
154
4.33 Army Statistical Organisation (ASO)

Army statistical organisation, under the ministry of defence is mainly responsible for all
statistical data related to the army and acts as a central clearing agency for statistics in
the army. The organisation also carries out statistical analysis for research problems and
surveys.
4.34 Census of India Organisation

The Indian Census is the largest single source of a variety of statistical information on
different characteristics of the people of India. With a history of more than 130 years,
this reliable, time tested exercise has been bringing out a veritable wealth of statistics
every 10 years, beginning from 1872 when the first census was conducted in India
non-synchronously in different parts. To scholars and researchers in demography,
economics, anthropology, socioogy, statistics and many other disciplines, the Indian
Census has been a fascinating source of data. The rich diversity of the people of India is
truly brought out by the decennial census which has become one of the tools to
understand and study India.
The responsibility of conducting the decennial Census rests with the Office of the
Registrar General and Census Commissioner, India under Ministry of Home Affairs,
Government of India. It may be of historical interest that though the population census
of India is a major administrative function; the Census Organisation was set up on an
ad-hoc basis for each Census till the 1951 Census. The Census Act was enacted in 1948
to provide for the scheme of conducting population census with duties and
responsibilities of census officres. The Government of India decided in May 1949 to
initiate steps for developing systematic collection of statistics on the size of population,
its growth, etc., and established an organisation in the Ministry of Home Affairs under
Registrar General and ex-Officio Census Commissioner, India. This organisation was
made responsible for generating data on population statistics including Vital Statistics
and Census. Later, this office was also entrusted with the responsibility of
implementation of Registration of Births and Deaths Act, 1969 in the country.
Population Census 2011
The Census 2011 is the 15 th National census survey conducted by the Census
Organization of India. Mr. C. Chandramouli is the Commissioner & Registrar General of
the Indian 2011 Census.
The 2011 Indian National Census has been conducted in 2 phases - house listing and
population. The national census survey covered all the 28 states of the country and 7
Union territories including 640 districts, 497 cities, 5767 tehsils & over 6 lakh villages.
155
The Indian Population Census 2011 covered a numbre of parameters during the survey.
These parameters include population, growth rate in population, rate of literacy, density
of population, sex ratio and child sex ratio (0-6 years). According to the census reports of
Indian Census 2011, the population of India is 1,210,193,422 with 623, 724, 248 males
and 586,469, 174 females. The total literacy rate in the country at present is 74.04%.
The density of population is 382 persons/sq.km. In regards to sex ratio, at present there
are 940 females on average on per 1000 males and the child sex ratio is 914 females per
1000 males.
The Indian Census survey is conducted to gather information from the grass root base
which is essential to launch different welfare schemes like Annual Plans, 5 Year Plans etc.
in the country for the privilege of common man here. The Houselisting & Housing
Census provides extensive information on the status of the human settlements, the
housing deficits as well as the different housing necessities to be looked for formulating
the housing welfare policies.
The Population Enumeration offers needed Census data about land & its people in the
present time. The survey reveals the current population trends, its varied characteristics
that are valuable inputs for planning sound programs and policies aimed towards the
welfare of India & her people and also for effective public administration.
4.35 Some Non-Government Statistical Organisations

Non-government statistical organizations include the statistical-cum-economic agencies
outside the government, constituted under statutory provisions such as :
1. Indian Statistical Institute (ISI)
2. Indian Agricultural Statistics Research Institute (IASRI)
3. National Council of Applied Economic Research (NCAER)
4. Institute of Applied Manpower Research (IAMR.)
5. Institute of Economic Growth (IEG)
6. Institute of Research of Medical Statistics (IRMS)
7. Indian Institute of Population Studies (IIPS)
8. Institute of Labour Research (ILR)
9. Statistics and Economics Departments of Reserve Bank of India.
Besides these non-government organisation, each ministry has a statistical unit which is
responsible for collection and publication of statistics related to the ministry.
All the states and union territories in India have Directorate of Economics and Statistics
of Bureau of Statistics. The state statistical organizations collect and publish the data
relating to the state. The State Agricultural Statistical Authority (SASA) is responsible
to agricultural statistics.
156
4.36 Statistical Organisation in States

On the recommendations of Pragori Committee in 1946, a State Statistical Bureau is
established in each state. At present there is a three stage decentralized statistical system
in force. In this system each district consists of district statistical officer who is the
incharge of district statitstics office. At departmental level there are statistical branches
in each ministry and at state, directorate of economics and statistics is established at apex
level. In some states it is called state Statistical Bureau.
4.37 Statistical Organisation in U.P. and Uttaranchal

The state of Uttaranchal is new and generated from the hill parts of U.P. The system in
Uttaranchal is same as that of U.P. There is a four tier system popular in both the states,
which are given below :
1. Nodal or Apex Level: The collection, analysis and publication of data at state level
is carried out by the Department of Economics and Statistics. Its head is Director
under which a number of Assistant Directors, Deputy Directors, Joint Directors
and Statistical Officers work.
2. Division Level : It is the middle stage between state and district which responsible
for the coordination and monitoring of the district statistical officers.
3. District Level : In each district, there is a District Statistical Office (DSO) which
works under the direct control and supervision of department of Economics and
Statistics. Its work is to assist in agricultural census, economics census and to
coordination the different surveys in the district. The results are also sent to
Department of Economics and Statistics by District statistical office.
4. Block Level : In each block of district there is an Assistant Development Officer

(Statistics) which is mainly responsible for collection of primary data.
Some of the publications are listed below :

(i) Statistical Abstract (U.P.)
(ii) Statistical Diary
(iii) Uttar Pradesh ki Arthik Samiksha
(iv) Shramjivi (Monthly)
(v) Labour Bulletin (Monthly)
(vi) Bulletin of Agricultural Statistics for U.P. (Annual)
(vii) Co-operation in U.P. (Annual)
Besides these publications, many more publications are released by different

co-operative societies, directorate, etc.
157
E XERCISE
1. What do you mean by vital statistics ? Mention the various sources from where the
vital statistics data can be obtained.
2. What are vital statistics ? How are they collected ?
3. What are vital events ? Which of these events is most important and why ?
4. Define crude death rate and standardised death rates. [C.C.S. Meerut 2000 (Back), 01]
5. Explain the terms 'Age specific fertility rate and ' Total Fertility rate' as used in vital
statistics. [C.C.S. Meerut 2002, 04]
6. Explain gross and net reproduction rates. Why are they unsatisfactory indices of
population growth ?
7. Describe the various columns of a life table and point out the relations that exist
between them. [C.C.S. Meerut 2005]
8. What is a life table ? Why is it constructed ? Explain its elements.
9. What is life table ? Describe the various columns of life table and establish the
relationship between them. [C.C.S. Meerut 2012]
10. Calculate the general fertility rate, total fertility rate and gross reproduction rate
from the following data, assuming that for every 100 girls 106 boys are born.
Age of women Number of women Age specific fertility rate
(per 1000)
15-19 212619 98.0
20-24 198732 169.6
25-29 162800 158.2
30-34 145362 139.7
35-39 128109 98.6
40-44 106211 42.8
45-49 86753 16.9

158
11. Compute the standarised death rate for the following data:
Age Group Population of Deaths in the Standard
(year) District A District A Population
0-5 1000 50 10000
5-10 800 20 8000
10-25 1200 12 12000
25-45 3000 15 30000
45 and over 4000 52 40000
Total 10000 149 100000
12. Find out the values of a, b, c, d, e and f of the following life table:
Age (x) lx dx qx px Lx Tx e0x
20 693435 2762 a b c 35081126 d
21 690673 — — — — e f

13. Fill in the blanks in a portion of life table given below:
Age in years lx dx px qx Lx Tx e0x
4 95000 500 ? ? ? 4850300 ?
5 ? 400 ? ? ? ? ?
14. Calculate e0x from the following data
Age (x) lx Tx
0 1273 —
1 1129 —
2 1090 58510
3 1066 57432
4 1049 56374
15. The values of l x i.e., the number of persons living at age x, are given below:
x: 102 103 104 105 106 107 108
lx : 97 59 32 75 6 2 0
Calculate the remaining entries of the life table for ³ 102.

159

1. Name some of the vital events.
2. Give the sources of collection of vital statistics.
3. Distinguish between stable and stationary population.
[C.C.S. Meerut 2001, 04, 06 Old, 06]
4. Explain in brief the gross and net reproduction rates. [C.C.S. Meerut 2001, 02, 04]
5. What do you mean by specific death rate ? [C.C.S. Meerut 2002]
6. Give the formula of gross reproduction rate. [C.C.S. Meerut 2002]
7. What is the relation between mx and q x ? [C.C.S. Meerut 2004, 06 Old]
8. Write the name of any two methods of obtaining vital statistics.
[C.C.S. Meerut 2004, 06 Old]
9. Cohort (radix) is denoted by which element ?
10. Write all the elements in sequence of a life table.
11. Define crude birth rate and total fertility rate. [C.C.S. Meerut 2002 (Back)]
12. Show that m = 2 q x .
x
2 - qx [C.C.S. Meerut 2004 (Back)]
13. What is meant by saying that the N.R.R. for a country is 1.1 ?
Hint: if N.R.R.=1, we may conclude that if the conduct fertility and female
mortality sites prevail in future, then a group of new born girls will exactly replace
itself in the next generation. On the other hand if N.R.R. in greater than unity then
the population has a tendency to increase, while N.R.R. is less than unity indicates
a declining population.
14. Define crude birth rate and general fertility rate. [C.C.S. Meerut 2005, 08]
15. Explain stable and stationary population. [C.C.S. Meerut 2005, 12]
16. What is a life table ? [C.C.S. Meerut 2005]
17. Write two uses of vital statistics. [C.C.S. Meerut 2005]
18. Give the formula of crude death rate. [C.C.S. Meerut 2006 (Old)]
19 What do you mean by Gross reproduction rate? [C.C.S. Meerut 2006 (Back)]
20. What is net production rate ? Define in brief. [C.C.S. Meerut 2007 (Back)]
21. Explain the term A.S.F.R and T.F.R as used in vital statistics. [C.C.S. Meerut 2012]
22. Write note on standardised death rate. [C.C.S. Meerut 2008]
23. Explain abridged life table, how it differ from complete life table ?

1. Which of the following is not a vital event ?
(a) Marriage (b) Education
(c) Migration (d) Birth [B.R.A. Agra 2006, 08, 12]
160
2. Net reproduction rate is used for the measurement of :

(a) Birth rate (b) Death rate
(c) Population growth (d) Stable population [B.R.A. Agra 2004, 11]
3. In life table e0x is calculated by :

l d
(a) x (b) x
dx lx
Tx
(c) (d) None of these
lx [B.R.A. Agra 2010]
4. A population maintaining a constant growth rate is said to be a :
(a) Stable population (b) Stationary population
(c) Mobile population (d) None of these [ B.R.A. Agra 2010]
5. Vital statistics are generally expressed as :

(a) Fraction of population (b) Per hundred of population
(c) Per thousand of population (d) None of the above [ B.R.A. Agra 2006, 09]
6. In life table q x is calculated by

l l
(a) x (b) x
dx ex
dx e
(c) (d) x
lx lx [ B.R.A. Agra 2009]
7. Gross reproduction rate is :
(a) Greater than net reproduction rate
(b) Less than net reproduction rate
(c) Equal to net reproduction rate
(d) None of the above [ B.R.A. Agra 2005]
8. Child bearing age in India is :

(a) 10-14 year (b) 15-19 year
(c) 50-65 year (d) None of the above [ B.R.A. Agra 2004]
9. Which age group is taken as child bearing age of women ?

(a) 15-40 years (b) 15-49 years
(c) 15-60 years (d) 20-49 years
10. Vital statistics is calculated by :

(a) Vital regestration (b) Demographic surveys
(c) Census (d) All of these
11. Which of the following is not covered in vital events ?

(a) Birth (b) Death
(c) Separation (d) Status
161
12. The registration of vital statistics in India suffers from :

(a) Non-reporting (b) Incomplete coverage
(c) Lack of accuracy (d) All the above
13. The death rate of babies under one month is known as :

(a) Infant mortality rate (b) Neonatal mortality rate
(c) Materral mortality rate (d) Fetal death rate
14. The death rate belonging to the age group (0 - 1) years is known as :
(a) Infant mortality rate (b) Crude death rate
(c) Crude mortality rate (d) None of these
15. NRR = 1 means :
(a) Female population will exactly replace itself
(b) Population remain constant
(c) There is no mortality in female birth till their child bearing age
16. Life table is also called :

(a) Survival table
(b) Mortality table
(c) Life expectancy table
17. Life table of human profile exhibits :

(a) Probability of dying at each age
(b) Numbers of cohorts living and dying
(c) Expectation of life at each age
18. A population is said to be stationary if :

(a) Number of deaths and births are exactly equal and distributed uniformly
throughout the year.
(b) The population is closed against migration
(c) The age and sex composition is fixed
19. A population is said to be stable if :

(a) The age and sex distribution is fixed
(b) Constant rate of mortality and fertility are experienced
(c) The population is closed against migration
162
20. Life table is based on :

(a) Stationary population (b) Stable population
(c) Population closed for migration (d) None of these
21. Life table are used for :
(a) Population projection (b) Size of the population at some future date
(c) (a) and (b) both (d) Neither (a) nor (b)
22. Life table is used for :

(a) Actuarial work
(b) Measurement of growth and population projection
(c) Analysing the effects of mortality on the age and sex composition of a population.
23. A life table consists of :

(a) Eight columns (b) Nine columns
(c) Seven columns (d) None of these
24. Normally a life table is constructed for an age interval of :

(a) 5-10 years (b) 5 years
(c) One year (d) None of these
25. An abridge life table can be constructed by the method suggested by :

(a) Greville (b) Reed menell
(c) King (d) All of these
26. The central mortality rate ' mx ' in terms of q x is given by :

qx 2 qx
(a) (b)
2 - qx 2 + qx
2 qx qx
(c) (d)
2 - qx 2 + qx
27. The death rate obtained for a segment of a population is known as :

(a) Specific death rate (b) Standard death rate
(c) Crude death rate (d) None of these
28. The total number of years lived by the Cohort to attaining the age x is given by
(a) Tx (b) Lx
(c) e x (d) d x [ B.R.A. Agra 2005, 08]
29. Infant mortality rate is computed for children:
(a) Above 5 years age (b) Between the age of 1 to 5 years
(c) Under the age of 1 year (d) None of the above.

163
30. Total fertility rate provides the basis for
(a) The expected maximum family size
(b) Population projection
(c) Population increase in a desired period
31. The ratio of births to the total deaths in a year is called:
(a) Survival rate (b) Total fertility index

(c) Vital index (d) Population death rate
32. If Tx = 3 and l x = 4 then what will be the value of e0x in life table:
(a) –0.1 (b) 12
(c) 7.0 (d) 0.75
33. The standard number of births 10,000 originating a life table is known as:
(a) Cohort (b) Initial population
(c) Radix (d) All the above
34. The death rate belonging to the age group 0-1 years is called as:
(a) Infant mortality rate (b) Crude death rate
(c) Crude fertility rate (d) None of the above
35. Crude birth rate is calculated with formula:

P B
(a) ´100 (b) ´1000
B P
B
(c) ´100 (d) None
Px
36. The formula of Gross Reproduction Rate (GRR) is:
f Bx fP
(a) fix = ´100 (b) fix = x ´100
f Px fB x
f Px f Bx
(c) fix = ´1000 (d) fix = ´100
f Bx f Px
37. Which is the sources of collection of vital statistics?
(a) Vital Registration (b) Demographic survey
(c) Census (d) All
38. If the value of net reproduction rate greater than unity then:
(a) Increase in population (b) Negative increases in population
(c) Zero increase in population (d) None of the above.

164
39. The probability of dying of a person of age between x and (x + 1) years is known as:
(a) Age specific death rate (b) Infant mortality rate
(c) Central mortality rate (d) None of the above
40. The range of vital index is :
(a) 0 to ¥ (b) 0 to 1
(c) 1 to 1 (d) None of the above
41. Vital rates are expressed as :

(a) Percentages (b) Per thousand
(c) Per million (d) Per trillion
42. Vital statistics data is collected by :
(a) Vital registration (b) Demographic surveys
(c) Census (d) All of the above
43. The total number of years lived by a cohort after attaining the age x is:
(a) Tx (b) I x
(c) e x (d) d x
44. In a life table q x is equal to
(a) Tx / l x (b) d x / l x
(c) 2 q x / 2 – q x (d) q x / 2 – q x
45. In life table l x + n is equal to
(a) l x + nPx (b) l x – nPx
(c) l x . nPx (d) l x / nPx
Fill in the Blank

1. In vital statistics we study events about .......................... life. [B.R.A. Agra 2007]
2. Tx = e0x ............................ . [ B.R.A. Agra 2007]
3. In life table q x is given by ............................. .
4. Vital rates are generally expressed in ............................. .
5. A population maintaining constant growth rate is called ..............................
population.
6. The child bearing span for a woman is .............................. .
7. N.R.R. is used to measure .............................. .
8. A life table consists of .............................. columns.
9. The death rate under one year child is called .............................. mortality rate.
10. A life table is used for population .............................. .
165
True or False
1. In a life table, the total number of years lived by a cohort after attaining the age x is
denoted by e x . [ B.R.A. Agra 2012]
2. A population maintaining constant growth rate is said to be stationary population.
[ B.R.A. Agra 2011]
3. N.R.R. can not be greater than G.R.R.
4. Vital statistics can be obtained by demographic surveys only.
5. Child bearing span for women is 10-42 years.
6. Most serious draw back of CDR is that it does not take into account the age and sex
distribution.
7. Marriage is a vital event.
8. Cohort (index) is denoted by L0 .
9. A population maintaining a constant growth rate is called stable population.
10. Infant mortality rate is considered for the children under one year of age.
A NSWERS
10. G.F.R. = 1145.6642, T.F.R. = 3619, G.R.R. = 1756.7961
11. 14.9 per thousand
12. a = 0.00398, b = 0.99602, c = 692054, d = 50.59, e = 34389072, f = 49.79
13. p4 = 0.9947, q4 = 0.0053, L4 = 94750, L5 = 94300, e40 = 5106

. , e50 = 50.32
p5 = 0.9958, q5 = 0.0042, T5 = 4755550
14. –, –, 53.6789, 53.8761, 53.7407j

1. (b) 2. (c) 3. (c) 4. (a) 5. (c)
6. (c) 7. (c) 8. (d) 9. (b) 10. (d)
11. (d) 12. (d) 13. (b) 14. (a) 15. (d)
16. (b) 17. (d) 18. (d) 19. (d) 20. (a)
21. (c) 22. (d) 23. (a) 24. (c) 25. (d)
166
26. (c) 27. (a) 28. (a) 29. (c) 30. (d)
31. (c) 32. (d) 33. (c) 34. (a) 35. (b)
36. (a) 37. (d) 38. (a) 39. (c) 40. (a)
41. (b) 42. (d) 43. (a) 44. (b) 45. (c)
Fill in the Blank

1. Human 2. lx 3. dx 4. thousands
lx
5. stable 6. 15-49 years 7. population 8. eight

growth
9. infant 10. projection
True or False
1. F 2. F 3. T 4. F 5. F
6. T 7. T 8. F 9. T 10. T
❍❍❍
UNIT
IV
5. Statistical Quality Control

169
U nit-IV
5
S tatistical Q uality C ontrol
5.1 Introduction
he object of all modern methods of manufacturing is to produce goods of absolute
T accuracy. But it is not always possible, particularly in mass production, to keep the
exact measurement. Given sufficient time, any operator could work to and maintain the
sizes within a close degree of accuracy, but there would still be small variations. A
product is of good quality, if it meets the required specifications, otherwise not. In the
age of competition, the success of a manufacturer mostly depends on quality of his
product. It is not enough to produce goods in the right quantity and at the right time; it is
important to ensure that the goods produced are of right quality. Without quality, the
other dimensions of quantity and time have little relevance.
In quality management, the quality of a product is so maintained that the reputation of

the company does not suffer. This helps the company to enter into a new product line
and various other marketing ventures. If a company wants to enter the international
market, it is necessary that the quality of his product is at par with the quality of other
countries. Quality management is thus, an important long-term marketing strategy. For
developing countries like India, quality management is an important aspect with
reference to liberalisation policies. In short, we say that our emphasis should be on
quality improvement even for domestic products, so that consumer gets better services.
We can say that it is a social aspect of quality.
170
5.2 What is Quality ?

Quality is a measure of how closely an item conforms to standards. The standards may
relate to time, materials, preformance, reliability, appearance of the product: As such, in
general, we can say that a product is considered to be of good quality if it gives the desired
services satisfactorily. In broader sense, the word 'quality' should posses the following
aspects:
1. Fitness for the purpose for which it is meant.
2. Quality characteristics i.e. its appearance, life, taste, odour, performance, reliability,
ease in maintenance etc.
3. Degree of general excellence of the product as compared to other products of the
same class.
4. Customer's preference, i.e., the degree or level to which it is preferred by the
customers over similar products from others sources.
5.3 How Quality is Measured ?

There are two methods to measure the quality of a manufactured product.
1. By numerical measurement
2. By attributal measurement
5.3.1 By Numerical Measurement

We decide about the quality of a product by making numerical measurements on it, such
as length, breadth, density, radius etc., and the items which are meeting with the
specification are accepted.
5.3.2 By Attributal Measurement

This type of measurement is applicable only when any item is divided into two classes,
defective (bad) or effective (good). In this case each item is inspected and decided
whether it is defective or effective. For example, colour, shape, cuts, spots etc.
5.4 What is Quality Control ?

It is an activity which assures that goods or, services conform to specified standards. It is
an assessment of quality standards in either or all three of the input, transformation, or
output phases of a production process. The control of quality involves, measurement,
feedback, comparison with standards and correctness when necessary. Thus, the basic
endeavours in quality control are:
171
1. To ensure that the desired specifications of the product are achieved.

2. To compare the product with standards and correct it when necessary.
5.5 Statistical Quality Control (S.Q.C)

We have explained the meaning of quality and quality control of product. Now,
statistical quality control is a method employed of quality of manufactured products by
using different statistical techniques. With the help of statistical methods, we can find
out whether the variation in the articles is of random nature, or whether some reason can
be assigned for this variation. In other words the main purpose of S.Q.C. is to device
statistical methods for separating the variation of random nature from the other type
(preventable variation), so that we may take appropriate action as quickly as possible
when preventable causes are present in the system. Let us discuss both type of variations.
5.6 Causes of Variation

We described above that the products manufactured by a certain manufacturer are not of
the same quality. Variation in their quality is bound to occur. If we analyse this variation,
we observe that there are certain causes responsible for this variation. These are classified
as:
1. Non-assignable causes (Random causes)
2. Assignable causes (Preventable causes)
5.6.1 Non-assignable Causes (Random Causes)

There may be some deviation from the desired specification of the product inspite of all
care. This type of variation is unavoidable and no reason is possible to assign for such
variation. We can say that this variation occurs in a random way. This type of variation
cannot be prevented and cannot be detected and therefore; is called allowable or
tolerable and does not affect the quality and utility of the product. For example, the
non-homogeneity in raw-material may result different surface quality or finishing in
machining. In quality control, chance variations are seldom considered and the process is
said to be under statistical control.
It has been established that if the variations are only due to chance causes, the
observations will follow "normal curve" described later on.
5.6.2 Assignable Causes (Preventable Causes)

When the causes of variation in quality are such that they can be detected, assigned to
some reasons and rectified to prevent further occurrence of such variability, they are
172
known as assignable or preventable causes. Variations due to assignable causes possess

greater magnitude as compared to those due to chance causes. This affects the quality
and utility of the product and causes worry to the manufacturer. This type of variation
occurs due to machine condition, changing working conditions, mistake on the part of
the worker, difference in skill of operations etc. Thus, these causes are detected and
eliminated and controlled. The control charts (to be discussed) tell when, and in some
cases where, to look.
5.7 Techniques of Statistical Quality Control

In statistical quality control out aim is to weed out the assignable causes of variation as
soon as they occur, so that the actual variation may be supposed to be due to chance
causes alone.
In this type of problem our aim is to control the manufacturing process so that the
proportion of defective items is not excessively large. This is known as "Process
Control". In the other type of problem, we like to ensure that lots of manufactured goods
do not contain an excessively large proportion of defective items. This is known as
'Product Control' or 'Lot Control'. Process control is achieved mainly through the
technique of control charts, whereas product control is achieved through sampling plans.
This is shown by the following chart:
–
X-Chart
Variables R-Chart
s-Chart
Process Control
p-Chart
Attributes np-Chart
Techniques
C-Chart
Variables
Product Control
Attributes
Fig.1
173
5.8 Control Charts and their Basis

One of the most important tool of production management and the control of quality in
the manufectured product is the 'Control Charts' technique. Statistical techniques for
controlling the quality of the finished product, were developed by W.A.Shewart (1924)
of Bell telephone laboratory. These charts (process control charts) are commonly used in
quality control to maintain a continuous evaluation of the manufacturing process. A
control chart is simply a frequency distribution of the observed values plotted as points
in order of occurrence so that each value has its own identity relative to the time of
observation. The points on the control charts may or may not be connected.
In control charts the quality characteristic Q (suppose) of manufactured product follows

a distribution if only random causes are present in the process. This distribution has the
mean Q and standard deviation sQ. If this distribution is normal the limits Q ± 3sQ
consists of 99.73 points of the distribution. Further, the limits Q ± 2sQ and Q ± sQ
consists of 95.44% and 68.26% points of the distribution. If the distribution of Q is
other than normal, the limits Q ± 3sQ consist of 88.89% points of the distribution. Thus
in every case (whether normal or other than normal) these limits consist practically all
the points of the distribution and if any point goes outside these limits, we expect that
not only random causes (non-assignable causes) are present in the process but the
assignable causes are also present. In such a case we say that the process is out of control.
If no point goes outside these limits we say that the process is within control. Thus
actually in SQC we do not control the quality but control the process.
Taking this idea, as stated earlier a central line Q, upper control limit Q + 3sQ and lower
control limit Q - 3sQ are drawn as horizontal straight lines on a graph paper and the
number of samples called homogeneous sub-groups taken from different shifts of the
process are plotted on the same graph paper. The sub-group number is taken on X -axis
and quality characteristic Q on Y -axis. The graph thus obtained is called Control Chart.
A typical control chart is shown below:
Y
Q + 3sQ U.C.L.
Quality characteristics, Q
Central
line
Q – 3sQ L.C.L.
O X
1 2 3 4
Sub-group number
Fig. 2
174
5.9 Types of Control Charts

Control charts can be divided under the following two heads:
1. Control charts for variables
2. Control charts for attributes
When quality is measured through numerical measurement, control charts are called
control charts for variables. In this case, measurements such as diameter of a screw,
specific resistance of a wire, density of liquid etc., are of continuous type and are regarded
to follow normal probability law. For quality control of such data, the following control
charts are used.
(i) Control chart for mean (X-chart)
(ii) Control chart for S-D (s-chart)
(iii) Control chart for range (R-chart)
On the other hand if quality characteristic is an attribute and each item is regarded as
either defective or non-defective (Good or bad), three types of control charts are used.
(i) Control charts for fraction defective (p-chart)
(ii) Control charts for number of defectives (np-chart)
(iii) Control charts for number of defects (C-chart)
Let us discuss both type of charts one by one.
5.9.1 Control Charts for Variables

As discussed earlier, these charts may be applied to any quality characteristic that is
measurable. In order to control a measurable characteristic we have to exercise control on
the measure of central tendency and the measure of dispersion. Usually X-chart is paired
with R-chart to control the mean (measure of central tendency) and range (measure of
dispersion) respectively of the characteristics.
5.9.1.1 Different Steps for Shewart Control Charts for Variables

The Shewart control chart technique consists of the following steps:
1. Collection of Data: The observations are taken at a regular intervals of time during
production and these sample observations are classified into k-subgroups having
n1, n2 ,... nk observations respectively. If possible sub-groups should be of equal size
i.e., n. A small size of the sample say 4 or 5 is preferred since there is less variation.
The minimum number of sub-groups is generally 20 or 25.
2. Calculation of Statistics: On the basis of the data collected above, the related
statistics such as X , R , s etc., are evaluated. It can be shown by the following table:
175
Sub-group No. Individual values Mean Range S.D.
1. x11 x12 K x1n x1 R1 s1
2. x21 x22 K x2 n x2 R2 s2
3. x31 x32 K x3 n x3 R3 s3
M M M M M M M
k x k1 x k 2 x kn xk Rk sk
where k sub-groups each of size n are taken.
3. Calculation of Control Limits: On the basis of the statistics upper control limit
(U.C.L.), lower control limit (L.C.L.) and the central line (C.L.) are calculated and
plotted as horizontal lines on the graph paper.
4. Conclusion: If the observed values fall within the control limits, the process is
assumed to be in control at that moment of production. If the observed values fall
outside the control limits, the process is said to be out of control at that moment
and presence of assignable causes is indicated in the process. However, points which
fall outside the control limits do not necessarily represent rejected material but only
signal that some corrective action is required to prevent manufacturing faulty parts.
5.9.1.2 Control Chart for Mean (X-Chart)

It is known that arithmetic mean is the most frequently used average. In X-chart,it is
observed that the means obtained by different samples lie within the control limits or
not. A control chart for X is frequently coupled with the control charts for either R-chart
or s-chart, because X-chart discovers the variability between rational sub-groups and
R-chart discovers the assignable causes within subgroups. Thus, the two charts reveal the
presence of irregularity both, between samples and within samples. For X-chart, the
following steps are taken:
1. Selection of samples or sub-groups: In order to make a control charts analysis

effective, it is necessary to select samples or sub groups by random procedure. It is
better that the size of each sample is equal, usually each sample should be of size 4
or 5. Further, there should be 20 or 25 samples. Normally, 25 samples of size 4 each
or 20 samples of size 5 each will give good estimate of the process average
2. Computation of averages: On the basis of the data or measurements the average

for each sub-group is calculated. That is,
n
S x1 j
x + x + ... + x1n j =1
x1 = 11 12 = = Mean of 1st sub-group
n n
M M
176
n
S x ij
x + x i2 + ... + x in j =1
x i = i1 = = Mean of i -th sub-group
n n
M M
n
S x
x k1 + x k 2 + ... + x kn j =1 kj
xk = = = Mean of the k-th sub-group
n n
n
where S x ij = sum of the sample units of i-th sample
j =1
x i = Mean of the i-th sub-group (sample), i =1, 2,... k
n = Number of units in the sub-group
Next, we find the average of the above k sub-group averages and call in x, i. e.,
k
S xi
x1 + x2 + ... + x k i = 1
x= =
k k
This x is called central line (C.L.)
3. Computation of Control Limits: It is known-that if sP is the process (universe)

standard deviation, then the standard error of sample mean is sP / n, where n is the
sample size, i. e.,
s
SE( x i) = P , (i =1, 2,... k)
n
Also m P is the mean of the process or process average. The limits are computed in
the following two cases:
Case 1: When standards are given i.e., both process mean m P and process s.d sP are
known. The 3-s control limits for X-chart are given by
s æ 3 ö
mP ±3 P or m P ± A sP çA= ÷
n è nø
If m ¢P and s¢P are known or specified values of m P and sP respectively, then
Upper Control Limit (U.C.L.) = m ¢P + A s¢P
Lower Control Limit (L.C.L) = m ¢P – A s¢P
The value of A can be obtained for different values of n.

177
Case 2: When standards are not given i. e., both m P and sP are unknown, then their
estimates based on sample information are used. From the observations we find,
Range of the i-th sub-group (R i) = x ij (max.) - x ij (min), i =1, 2,... k
further, we find mean of the ranges of all sub-groups (R ) = Mean of the k values of range
1 k
= . S Ri
k i =1
Similarly, we find (si) = S.D. of the i-th sub-group, i =1, 2,... k and consequently
1 k
Mean of the s.d.'s of all sub groups (s) = S S
k i =1 i
There are two ways of obtaining these limits. One way is to use the value of
ranges and another way is to use the values of standard deviations.
On the basis of R (range) On the basis of s (S.D.)
From the sampling distribution of range, If the control limits are to be obtained on
E (R ) = d2 sP the basis of s rather than R, then the
where d2 is a constant depending upon estimate of sP can be obtained from the
the sample size. Thus an estimate of s relation
P
can be obtained from R by the relation s = C2 sP
R s
sP = or sP =
d2 C2
Also x gives an unbiased estimate of the æ n-2ö

ç ÷!
population mean m P , since 2 çè 2 ÷ø
Where C2 = =
n æ n-3ö
çç ÷÷ !
è 2 ø
1 k
E ( x) = S E( x i) = m P
k i =1
Control limits for X-chart
On the basis of R On the basis of s

R æ 3 ö
U.C.L. = x + = x + A2 R U.C.L. = x + ç ÷s=x+ A s
d2 ç nC ÷ 1
è 2ø
R æ 3 ö
L.C.L. = x – = x - A2 R L.C.L. x - ç ÷s=x+ A s
d2 ç nC ÷ 2
è 2ø
C.L. = x C.L. = x
where A2 is again a value obtainable from where A1 is again a value obtainable from
the table for certain value of n. the table for certain value of n.
178
4. After obtaining the upper and lower control limits and all the values of
statistics,sub-groups (samples) numbers are taken on x-axis and the values of
statistics on y-axis. all the points so plotted are connected and chart is obtained.
5. If all the points (obtained by plotting means of different samples) lie within the
control limits, we say that process is within control and if any point goes outside the
control limits, we say that process is out of control and some assignable causes are
present in the system, which needs a correction.
Solved Examples
Example 1: A drilling machine bores holes with a mean diameter of 0.5230 cm and a
standard deviation of 0.0032 cm. Calculate the 2-sigma and 3-sigma control limits for
means of sample of 4.
Solution: Given m P = 0.5230 cm; sP = 0.0032 cm; n = 4
2-sigma control limits for X-chart:
s s
U.C.L = m P + 2 P L.C.L = m P - 2 P
n n
2 ´ 0.0032 2 ´ 0.0032
= 0.5230 + = 0.5230 -
4 4
= 0.5230 + 0.0032 = 0.5230 – 0.0032
= 0.5262 cm = 0.5198 cm
3-sigma control limits for X-Chart:
s s
U.C.L = m P + 3 P L.C.L. = m P - 3 P
n n
3 ´ 0.0032 3 ´ 0.0032
= 0.5230 + = 0.5230 -
4 4
0.0096 0.0096
= 0.5230 + = 0.5230 -
2 2
= 0.5230 + 0.0048 = 0.5230 – 0.0048
= 0.5278 cm = 0.5182 cm
Example 2: In the production of certain rods, a process is said to be in control if the outside
diameters have a mean 2.5" and a s.d. of 0.002" , (i) find the control limits for the means
of random samples of size 4, (ii) mean of 10 random samples taken at regular interval
were : 2.5014, 2.5022, 2.4995, 2.5076, 2.5040, 2.5001, 2.4993, 2.4962, 2.4966
and 2.4971, was the process even out of control.
179
Solution: Given : m P = 2.5" ; sP = .002" ; n = 4
1. Control limits for X-chart:

The central line (C.L.) = m P = 2.5"
3 sP 3 ´ 0.002
U.C.L. = m P + = 2 .5 +
n 4
0.006
= 2 .5 + = 2 .5 + 0.003
2
= 2.503"
3 sP 3 ´ 0.002
L.C.L = m P - = 2.5 -
n 4
0.006
= 2.5 - = 2.5 - 0.003
2
= 2.497"
2. On the basis of central line, U.C.L and L.C.L and plotting the values of 10 means
(given), it seems that process is out of control corresponding to sample number 4, 5,
8 and 9.
X-chart
2.508
2.506
2.504
2.503 UCL
2.502
2.500 CL
2.498
2.497 LCL
2.496
2.494
2.492
1 2 3 4 5 6 7 8 9 10
Sample Number
Fig. 3: Mean chart showing process out of control
5.9.1.3 Control Chart for Range (R-chart)

In order to control a measurable characteristic, we have to exercise control on measure of
central tendency as well as measure of dispersion. Usually X and R charts are used to
control mean and range respectively. The following steps are adopted:
1. Calculation of Range of Each Sample: Firstly, we find the range of each of the
sample, by the formula R = L - S, where ' L' stands for the largest observation and 'S'
180
stands for the smallest observation of the sample. Thus, if we have k sub-groups, we
will have k values of the range denoted by R1, R2 .... R i.... R k (i = 1, 2,.... k).
2. Computation of Central Line: For obtaining central line we find the average of
the above obtained ranges, i.e.,
k
S Ri
R + R2 + .... + R k i =1
R= 1 =
k k
3. 3-sigma Control Limits: 3-sigma control limits for R-chart are given by
E (R ) ± 3 sR .
Case 1: When standards are given i.e., R ¢ and sR ¢ are known as the estimates of
E (R ) and sR . The 3-sigma control limits for R-chart are given by
U.C.L. = R ¢ + 3 sR ¢ = D2 s x ¢
L.C.L. = R ¢ - 3 sR ¢ = D1s x ¢
where, D2 and D1 are constants whose values can be obtained from the table.
Case 2: When standards are not given then

E(R) is estimated by R and sR is estimated by the relation
sR = CE(R ) = CR
where C is a constant depending upon the value of n.
\ U.C.L. = R + 3 C R = R (1 + 3 C)
= D4 R
L.C.L. = R - 3 C R = R (1 - 3 C)
= D3 R
The values of D4 and D3 can be seen from the table for different values of n.
4. Drawing of Control Chart: Like X-chart, the chart for range is also drawn in the
same fashion except that the values of ranges (R) are taken on Y -axis.
5. Conclusion: If any point goes out side the control limits, we say that process is out
of control. Other wise not.
Example 3: 20 random samples of 5 units drawn from each lot of wire gave the following
mean diameters and ranges. Find out control limits for X and R-chart and comment.
Subgroups X R Subgroup X R
1 61.2 10 11 60.8 13
2 60.8 9 12 65.2 10
3 61.8 5 13 67.8 10
181
4 62.2 9 14 66.0 9
5 59.2 11 15 64.2 6
6 62.0 14 16 64.8 8
7 61.6 12 17 69.6 13
8 66.0 6 18 59.4 11
9 62.0 15 19 61.0 15
10 66.0 16 20 61.8 13
Solution: It is given that 20 samples each of size 4 are selected.
Thus, k = 20, n = 5
S X =1263.4 and SR = 215
S X 1263.4
\ X= = = 6317
.
k 20
SR 215
and R= = = 10.75
k 20
Control Limits for X-Chart:
Central line ( X ) = 6317

.
U.C.L.= X + A2 R = 6317
. + 0 . 58 ´10.75 (for n = 5, table value of A2 = 0.58)
= 6317
. + 6.235 = 69.41
L.C.L. = X - A2 R = 6317
. - 0.58 ´10.75
= 6317
. - 6.235 = 56.94
Control Limits for R - Chart:
Central line (R ) = 10.75
U.C.L. = D4 R = 211
. ´10.75 (for n = 5, table value of D4 = 211
. )
= 22.68
L.C.L = D3 R = 0 ´10.75 (for n = 5, table value of D3 = 0)
=0
182
1. From the value of X given above, it is observed that the process is out of control
corresponding to sample number 17.
2. From the values of R given above, it is observed that all the values of R lies within
the control limits and thus the production process is in a state of statistical control.
Example 4: Construct a control chart for mean and the range for the following data on the
basis of fuses, samples of 5 being taken every hour (each set of 5 has been arranged in
ascending order of magnitude). Comment on whether the production seems to be under
control, assuming that these are the first data.
42 42 19 36 42 71 60 18 15 69 64 61
65 45 24 54 51 74 60 20 30 109 90 78
75 68 80 69 57 75 72 27 39 113 93 94
78 72 81 77 59 78 95 42 62 118 109 109
87 90 81 84 78 132 138 60 84 153 112 136
Solution:
Sample No. Sample Observations Total Sample Mean Sample Range

X R
1 42 65 75 78 87 347 69.4 45
2 42 45 68 72 90 317 63.4 48
3 19 24 80 81 81 285 57.0 62
4 36 54 69 77 84 320 64.0 48
5 42 51 57 69 78 297 59.4 36
6 71 74 75 78 132 430 86.0 61
7 60 60 72 95 138 425 85.0 78
8 18 20 27 42 60 167 33.4 42
9 15 30 39 62 84 230 46.0 69
10 69 109 113 118 153 562 112.4 84
11 64 90 93 109 112 468 93.6 48
12 61 78 94 109 136 478 95.6 75
Total 865.2 696

183
From the above data, we get
1 k 865 .2
X= S Xi = = 72.1
k i =1 12
1 k 696
R= S R = = 58
k i =1 i 12
Control limits for X-chart:
U.C.L. = X + A2 R
= 72.1 + 0.58 × 58 (The value of A2 for n = 5 = 0.58)
= 105.74
L.C.L.= X - A2 R
= 72.1 – 0.58 × 58
= 38.46
C.L. = X = 721
.
Y
110 X-Chart UCL = 105.74
100
90
80
CL = 72.1
Sample Mean
70
60
50
40 LCL = 38.46
30
20
10
0 X
1 2 3 4 5 6 7 8 9 10 11 12
Sample Number
Fig. 4
From the above chart it is clear that process is out of control corresponding to the 8 th and
10 th samples.
184
5.9.1.4 R-Chart
U.C.L. = D4 R = 2.11 × 58 = 122.38 (The value of D4 for n = 5 is 2.11)
L.C.L = D3 R = 0 ´ 59.67 = 0
C.L. = 58
Y
130 R-Chart UCL = 122.38
120
110
100
90
Sample Range
80
70
60 CL = 58
50
40
30
20
10
LCL = 0
0 X
1 2 3 4 5 6 7 8 9 10 11 12
Sample Number
Fig. 5
Since all the sample points fall with in the control limits, the process is within control.
Example 5: Give control limits to control the variability in production when s is not
known. Use the following values to study the variation in the production of screws when
D3 = 0 and D4 = 2.282 for n = 4
Group No. : 1 2 3 4 5 6 7 8
Range (m.m) 3 4 2 5 3 4 5 2
1 k 28
Solution: We find R = S Ri = = 3.5
k i =1 8
Thus, control limits for R-chart
U.C.L. = D4 R = 2.282 × 3.5 = 7.987
L.C.L. = D3 R = 0 × 3.5 = 0
C.L. = R = 3.5
185
Example 6: For the following data draw X-chart and also give your comments.
Sample No. Sample observations
1 62 68 66 68 73
2 50 58 52 58 65
3 67 70 68 56 61
4 64 62 57 62 63
5 49 98 65 64 66
6 63 75 62 58 68
7 61 71 66 69 77
8 63 72 61 53 55
Given for : n = 5 A2 = 0 .58
n = 8 A3 = 0 . 373 [C.C.S., Meerut 2001]
Solution:
Sample No. Sample Observations Total Mean X Range R
1 62 68 66 68 73 337 67.4 11
2 50 58 52 58 65 283 56.6 15
3 67 70 68 56 61 322 64.4 14
4 64 62 57 62 63 308 61.6 7
5 49 98 65 64 66 342 68.4 49
6 63 75 62 58 68 326 65.2 17
7 61 71 66 69 77 344 68.8 16
8 63 72 61 53 55 304 60.8 19
Total 513.2 148
1 k 513 .2
X= S Xi = = 64 .15
k i =1 8
186
1 k 148
R= S R = = 18.5
k i =1 i 8
Control limits for X-chart
U.C.L = X + A2 R = 64.15 + 0 .58 ´18.5 = 74 .88
L.C.L = X - A2 R = 64 .15 - 0 .58 ´18.5 = 53 .42
C.L. = X = 64.15
Y
X-Chart
100
90
80 UCL = 74.88
Sample Mean, X
70 CL = 64.15
60
LCL = 53.42
50
40
30
20
10
0 X
1 2 3 4 5 6 7 8 9 10 11 12
Sample Number
Fig. 6
Since all the sample points fall within the control limits, the process is within control.
5.9.1.5 Standard Deviation Chart (s-chart)

The object of s-chart is the same as that of range chart (R-chart) because both are the
measures of dispersion. The method is more or less same as of R-chart. The following
steps are taken:
1. Calculation of Standard Deviation of Each Sample: Standard deviations

s1, s2 ,... si ... sk (i = 1, 2 ... k) are calculated for each sample.
2. Control Limits: The 3-sigma control limits are obtained for the two cases as follow:
Case I: When standards are given i.e., sP is known, then E(s) = C2 sP and
n -1
ss = - C22
n
If the standard value of sP is s p¢, then the chart will be based on
187
n -1
U.C.L. = C2 sP ¢ + 3 sP ¢ - C22 = B2 sP ¢
n
n -1
L.C.L = C2 sP ¢ - 3 sP ¢ - C22 = B2 sP ¢
n
Central Line (C.L.) = C2 sP ¢
The values of B1, B2 and C2 are obtainable from the table.
Case 2: When standards are not given, i.e., sP is estimated by
s 1 æ1 k ö
= ç S s ÷ where E(s) = C s
C2 C2 ç k i =1 i ÷ 2 P
è ø
Thus, the control limits are
s n -1
U.C.L. = s + 3 - C22 = C2 B4 s
C2 n
s n -1
L.C.L. = s - 3 - C22 = B3 s
C2 n
Central line (C.L.) = s
The values of B3 and B4 may be obtained from the table.
3. Drawing of Chart: Like X and R-chart, the chart for s.d (s-chart) is also drawn in
the same fashion except that values of s.d. (s) are taken on Y -axis.
4. Conclusion: If any sample point goes outside the control limits, we say that process
is out of control otherwise not.
Note: In either case, if L.C.L., as given by the stated formula, comes out negative, then it is to be
taken as zero. This is because in no case it can be a negative quantity.
Example 7: The random sample each of 6 items were taken from the output of a process ad
the means, ranges and s.ds of 10 subgroups were obtained as under:
Sample: 1 2 3 4 5 6 7 8 9 10
X: 26.2 22.1 23.3 24.5 22.0 19.3 20.8 25.4 23.2 21.2
R: 4 7 2 5 11 9 8 6 7 5
s: 2.6 3.2 1.9 2.4 5.1 4.3 4.4 5.0 1.0 17
Compute 3-s limits for three charts respectively. You may use the following control factors
for n = 6:
A2 B3 B4 D3 D4
0.48 0.03 1.97 0 2.004

188
1 k 228 1 k 64
Solution: We find, X= S Xi = = 22.8, R = S R i = = 6.4,
k i =1 10 k i =1 10
1 k 316
.
s= S s = = 316
.
k i =1 i 10
Control limits for X-chart:
U.C.L. = X + A2 R L.C.L = X - A2 R
= 22.8 + 0.48 × 6.4 = 22.8 – 0.48 × 6.4
= 22.8 + 3.07 = 22.8 – 3.07
= 25.87 = 19.73
Control limits for R-chart:
U.C.L = D4 R L.C.L = D3 R
= 2.004 × 6.4 = 0 × 6.4
= 12.83 =0
Control limits for s-chart:
U.C.L = B4 s L.C.L = B3 s
= 1.97 × 3.16 = 0.03 × 3.16
= 6.23 = 0.095
Conclusion
1. X -chart: The sample value corresponding to first sample is out of upper control
limit, thus process is out of control.
2. R-chart: As no point is outside the control limits, process is within control.
3. s-chart:As no point is outside the control limits, process is within control.
5.9.1.6 Limitations of the Control Charts for Variables

1. These charts are available for variables only i.e., only for measurable characteristics.
2. If the measurable characteristics are large, then it is not possible to construct the
control chart for each of the dimension of a manufactured product.
3. In certain cases these charts are impracticable and uneconomical.

189
5.9.2 Control Charts for Attributes

Keeping in view of the above limitations we have another type of charts known as control
charts for attributes. When it is not possible to measure the quality of a product
numerically, we call it good or bad. For example, the circular disc produced by a factory is
perfectly circular or not, the length of bolts produced by a factory is proper or not. If disc
is not perfectly circular, it is called defective, similarly, if length of bolt is not upto the
mark, it is called defective and so on. If a product is not defective, we say that it is
acceptable. Thus, in control charts for attributes, we see whether it is under control with
regard to fraction defectives. There are three kinds of control charts for attributes.
1. Control charts for fraction defectives (p-chart)
2. Control charts for number of defectives (d-chart or np-chart).
3. Control charts for number of defects (c-chart).
Let us discuss them one by one.
5.9.2.1 Control Chart for Fraction Defectives (p-chart)

This chart is based upon the binomial distribution in which we find the proportion of
defective units in the sample. Suppose we have k samples each of size n then, proportion
of defectives in a sample of size n is given by:
d
p=
n
where, p = sample proportion defective
d = number of defectives in a sample
n = size of the sample
The following steps are taken:
1. To find the proportion of defectives: As per the above rule, the proportion of
defectives for each sample is calculated and these are denoted by p1, p2 ,... pi ... pk
(i =1, 2,... k).
2. To find the average of proportion defectives: The average of the above values of
pi is taken, it is denoted by p and is given by:
k
p + p2 + .... + pi + .... + pk
å pi
i =1
p= 1 =
k k
p can be found by another way also in which we add all the number of defectives, i.e,
Total number of defectives d1 + d2 + ... + di + ... + dk
p= =
n´ k nk
1 k
= S P
k i =1 i
190
where di is the number of defectives in the i-th sample (i = 1, 2 .... k).
3. To find control limits : If ' d' is the number of defectives in a sample of size n then
d
the sample proportion defective is p = . Hence ' d ' is a binomial variate with
n
parameters n and P.
\ E(d) = nP and V (d) = nPQ, Q =1 - P
æ dö nP
\ E( p) = E ç ÷ = =P
è nø n
ædö 1 nPQ PQ
V ( p) = V ç ÷ = V (d) = =
è n ø n2 n2 n
Thus, 3-sigma control limits for p-chart are given by
PQ
E( p) ± 3 S. E.( p) = p ± 3 = P ± A PQ
n
where A = 3 / n can be obtained from the table.
Case I: When standards are given i.e., the value of P is given as P ¢

P ¢ Q¢ P ¢ (1 - P ¢ )
U.C.L. = P ¢ + A = P¢ + A
n n
P ¢ Q¢ P ¢ (1 - P ¢ )
L.C.L. = P ¢ - A = P¢ - A
n n
Central line (C.L.) = P ¢
Case II: When standards are not given, i.e., the value of P is unknown, it is
estimated by the statistic P with the help of the information obtained from the
sample. It may be seen that p is an unbiased estimate of P, since
æ k ö
ç S di ÷
ç i =1 ÷ knP
E( p) = E ç = =P
nk ÷ nk
ç ÷
ç ÷
è ø
Thus 3-sigma control limits for p- chart are
pq p(1 - p)
U.C.L. = p + 3 = p+ A , q =1 - p
n n
pq P (1 - p)
L.C.L. = P - A =P - A
n n
Central line (C.L.) = p
Example 8: The following are the figures of defectives in 22 lots each containing 2000
rubbers belts:
425, 430, 216, 341, 225, 322, 280, 306, 337, 305, 356, 402, 216, 264, 126, 409,
193, 326, 280, 389, 451, 420
Draw control charts for fraction defective and comment. [C.C.S., Meerut 2005]
191
Solution: Here we have a fixed sample size n = 2000 for each lot. If di and pi are respectively
the number of defectives and the sample fraction defective for the i-th lot then
di
Pi = (i =1, 2,...22)
2000
These are shown in the following table:
S.No. d p S.No. d p
1 425 0.2125 12 402 0.2010
2 430 0.2150 13 216 0.1080
3 216 0.1080 14 264 0.1320
4 341 0.1705 15 126 0.0630
5 225 0.1125 16 409 0.2045
6 322 0.1610 17 193 0.0965
7 280 0.1400 18 326 0.1630
8 306 0.1530 19 280 0.1400
9 337 0.1685 20 386 0.1945
10 305 0.1525 21 451 0.2255
11 356 0.1780 22 420 0.2100
Spi 3 .5095
1 k 3.5095
Thus, p= S p = = 0 .1595
k i =1 i 22
q = 1 - 01595
. = 0.8405
\ Central line (C.L.) = p = 01595

.
pq
U.C.L. = p + 3
n
01595
. ´ 0.8405
= 01595
. +3
2000
= 01595
. + 3 0.000067
= 0.1595 + 0.0246 = 0.1841
pq
L.C.L. = p - 3
n
192
01595
. ´ 0.8405
= 01595
. -3
2000
= 0.1595 – 3 0.000067
= 0.1595 – 0.0246
= 0.1349
Y p-chart for fixed sample size

230
220
Fraction defective in Rubber belts
210
200
190 UCL
180
170
160 CL
150
140 LCL
130
120
110
100
X
2 4 6 8 10 12 14 16 18 20 22
Sample Number
Fig. 7
From the chart it is seen that a number of points fall outside the control limits, hence the
process is out of control.
Example 9: The average number of defectives in 22 sampled lots of 2000 rubber belts each
was found to be 16%. Indicate how to construct the relevant control limits.
Solution: Number of lots i.e., k = 22
Number of rubber belts inspected per lot = 2000
Average number of defectives per lot = 16%

16
Hence, proportion of defective per lot = = 016
.
100
That is, p = 016

.
193
Control limits for p-chart:
p(1 - p) p(1 - p)
U.C.L. = p + 3 L.C.L. = p - 3
n n
. ´ 0.84
016 . ´ 0.84
016
= 0.16 + 3 = 0.16 – 3
2000 2000
01344
. 01344
.
= 0.16 + 3 = 0.16 – 3
2000 2000
= 0.16 + 3 0.0000672 = 0.16 – 3 0.0000672
= 016
. + 3 ´ 0.0082 = 016
. - 3 ´ 0.0082
= 0.16 + 0.0246 = 0.16 – 0.0246
= 0.1846 =0.1354
5.9.2.2 p-Chart for Variable Sample Size

p-chart discussed earlier was meant for fixed sample size. If the number of items
inspected (n) in each sample varies, separate control limits are to be calculated for each
sample. The central line will remain same. This type of limits are known as variable.
control limits. Let us illustrate the method.
Example 10: Draw a suitable control chart from the following and comment on your
findings
Sample No. No. Inspected No. of defectives
1 200 5
2 150 6
3 300 8
4 150 4
5 100 3
6 200 6
7 250 8
8 200 6
9 200 6
10 150 7
194
Solution : Since we have variable sample size, we can draw the control chart for fraction
defective by variable control limits. In this case we calculate 3-s limits for each sample
separately by using the formula :
pq
U.C.L. = p + 3
ni
pq
L.C.L. = p - 3
ni
k
S di
i =1 59
where p= = = 0.03
k 1900
S ni
i =1
di = number of defectives in the i-th sample

ni = number of units in the i-th sample
\ q = 1 - p = 1 - 0.03 = 0.97
\ p q = 0.03 ´ 0.97 = 0.0291
Computation of variable control limits for p-chart
n d d pq pq pq U.C.L. L.C.L.
p= 3´
n n n n
200 5 0.025 0.0001455 0.01206 0.03618 0.06618 – 0.00618 = 0
150 6 0.04 0.000194 0.01392 0.04176 0.07176 – 0.04176 = 0
300 8 0.026 0.000097 0.0984 0.02952 0.05952 – 0.02952 = 0
150 4 0.026 0.000194 0.01392 0.04176 0.07176 – 0.04176 = 0
100 3 0.03 0.000291 0.01705 0.05115 0.08115 – 0.05115 = 0
200 6 0.03 0.0001455 0.01206 0.03618 0.06618 – 0.00618 = 0
250 8 0.032 0.0001164 0.01078 0.03234 0.06234 – 0.00234 = 0
200 6 0.03 0.0001455 0.01206 0.03618 0.06618 – 0.00618 = 0
200 6 0.03 0.0001455 0.01206 0.03618 0.06618 – 0.00618 = 0
150 7 0.046 0.000194 0.01392 0.04176 0.07176 – 0.04176 = 0
1900 59
195
Y
p-chart for variable sample size
Fraction defective in different samples

0.09
0.08
0.07 UCL
0.06
0.05
0.04
0.03 CL
0.02
0.01
LCL
X
1 2 3 4 5 6 7 8 9 10
Sample Number
Fig. 8
From the above chart it is seen that no sample point goes out side the limits and thus
process is within control.
5.9.2.3 Control Chart for Number of Defectives (np-chart or d-chart)

We observed that in p-chart plotting of fraction defective was somewhat difficult. thus, if
we have samples of equal size, it is easy to plot the number of defectives. If we plot the
number of defectives on Y -axis, it is called d -chart (d = np). For obtaining the central line
and control limits, p is multiplied by n. That is,
Central line (C.L.) = np
U.C.L. = n p + 3 n p q
L.C.L. = n p - 3 n p q
Example 11: Twenty five boxes, each containing 20 electric switches were randomly
selected and inspected for the number of defectives in each box. The number of defectives
found in each box were as follows:
Box Number: 1 2 3 4 5 6 7 8 9 10 11 12 13
No. of defectives: 3 2 1 0 4 2 1 2 3 0 2 1 2
Box Number: 14 15 16 17 18 19 20 21 22 23 24 25
No. of defectives: 0 3 5 4 2 1 3 0 3 1 2 1
Set up control limits for a number of defectives.

196
k
Solution: Total number of defectives = S di = 48
i =1
k
S di
i =1 48
\ p= = = 0.096
nk 20 ´ 25
q = 1 - p = 1 - 0.096 = 0.904
np = 20 ´ 0.096 = 192
.
Control limits for number of defectives are:
Central line = np = 192
.
U.C.L. =n p+3 n pq L.C.L. =n p–3 n pq
. + 3 192
= 192 . ´ 0.904 = 192 . ´ 0.904
. – 3 192
= 1.92 + 3 173568
. = 1.92 – 3 173568
.
= 1.92 + 3 × 1.31745 = 1.92 – 3 × 1.31745
= 1.92 + 3.95 = 1.92 – 3.95
= 5.87 = –2.03
=0
Since the number of defectives can never be negative, the lower control limit is taken as
zero. The control chart for number of defectives is drawn below:
From the figure (9) it is clear that no plotted point lies outside the control limits, hence it
is concluded that process is under statistical control and there is no assignable causes
present in the system.
Example 12: A manufacturer finds that on an average 1 in 10 of the items produced by

him is defective. A few days letter he finds 20 items in a sample of 100 items defective. Is
the process out of control ? Give reasons for you answer.
Solution: Given that, p = 1 / 10 = 01

.
\ q = 1 - p = 1 - 01
. = 0.9
Central line = np = 100 ´ 01

. = 10
197
Control limits for np-chart:
U.C.L. =n p+3 n pq L.C.L. = n p -3 n p q
= 10 + 3 10 ´ 0.9 = 10 – 3 10 ´ 0.9
= 10 + 3 9 = 10 – 3 9
= 10 + 9 = 10 – 9
= 19 =1
Here upper control limit is 19 while the number of defective in a sample is 20, which is
greater than 19. Thus process is out of control.
CL
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
UCL
LCL
Sample Number
np-chart
Fig. 9
8
7
6
5
4
3
2
1
0
3
1
4
5
6
No. of Defectives
198
5.9.2.4 Control Chart for Number of Defects Per Unit (c-chart)

When our aim is to control the number of defects in each unit, c-chart is used. In this
chart each unit is observed for the number of defects it contains. Before, discussing
c-chart it is better to understand between defective and defects. A defective is an item that
fails to fulfil one or more of the given specifications. A defect is any instance of the items lack of
conformity to specifications. Every defective item thus contains one or more defects. The
defect may be a spot on the surface of a item or bubbles in material, cut on a piece of cloth
etc. Sample size for c-chart may be a single unit like washing machine, calculator or group
of units. However, defined sample size should be constant in the sense that different
samples have essentially equal opportunity for the occurrence of defects.
In a manufacturing process, it may be possible that the opportunities for defects to occur
are numerous, while the probability of occurrence of a defect is very low (i.e., occurrence
of defect is a rare event). In other words, we say that probability, p of occurrence of such
defect is very small such that np is finite. In such a situation, it is known that such pattern
of variation in data follows Poisson distribution and consequently 3-sigma limits are
obtained on the basis of Poisson distribution. It is known that mean and variance of a
Poisson distribution are equal. Here, we assume that c (number of defects) is a Poisson
variate with parameter, l, we get
E(c) = l and V (c) = l
The 3-sigma limits for c-chart are given by
U.CL. = E(c) + 3 V (c)
L.C.L = E(c) - 3 V(c)
= l + 3 l ; = l -3 l
and Central line = l
Case I: Standards are given i.e., the value of l is given as l¢, the 3-sigma limits are :
U.C.L = l ¢ + 3 Ö l ¢
L.C.L. = l ¢–3 Ö l ¢
Central line = l¢
Case 2: Standards are not given. In such case it is estimated by the help of sample
observations i.e., by the mean number of defects per unit. if ci is the number of defects
observed in i-th unit, then an estimate of l is given by c as it is an unbiased estimate of l
i.e., E(c) = l.
199
The value of c is calculated by the following formula:
Total number of defects c1 + c2 + .... + ci + .... + ck

c= = i =1, 2,... k
No. of units observed k
Thus, the 3-sigma are given by:
U.C.L. = c + 3 c ; L.C.L. = c - 3 c
Central line = c
Note: If L.C.L. comes out to be negative, it is considered as zero as the numbre of defects cannot be
negative.
Example 13: The number of defects in 24 rolls of cloth each of 100 metres length is given
below:
3, 5, 8, 9, 4, 2, 5, 9, 6, 4, 8, 12, 7, 5, 10, 10, 7, 5, 4, 10, 9, 5, 8, 13
Draw a suitable control chart and give your comments. [C.C.S, Meerut, 2004, 07 (Back)]
Solution: If ci denotes the number of defects in the i -th group, then we have the average
number of defects, c, given by
3 + 5 + 8 + 9 + .... + 9 + 5 + 8 + 13 168
c= = =7
24 24
Thus, central line (c) = 7
3-sigma control limits for C-chart
U.C.L. = c + 3 c L.C.L. = c – 3 c
= 7 + 3Ö 7 = 7 – 3Ö 7
= 7 + 3 × 2.645 = 7 – 3 × 2.645
= 7 + 7.937 = – 0.937
= 14.937 =0
200
Y c-chart
UCL
15
14
13
12
11
Number of defects
10
9
8
7 CL
6
5
4
3
2
1
LCL
0 X
1 3 5 7 9 11 13 15 17 19 21 23 25
Sample Numbers
Fig. 10
Since all the points are with in the control limits, the process is in a state of statistical
control.
Example 14: The number of defects in 20 items each of 100 meters length is given below:
1, 3, 3, 1, 6, 4, 3, 7, 10, 2, 2, 6, 4, 3, 2, 7, 1, 5, 6, 4
Draw the appropriate chart and give comment on your result.
Solution: (Left to the reader as an exercise)
Application of c-chart; As discussed earlier, it is apparent that c-chart is applicable only

when the probability of occurrence of defects, at any stage, is small. There are a number
of situations in industry when c-chart is needed. The following are the applications of
c-chart:
1. c is the number of imperfections observed in a bale of cloth or a number of defects

per role of 20 mt (say) length.
2. c is the number of defects of all types observed in aircraft assembly or missing rivets
in an aircraft.
3. c is the number of surface defects observed in (a) galvanised sheet of metal (b)
castiron billets (c) a sheet of paper (d) a photographic film etc.
4. c-chart can also be used for non- manufacturing units, e.g., for aircraft accidents at a
particular place in a fixed direction, in chemical laborations, where c is the number
201
of surface, defects on a sheet. It can be applied to an epidemic disease in a particular

region in a fixed span of time etc.
5.10 Control Charts for Variables Versus Control Charts

for Attributes
Variable charts Attribute chart
1. Separate charts are constructed for 1. A single chart is considered sufficient to

each quality characteristics. serve the purpose.
2. Usually a sample of size 4 or 5 2. Usually a large sample size work for

serves the purpose. correct conclusions.
3. The data is collected purposively 3. Data collected for other purposes may
for the use of variable charts. also be used.
4. Computational procedure is 4. Computational procedure is easy and

lengthy and time consuming. simple.
5. Cost of collecting information on 5. Cost of collecting information on units

units is large. is very small.
6. In general, sub-group size is taken 6. The need for equal size of samples is not
to be equal. much required as in case of variables.
7. As the control charts are sensitive 7. These charts are less sensitive to
to assignable causes, these charts assignable causes, these charts are not
provides better quality control. much effective in quality control.
5.11 Advantages of Statistical Quality Control

We know that by S.Q.C., the quality of a certain product is improved by bringing the
process under control. The following are some advantages of S.Q.C.
1. S.Q.C. keeps continuous vigilance on the quality of the product. As soon as the
presence of assignable causes is felt, the concerned engineer is informed and the
related part is corrected. In this way, the quality is improved and probability of the
company increases.
2. By correcting (repairing) the plant at an appropriate time, a considerable amount of
cost per unit is reduced.
3. The number of defective items become less and amount of scrap reduces.
4. A considerable amount of excess raw-material is avoided.
5. It provides protection against losses to the producer as well as to consumer.
202
6. The inspection of each and every unit is not practicable and reduce the efficiency of
the quality inspectors S.Q.C. keeps the workers alert.
7. S.Q.C. is economical in the sense that only a fraction of items produced is inspected.
8. It helps in quality awareness in employees
9. Better consumers relations are formed though general improvement in product and
higher share of the market.
5.12 Limitation of Statistical Quality Control

A number of advantages are discussed above and the usefulness of S.Q.C is focused. But
S.Q.C is not only the single technique for controlling all type of evils in production.
Apart from S.Q.C. many more considerations are made in the production of units. We
can say that S.Q.C. is one of the many considerations which are made in the production
system. That is S.Q.C. is a part of the production process rather than the production
being a part of S.Q.C.
5.13 Sampling Inspection (Acceptance Sampling)

From the control charts given by Shewart we judged the status of a process. if some
assignable causes are detected in the process, they are immediately removed and process
is brought into the state of statistical control. If a process is under control, it is not
necessary that all the items produced are good and there is no need of inspection. to keep
a close vigilance over the production process, inspection is essential. However, if process
is under statistical control, 100% inspection is not necessary. To inspect samples from a
lot of articles is called sampling inspection or Acceptance Sampling. Acceptance
sampling is based upon the decision to accept a mass of items (called lot) as conforming
to given standard of quality, or to reject the mass (lot) as not conforming. Thus decision
is taken through sampling as 100% inspection has the serious disadvantage of greater
fatigue. Further if testing is destructive it is absurd to use 100% inspection. For example,
in case of crackers, bulbs, glass containers, destruction of 100% items is not desirable.
Further, sampling reduces the number of inspectors on the job and saves money as well as
time. However, in certain case, 100% inspection is necessary. For example, while testing
rifles to be used by soldiers, no risk is taken. 100% inspection is essential here. Excepting
the special cases of 100% inspection it is not practable to inspect the entire lot. One has
to take resource of sampling inspection. Sampling inspection may be of two types:
1. Sampling inspection by attributes and
2. Sampling inspection by variables. For simplicity, we shall discuss here with
sampling inspection for attributes; in which items are judged as good or bad,
effective or defective and quality of the lot is adjudged from the sample fraction
defective the sampling plans. Before discussing the different sampling plans let us
discuss some basic concepts.
203
5.14 Producer's Risk

In sampling inspection it may possible that a lot of good quality is rejected. By rejection
of a good lot, the producer will be in loss. Suppose the producer claims that he has
standardised the quality at a level of fraction defective p which is called producer's
process average. The probability of rejecting a lot under the sampling plan when the
fraction defective is actually p is called the producer's risk and is denoted by Pp. That is,
Pp = Probability of rejecting a lot of quality p.
It is some times denoted by a also.
5.15 Consumer's Risk

On the same lines in sampling inspection it may be possible that a lot of bad
(unsatisfactory) quality is accepted. By acceptance of a bad lot, the consumer will be at
loss. if pt be the lot tolerance fraction defective i.e., maximum fraction defective in the lot
that he will tolerate, then the probability of accepting a lot with fraction defective Pt ,
under the sampling inspection plan is called the consumer's risk and is denoted by Pc ,
That is, Pc = Probability of accepting a lot of quality Pt .
It is some times denoted by b also.
5.16 Acceptance Quality Level (A.Q.L.)

The producer wishes to maintain the standard of quality at a specified level agreed to by
the consumer. That is a quality level that is acceptable. This level is commonly denoted
by p1. It is the fraction defective that can be tolerated without serious effects upon further
processing operations or customer reaction. Nearly all such lots should be accepted (only
perfect inspection would assure acceptance of all of them). The A.Q.L. must be
determined by practical considerations.
5.17 Rejectable Quality Level (R.Q.L.) or Lot Tolerance

Proportion Defective (L.T.P.D.)
It is the lot quality which is considered to be bad by the customer. This is a maximum
percentage or proportion of defectives that a consumer can tolerate. Lots of the quality
level are denoted by pt . The lot tolerance must also be determined by practical
considerations.
204
5.18 Average Outgoing Quality Limit (A.O.Q.L)

It gives a idea of the possible limit of the average quality in the accepted lots over a long
period. The expected fraction defective remaining in the lot after the application of
sampling inspection is known as average outgoing quality. It is a function of p, i.e., lot
quality before inspection or incoming quality (quality maintained by the producer). The
maximum value of the average outgoing quality, the maximum being taken with respect
to p, is known as the average outgoing quality limit or AOQL. This is useful in
planning for stores or long period purchasing. This type of quality assurance should be
decided on economic and production considerations.
5.19 Average Sample Number (ASN)

It is the expected value of the sample size required for coming to a decision (acceptance
or rejection of a lot) under a sampling inspection plan. It is always a function of incoming
quality i.e., p (the actual fraction defective of the lot) defined above. If we plot different
values of ASN corresponding to different value of p, we get ASN curve. It is obvious that
lower the ASN curve, the better is the sampling inspection plan (if other factors remain
unchanged).
5.20 Operating Characteristic (OC)

It is the mathematical expression L( p) which states the probability of accepting a lot as a
function of p (fraction defective of the lot). By plotting the value of L( p) against different
values of p, a curve is obtained which is called OC curve. It is clear that, the steeper the
OC curve, the greater is the protection to the customer.
5.21 Single Sampling Plan

If a decision to accept or to reject a plan is taken on the basis of single sample, it is called as
a single sampling plan. A single sampling plan for attributes may be described as follows:
1. Inspect a random sample of size n.
2. Find the number of defectives ' d' in the sample.
3. If x £ C, accept the lot, replacing all defectives found in the sample by effectives
(non-defectives)
4. If x > C, inspect the entire lot (size N) replacing all defectives by effectives.
205
In the plan discussed above consists of 4 numbers, N , n, d and C. In these 4 numbers N

and d are known numbers and n and C are to be determined. For finding the values of n
and C we apply average quality protection approach.
If p be the actual fraction defective in the lot of size N, the number of defective is N p and
the number of effective is N - N p = N (1 - p). The probability of getting exactly x
defectives in a sample of size n from this lot is obtained by hypergeometric distribution:
æ Np ö æ N - Np ö
ç x ÷ç n- x ÷
g( x, p) = è øè ø
æN ö
çç n ÷÷
è ø
Probability of accepting a lot of quality p is given by
æ Np ö æ N - Np ö
C C ç x ÷ç n- x ÷
Pa ( p) = S g( x, p) = S è øè ø ...(1)
x =0 x =0 æN ö
çç n ÷÷
è ø
The average outgoing quality under the single sampling scheme is given by
C æ N - xöæ N - N öæ N ö
S ç pN ÷ ç n - x p ÷ ç x p ÷
x =0 è øè øè ø
AOQ = ...(2)
æN ö
çç n ÷÷
è ø
since the fraction defective in the lot after inspection is (N p - x) / N , where x is the
number of defectives found in the sample, provided x does not exceed C and it is zero if it
exceeds C.
The maximum value of the expression at (2) with respect to p is the average outgoing
quality limit (AOQL).
The OC curve for the incoming quality 'p' is given by:
1. When p < 010

. , a good approximation to (1) is given by the first (C + 1) terms of the
Np
é nù
binomial expansion ê æç1- ö÷ + ú
n
ëè N ø N û
C é N pù é n ù x
n ö Np– x
i.e. L( p) ~
- åêxê ú æ
ê ú ç1 – ÷
úû êë N úû è N ø
x =0 ë
2. When p < 010

. and also n / N < 010,
. a good approximation to (1) is given by the
Poisson distribution viz.,
C ( np) x
- np
L( p) ~ å e x!
x =0
206
5.22 Double Sampling Plan

On the basis of single sample, three decisions namely accepting the lot, rejecting the lot
and taking another sample are taken. If the decision for another sample is taken it is
called double sampling plan. Under this sampling plan following numbers are taken;
N = Lot size
n1 = size of first sample
C1= acceptance number of first sample
n2 = size of the second sample
C2 = acceptance number for both samples combined
d1 = number of defectives in the first sample
d2 = number of defectives in the second sample.
Procedure
1. Take a sample of size n1 from the lot of size N.
2. Sample is inspected and classified as defective or effective.
3. If d1 £ C1, accept the lot, replace the defectives found in the sample by effectives.
4. If d1 > C2 , reject the whole lot, 100% inspection is done and all defectives are
replaced by effectives.
5. If C1 + 1 < d1 £ C2 take a second sample of size n2 from the remaining lot.
6. If d1 + d2 £ C2 , accept the lot replacing defectives by effectives.

7. If d1 + d2 > C2 , reject the whole lot. 100% inspection is done and all defectives are
replaced by effectives.
5.23 Single Sampling vs Double Sampling Plans

1. Single sampling plans are simple, easy to design and administer.
2. It provides maximum information about the lot.
3. In single sampling plan a lot is rejected on the basis of one sample only, which seems
to be unfair.
4. In double sampling plan, lots of a marginal quality always gets a second chance (2 nd
sample) of being accepted.
5. Double sampling plan often requires 25% – 33% less inspection on the average.
6. The training of inspectors to use sampling plans is easiest and somewhat difficult in
double sampling plan.
207
E XERCISE
1. What do you understand by Statistical Quality Control ? Explain C-chart in detail
with advantages. [C.C.S. Meerut 2002]
2. What do you understand by statistical quality control ? Discuss briefly the need
and utility of control charts in industry.
3. What are the main control charts for attributes ? Obtain their control limits.
[C.C.S. Meerut 2006, 06 (Back)]
4. Distinguish between:
(i) Process and Product control. [C.C.S. Meerut 2004]
(ii) Chance causes and assignable causes. [C.C.S. Meerut 2005, 08]
5. Explain the uses of control charts in industry, stating the conditions under which
different types of control charts should be used.
6. Write short notes on the following:
(i) 3-s limits (ii) AQL and LTPD
7. Distinguish between defect and defective. How would you calculate control limits
for p-chart ? Give application of such chart, in industry. Given the process fraction
defective p = 0.03 and n = 25. Calculate the control limits for p -chart.
8. Write short notes on:
(i) X-chart (ii) Producer's and Consumer's risk
[C.C.S. Meerut 2006 Old, 08]
(iii) OC curve (iv) ASN curve
(v) Sampling plan
9. Write short notes as:
(i) AOQ (ii) AOQL (iii) Sampling plan (iv) Application of C-chart
10. What are main control charts for variables ? Obtain their control limits.
11. Describe the single sampling inspection plan. Give a general out-line of the methods
for determining the constants involved in this sampling plan. [C.C.S. Meerut 2002]
12. Describe single and double sampling inspection plan. Give a general outline of methods
for determining the constants involved in these sampling plans. [C.C.S. Meerut 2001]
13. Make out difference between producer's risk and consumer's risk. Describe a single
sampling plan and give expression for the consumer's risk of this plan.
14. Explain the construction of the control charts for variables. Give 3-s lower and
upper limits for X-chart, where the sample size is n. The following number of defects
were found on 20 items during an inspection schedule. Calculate control limits,
draw a control chart and comment on your result:
3, 8, 5, 8, 6, 3, 4, 7, 8, 10, 2, 8, 6, 4, 5, 3, 0, 2, 6, 6
15. (a) Discuss the advantages of double sampling plan over a single sampling plan.
(b) Distinguish between AQL and AOQL. [C.C.S. Meerut 2002]
208
16. What do you mean by acceptance sampling procedure ? State its uses and explain
(i) Producer's risk (ii) Consumer's risk [C.C.S. Meerut 2008]
17. Discuss the uses of single sampling plan in acceptance sampling. How will you
obtain O.C. curve of such plan.
18. Discuss the advantages and disadvantages of control charts for attributes. Explain
in detail p-chart and c-chart. [C.C.S. Meerut 2012]
19. What do you understand by statistical quality control ? How is it obtained ?

Explain in detail X and R chart. [C.C.S. Meerut 2012]
20. A machine is set to deliver packets of a given weight. 10 samples of size 5 were
recorded. Below are given the relevant data:
Sample No: 1 2 3 4 5 6 7 8 9 10
Mean X : 15 17 15 18 17 14 18 15 17 16
Range (R) 7 7 4 9 8 7 12 4 11 5
Calculate the values for central line and control limits for the mean chart and range
chart and comment on the state of control. (Conversion factors for n = 5 are
A2 = 0.58, D3 = 0, D4 = 211
. ) [C.C.S. Meerut, 04 (Back)]
21. The following data show the values of sample mean (X) and range (R) for 10
samples of size 6 each. Calculate the values for central line and control limits for
mean and range chart. Draw control charts and comment on the state of control.
Sample No.: 1 2 3 4 5 6 7 8 9 10
Mean (X): 43 49 37 44 45 37 51 46 43 47
Range (R) 5 6 5 7 7 4 8 6 4 6
(Conversion factors for n = 6 are A2 = 0.483, D3 = 0, D4 = 2.004)

[C.C.S. Meerut, 04 (Back)]
22. A company manufactures a product which is packed in cans. It utilitizes an automatic

filling equipment. It takes a sample of 5 cans every two hours and measures the filling
in each of the 5 cans. The following table gives the measurements of filling (gms) in
the last 5 samples. Set up control chart and comment.
Assume : A2 = 0.58, D3 = 0, D4 = 2115

.
Sample Individual Measurements

Number 1 2 3 4 5
1 1001 1002 1000 998 999
2 999 998 1000 998 999
3 995 1002 1003 1001 1002
4 1000 1001 999 999 1002
5 994 996 996 1000 999
209
23. The following data gives the defective items in a sample of size 50 presented for
inspection. Construct p-chart and comment.
Sample No.: 1 2 3 4 5 6 7 8 9 10
Defectives found: 2 1 2 0 2 3 4 2 0 3
Sample No.: 11 12 13 14 15 16 17 18 19 20
Sample No.: 21 22 23 24 25 26 27 28 29 30
Sample No.: 31 32 33 34 35 36 37 38 39 40
24. A company produces fuses for automobile electric systems. Five hundred of the
fuses found per day for 30 days. The following table gives the number of defective
fuses found per day for 30 days. Determine the central line and 3 s-control limits
for p-chart. Does the process appear to be in control ?
Days: 1 2 3 4 5 6 7 8 9 10
No. of Defectives: 3 3 3 3 1 1 1 1 6 1
Days: 11 12 13 14 15 16 17 18 19 20
No. of Defectives: 1 1 5 4 6 3 6 2 7 3
Days: 21 22 23 24 25 26 27 28 29 30
No. of Defectives: 2 3 6 1 2 3 1 4 4 5
25. The following are the number of defects found in 1000 items of some industry
goods inspected every day in a month. Draw c-chart.
Days: 1 2 3 4 5 6 7 8 9 10
No. of Defectives: 1 1 3 7 8 1 2 6 1 1
Days: 11 12 13 14 15 16 17 18 19 20
No. of Defectives: 10 5 0 19 16 20 1 6 12 4
Days: 21 22 23 24 25 26 27 28 29 30
No. of Defectives: 5 1 8 7 9 2 3 14 6 8
26. The following table shows the number of missing rivets observed at the time of the
inspection of 12 aircrafts. Find the control limits for the number of defects chart
and comment on the state of control.
Aircraft No.: 1 2 3 4 5 6 7 8 9 10 11 12
No. of missing rivets: 7 15 13 18 10 14 13 10 20 11 22 15

210

1. Name the two types of causes of variation in the quality of manufactured products.
2. Write down the limits for X and R-chart
3. Differentiate between defect and defective.
4. Define quality, quality control and statistical quality control.
5. Explain the basis of control chart given by Shewart.
6. Differentiate between numerical and attributal measurement.
7. What happens when lower control limit comes out to be negative in p-chart or
c-chart ?
8. Define product control or lot control. [C.C.S. Meerut 2006 (Old)]
9. What is average outgoing quality ?
10. What happens when sample size of each sample is not constant ?
11. What do you mean by acceptance sampling ? [C.C.S. Meerut 2008]
12. Find control limits from the following data of 18 samples each of size 4 : X = 911, R
= 17.4
13. Name the distribution at which p-chart is based.
14. Name the distribution at which c-chart is based.
15. Explain the meaning of double sampling plan. [C.C.S. Meerut 2005]
16. Write down the formulae for the limits of c-chart. [C.C.S. Meerut 2005]
17. Who developed control charts ? [C.C.S. Meerut 2005, 06 (Old)]
18. Write the name of any two control charts. [C.C.S. Meerut 2005]
19. Discuss the causes of quality variation. [C.C.S. Meerut 2006 (Old)]
20. Explain the meaning of single sampling plan. [C.C.S. Meerut 2006 old]
21. When a production process is said to be under statistical control ?
[C.C.S. Meerut 2006 (Old)]
22. What is meant by random variation ? [C.C.S. Meerut 2006]
23. Define two types of causes of variation in the quality of manufactured goods.
24. What do you mean by statistical quality control ? How is it obtained ?
25. Explain in detail p and c chart [C.C.S. Meerut 2008]
26. Differentiate between R and S chart [C.C.S. Meerut 2008]
211

1. The causes of variation in the quality of a manufactured product is due to
(a) Assignable causes (b) Chance causes
(c) (a) and (b) both (d) None of the above [B.R.A. Agra 2004]
2. The upper control limit of x and R chart is when the standards are not given:
sx
(a) X = As x (b) x = 3
n
(c) x + A2 R (d) None of these [B.R.A. Agra 2004]
3. Constants of mean charts are:
(a) A2 (b) B2
(c) D3 (d) D4 [B.R.A. Agra 2006,08]
4. When L.C.L of c-chart is zero, then c is :
(a) 9 (b) 10
(c) 11 (d) 12 [B.R.A. Agra 2006]
5. If one sample point falls out side the control limits then it means:
(a) Process is under control (b) Process is out of control
(c) Neither (a) nor (b) (d) None of the above [B.R.A. Agra 2005, 08]
6. The L.C.L. of control charts for fraction defective when standards are not given:
(a) p + A p (1 - p) (b) p
(c) p - A p (1 - p) (d) None of the above [B.R.A. Agra 2005]
7. Quality control is used in :
(a) Agriculture (b) Industries
(c) Education (d) None of these [B.R.A. Agra 2009]
8. The upper control limit of c-chart is given by :
(a) c - 3 c (b) c + 3 c
(c) c + 3 c (d) c - 3 c [B.R.A. Agra 2009]

212
9. X -chart indicates variation :
(a) Between samples (b) Within samples
(c) Both (a) and (b) (d) None of these [B.R.A. Agra 2012]
10. The U.C.L. of R-chart when standards are not given is:
(a) D4 R (b) D3 R
(c) R (d) None of these [B.R.A. Agra 2011]
11. In the construction of which chart, poisson distribution is used ?
(a) X-chart (b) R-chart
(c) p-chart (d) c-chart [B.R.A. Agra 2010]
12. The U.C.L. for X-chart is :
(a) X + A2 R (b) X + A1 R
(c) x = D2 R (d) x + D1 R
[B.R.A. Agra 2010]
13. Which of the following is not a control chart for variables ?
(a) X-chart (b) R-chart
(c) s-chart (d) p-chart
14. Which chart is used to control the number of defects in each unit of product ?
(a) p-chart (b) c-chart
(c) np-chart (d) None of these
15. For constructing a p-chart in case of observed data, if n = 9, p = 0.36, then U.C.L. is :
(a) 0.12 (b) 0.48
(c) 0.84 (d) 0.60
16. In a c-chart if average number of defects c is 9, then L.C.L is :
(a) 9 (b) 27
(c) 0 (d) None of these
17. Variation in the items produced in a factory may be due to :
(a) Non-assignable causes (b) Assignable causes

(c) (a) and (b) both (d) Neither (a) nor (b)
213
18. Non-assignable causes are those :
(a) Which can not be detected (b) Which can not be controlled
(c) Neither (a) nor (b) (d) (a) and (b) both
19. Chance variation in respect of quality control of a product is :
(a) Uncontrollable
(b) Tolerable
(c) Not effecting the quality of a product.
20. Variation in the quality of a product due to assignable causes arises due to :
(a) Defective process (b) Inconsistent (poor) raw material
(c) Carelessness of operator (d) All of the above
21. The basis of control chart is :
(a) The distribution of quality characteristics follow nearly normal distribution
(b) The distribution of quality characteristics follow Poisson distribution
(c) Neither (a) nor (b)
(d) The distribution of quality characteristic follow binomial distribution
22. The process control is achieved through :
(a) Control charts (b) Acceptance sampling plans
(c) Sequential sampling plan (d) None of these
23. Product control is done by :
(a) Control charts (b) Acceptance sampling plans
(c) Both (a) and (b) (d) None of these
24. The probability of accepting a lot of quality p is known as :
(a) ASN curve (b) OC-curve
(c) L.T.P.D (d) AOQL
25. The probability of rejecting a lot of good quality is known as :
(a) Producer's risk (b) Consumer's risk

214
(c) Average outgoing quality (d) Average sample number
26. The probability of accepting a lot of bad quality is known as :
(a) Consumer's risk (b) Producer's risk
(c) AOQL (d) ASN
27. Statistical quality control takes care of the yariation due to
(a) Assignable causes (b) Random causes
28. Quality of a manufactured product is measured by:
(a) Numerical measurement (b) Attributable measurement
29. x and R-Charts are used to find out
(a) Production control (b) Cost control
(c) Process control (d) Material control
30. The expected sample size required to arrive at a decision about the lot is called:
(a) A random variable (b) Average sample number
(c) OC function (d) None of these
31. X Chart discovers assignable causes
(a) Within samples (b) Between samples
32. Which of the following is a control chart for attributes ?
(a) X-Chart (b) R-Chart
(c) s-Chart (d) p -Chart
33. In case of a s-chart the U.C.L. is given by:
(a) B4 . s (b) B3 . s
(c) s + B4 (d) s + B3
215
34. Which of the following is not a control chart for variables?
(a) X -Chart (b) R- Chart
(c) s - Chart (d) p- Chart
35. The 3-s control limits for a p-chart are given by:
(a) P ± A PQ (b) P ± A Q
(c) P ± 3 A PQ (d) P ± 3 / A PQ
36. p-Chart is a control chart for
(a) Number of defects per unit (b) Fraction defectives
(c) Number of defectives (d) all of the above
37. The relation between expected value of R and S.D. s with usual constant factors is:
(a) E(R ) = d1s (b) E(R ) = d2 s
(c) E(R ) = D1s (d) E(R ) = D2 s
38. The upper control limits of x and R charts when standards are not given are
(a) x + As1 (b) x + 3 As1
(c) x + A2 R (d None of the above
39. A typical control chart consists of :
(a) Two horizontal lines (b) Three horizontal lines
(c) Five horizontal lines (d) None of the above
40. If the lower control limit is found negative in case of control charts for attributed, it
is taken as:
(a) 0 (b) 1
(c) 1,3 (d) 2
41. R-Charts are preferable over s-charts because:
(a) R and s - fluctuate together in case of small samples
(b) R-charts are easily calculable
(c) R-charts are economical

216
42. The relation between expected value of R with S.D. s with usual constant factors is:
(a) E(R ) = d1s (b) E(R ) = d2 s
(c) E(R ) = D1s (d) E(R ) = D2 s
43. The probability of accepting a lot with fraction defective Pl is called:
(a) Consumer's risk (b) Producer's risk
(c) Type - I error (d) None of the above
44. Variation in te quality of manufactured product is tolerable due to
(a) Chance causes (b) Assignable causes
45. Control chart is developed by
(a) C.R. Rao (b) W.A. Shewart
(c) J.Bernoulli (d) R.A. Fisher
46. Control limits for C-chart are:
(a) c ± 3 c (b) c + 3 c
(c) c ± 3 1 / c (d) c ± 3 1 / c
47. In the construction of which chart Poisson distribution is used?
(a) x - chart (b) R-chart
(c) P-chart (d) c-chart
48. In a C-chart if average number of defects C is 9, then lower control limit is:
(a) 9 (b) 18
(c) 0 (d) 12
49. For constructing a p-chart in case of observed data, if n=9, p = 0.36 then upper
control limit limit is:
(a) 0.12 (b) 0.36
(c) 0.48 (d) 0.84

217
50. When the value of population process range R ¢ is not known, then with usual
notation, the upper control limit for x-chart is:
(a) x + A1R (b) x + A2 R
(c) x + D2 R (d) x + D1 R

1. The process control is achieved through .................... .
2. The product control is achieved through .................... .
3. The idea of S.Q.C was given by .................... .
4. X and R charts are charts for .................... .
5. C chart is meant for number of defects per ................... .
6. .................... distribution is used in c-chart.
7. .................... causes can not be detected.
8. Variation in quality of a product is due to .................... and .................... causes.
9. Carelessness of the operator comes under .................... causes.
10. The probability of accepting a lot of bad quality is ................... .
True or False
1. Statistical quality control takes care of variation due to non-assignable causes.
2. Variation in quality of a certain product may be due to assignable causes and
non-assignable causes.
3. c-chart is meant for number of defects per unit.
4. In p-chart Poisson distribution is used.
5. If any sample point is out side the control limits, the process is said to be within
control.
6. Shewart gave the idea of control charts.
7. Binomial distribution is the basis of c-chart.
8. The quality variation due to uneven raw material are called chance variations.
9. Variation in product quality is inevitable.
10. X + A R are the control limits for R-chart.
2
218
A NSWERS

1. (c) 2. (c) 3. (a) 4. (a) 5. (b)
6. (c) 7. (b) 8. (b) 9. (b) 10. (a)
11. (d) 12. (a) 13. (d) 14. (b) 15. (c)
16. (c) 17. (c) 18. (d) 19. (d) 20. (d)
21. (a) 22. (a) 23. (b) 24. (b) 25. (a)
26. (a) 27. (a) 28. (c) 29. (c) 30. (b)
31. (b) 32. (d) 33. (a) 34. (d) 35. (a)
36. (b) 37. (b) 38. (c) 39. (b) 40. (a)
41. (d) 42. (d) 43. (a) 44. (a) 45. (b)
46. (a) 47. (d) 48. (c) 49. (d) 50. (b)

1. control chart 2. acceptance sampling plan 3. Shewart
4. variables 5. unit 6. Poisson 7. Non-assignable
8. assignable, non-assignable 9. assignable 10. producer's risk
True or False
1. F 2. T 3. T 4. F 5. F
6. T 7. F 8. F 9. T 10. F
❍❍❍

Applied Statistics

Uploaded by

Copyright:

Available Formats

You might also like

Applied Statistics

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Applied Statistics

Uploaded by

Copyright:

Available Formats

Applied

Dr. Arun Kumar

Dr. Alka Chaudhary Dr. Qaim Akbar Rizvi

KRISHNA Prakashan Media (P) Ltd.

The key features are:

To improve the scope of the book Objective type questions

Special thanks to Mr. Satyendra Kumar Rastogi

B.A/B.Sc IIIrd Year Paper-II (w.e.f. 2013-2014)

* Sequence of contents is as per U.P. Unified Syllabus

Chapter 1: Time Series.........................................................................................(01-48)

Chapter 2: Index Numbers..................................................................................(49-98)

Chapter 3: Statistical Methods for Psychology & Education

Chapter 4: Demographic Methods.................................................................(115-166)

Chapter 5: Statistical Quality Control........................................................(167-220)

Chapter 1: Time Series ................................................................................... (01-48)

Chapter 2 : Index Numbers.............................................................................(49-98)

2.5.3 Value Index Numbers 53

2.5.4 Special Purpose Index Numbers 53

2.6 Methods of Constructing Index Numbers 55

2.6.2 Weighted Index Numbers 61

2.7 Test for Index Numbers 65

2.7.2 Factor Reversal Test 66

2.7.3 Circular Test 67

2.8 Quantity Index Number 71

2.12.2 From Chain Base to Fixed Base 74

2.13 Cost of Living Index Numbers 76

2.14.2 Method of Weighted Relatives or Method of Family Budget 77

2.15 Limitation of Index Numbers 84

Chapter 3 : Statistical Methods for Psychology

Chapter 4 : Demographic Methods........................................................... (115-166)

Ø Solved Examples 124

Chapter 5 : Statistical Quality Control........................................................(167-220)

5.4 What is Quality Control ? 170

5.7 Techniques of Statistical Quality Control 172

1.3 Applications of Time Series Analysis

1. In Financial Decision Making: Econometrics is vital in finance and in financial

2. To Understand Past Behaviour: The analysis of a time series enables us to

3. Helpful in Planning: A time series analysis helps directly in business planning. A

4. To Understand Cyclical Variations: By time series analysis we can judge the

5. To Analyse Meaningful Comparisons: A time series analysis enables one to make

The Examples of Times Series are as Under :

5. The daily closing price of a share in Delhi stock exchange.

Requirements of a Time Series: The essential requirements of a time series are

1.4 Components of a Time Series

Hence, the main problems in time series analysis are:

It is shown below by a chart:

Components of a Time series

Long Term Short Term

1.4.2 Seasonal Variations

1. Those resulting from natural forces and

1.4.3 Cyclical Fluctuations

The purpose of studying the cyclical variation is:

1.4.4 Irregular or Random Fluctuations

2. It helps in planning and forecasting. It is very essential in business and economics

1.5 Analysis of Time Series

Another approach is to consider an additive model. Symobolically

1.6 Measurement of Trend

1.6.1 By Inspection or Graphic Method