Professional Documents
Culture Documents
Applied Statistics
Applied Statistics
Applied Statistics
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
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.
(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.
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.
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
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)
(xi)
Metrization Theorems and Paracompactness 11
D etailed C ontents
(xii)
12 Topology
(xiii)
Metrization Theorems and Paracompactness 13
(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 Objective Type Questions 159
Ø Answers 165
(xv)
Metrization Theorems and Paracompactness 15
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 Objective Type Questions 211
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".
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
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.
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.
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.
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.
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
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.
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.
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
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
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:
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.
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. 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.
Example 3: Find 3-yearly and 4-yearly moving average for determining trend values in
the following time series data:
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 — —
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
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.
1984 86 — —
1985 84 — —
2002 108 — —
2003 126 — —
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:
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
Example 5: Below are given the figures of production ('000' quintals) of a sugar factory:
Production 80 90 92 83 94 99 92
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
The values of ' a ' and ' b ' are given by:
21 630
a = Y - b X = 90 - 2 ´ 3 = 84 where X = = 3, Y = = 90
7 7
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.
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:
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.
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:
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
(i) For finding the values of ' a ' and ' b ', we solve the following normal equations:
Y t = 90 .01 + 7 .11X
The trend values are obtained by giving different values of X, which are shown in
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.,
Y t = 90 .01 + 7 .11 ´ 6
Y t = 90 .01 = 7 .11 ´ 8
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)
24 a - 36 b = 143.4
- 24 a + 120 b = - 134.8
84b = 86
.
86
.
Þ b= = 0102
.
84
. = 8a - 12 ´ 0102
478 .
8a = 1.224 = 47 .8 or 8a = 47 .8 + 1.224
49.024
8a = 49.024 or a = = 6128
.
8
y = 6128
. + 0102
. x
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:
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.
Values of Y 6 10 15 39 25 20 15
Y = a + bX + cX 2
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
130 = 7a + b × 0 + c × 28 ...(1)
57 = a ´ 0 + b ´ 28 + c ´ 0 ...(2)
57 = 28b ...(2)
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
171
\ c=– = - 2.04
84
130 = 7a + 28 (– 2.04)
7a = 187 .12
18712
.
\ a= = 26.73
7
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
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)
When middle year is taken as origin the above equations reduce to:
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 .
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.
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
1.6.6.2 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.
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
Example 12: Assuming that trend is absent, determine if there is any seasonality in the
data given below:
Years Questions
I II III IV
Years Quarters
I II III IV
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
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.
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
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
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
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 I II III IV
Year I II III IV
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.
I 30 35 31 31 34
II 26 28 29 31 36
III 22 22 28 25 26
IV 31 36 32 35 33
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
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
= 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
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
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
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]
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)
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
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.
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:
(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:
21. Fit a second degree parabola and find the trend vaules for the following data:
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
24. Calculate the seasonal indices by the "Ratio-to-Moving Average" method from the
following data:
1998 I 75 2000 I 90
II 60 II 72
III 54 III 66
IV 59 IV 85
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.
(a) Less than one year (b) More than one year
(c) More than five years (d) None of these [B.R.A. Agra 2012]
(c) Method of moving averages (d) None of these [B.R.A. Agra 2005]
(c) Cyclic variation (d) Irregular variation [B.R.A. Agra 2006, 09]
(c) Ratio to moving average method (d) Ratio to trend method [B.R.A. Agra 2006]
10. A time series has :
11. The method of minimum least squares is connected with the analysis of time series
for measuring :
(a) The allow the elimination of the components from the series
(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 :
38. For the given five values 15, 24, 18, 33, 42 the three year moving averages are:
(c) Both (a) and (b) (d) Neither (a) nor (b)
(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
45. In the time series 2, 0, –2, 0, 2, 0, –2, 0, 2 all the four yearly moving averages are:
46. The regular movements occurring in a period of more than one year are called as:
(a) Vt = T + S + C + R (b) Vt = T ´ S + S ´ R
(c) Vt = T ´ S ´ S ´ R (d) Vt = T + S + C ´ R
57. The sales of a departmental store on Christman are associated with the component
of a time series:
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
Objective Type Questions
Multiple Choice Questions
1. (b) 2. (d) 3. (c) 4. (c)
57. (b)
10. seasonality
True or False
1. False 2. True 3. 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.
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.
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.
merit of measuring relative changes and also not affected by the extreme
observations.
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.
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
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.
Solution:
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.
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:
Commodities Prices
1998 2000
A 100 140
B 80 120
C 160 180
D 220 240
E 40 40
56
Commodities Prices
A 100 140
B 80 120
C 160 180
D 220 240
E 40 40
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
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
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
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
å p01 ´100
p
P01 =
N
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.
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.
B 40 45 60
C 30 35 45
D 10 12 15
E 20 22 23
p0 p1 p2
= 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
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
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:
A 1 6 3 5
B 3 5 8 5
C 4 8 10 6
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
é Sp q Sp q ù
P01 = ê 1 0 ´ 1 1 ú ´100
êë Sp0 q0 Sp0 q1 úû
æ138 115 ö
= ç ´ ÷ ´100
è 53 44 ø
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:
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
êë Sp0 q0 Sp0 q1 úû
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.
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
êë Sp0 q0 Sp0 q1 úû
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.
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
p0 q0 p1 q1
B 15 5 20 5 75 75 100 100
D 5 16 5 14 80 70 80 70
é Sp q Sp q ù
Fisher's ideal index = P01 = ê 1 0 ´ 1 1 ú ´100
êë Sp0 q0 Sp0 q1 úû
é 470 530 ù
= ê ´ ´100
ë 513 595 úû
é 249100 ù
= ê ´100
ë 305235 úû
é Sp q Sp q ù é 595 513 ù
P10 = ê 0 1 ´ 0 0 ú = ê ´
êë Sp1 q1 Sp1 q0 úû ë 530 470 úû
According to test,
é Sp q Sp q ù é 470 530 ù
Here, P01 = ê 1 0 ´ 1 1 ú = ê ´
êë Sp0 q0 Sp0 q1 úû ë 513 595 úû
Wheat 30 50 32 50
Gram 25 40 30 35
Barley 18 50 16 55
p0 q0 p1 q1
æ 3600 3530 ö
= ç ´ ÷ ´100
è 3400 3365 ø
= 1111
. ´100 = 1054
. ´100 = 105.4
A 6 50 10 60
B 2 100 2 120
C 4 60 6 60
[C.C.S. Meerut 2005, 06]
p0 q0 p1 q1
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.
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.
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.
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:
Solution:
1996 80 — 100
Example 14: From the fixed base index numbers given below, prepare chain base index
numbers.
Solution:
Year Fixed base index no. Conversion Chain base index no.
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
1982 93 . ´ 93
10167 94.55
100
Example 16: Construct fixed-base index numbers from the following chain base index
numbers:
Solution: Conversion of chain base index number into fixed base index number.
Year Chain base index no. Conversion Fixed base index no.
1982 71 . ´ 71
7875 55.91
100
1984 95 . ´ 95
5871 55.77
100
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.
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.
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.
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:
Sugar per kg 50 kg 13 16
q0 p1 p1q0
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.
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
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.
Food 300 50
Clothing 350 10
Fuel 200 5
Rent 280 15
Miscellaneous 140 20
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
Cloth 75 100 x x
´100 = 1.333 x
75
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
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:
I 210 1,680
II 215 3,000
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:
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:
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
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.
4. Index numbers indicate only average or approximate trend of changes. Hence, they
should be interpreted keeping this limitation into consideration.
E XERCISE
Long Answer Type Questions
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)]
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.
[C.C.S. Meerut 2001]
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]
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
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.
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.
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
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:
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.
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
1995 12 4 5
1995 22 27 7
20. From the following data, which index do you consider appropriate to construct and
why ? Prepare that index.
88
21. Compute Fisher's ideal index number from the following data and check whether
the time reversal test is satisfied.
P 2 7 6 6
Q 3 6 2 3
R 4 5 8 5
S 5 4 2 4
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.
[C.C.S. Meerut 2004]
4. Which index number is equal to A.M. of Paasche's and Laspeyre's index numbers.
[C.C.S. Meerut 2004]
5. Which index number is used to determine the grant of D.A. to the workers ?
[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]
15. How will you choose a base year for constructing an index number.
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)]
23. Define Laspeyre's Paache's and Fisher Index numbers. Show that Fisher's formula is
an ideal formula [C.C.S. Meerut 2008]
(c) Average of base and current (d) Product of base and current
(c) Both (a) and (b) (d) Neither (a) nor (b) [B.R.A. Agra 2006, 11]
(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:
(c) Both (a) and (b) (d) None of these [B.R.A. Agra 2009]
90
Sp1 q1
(c) (d) None of these
Sp0 q0 [B.R.A. Agra 2008, 09]
12. The first and fore most step in the construction of Index number is :
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 :
30. Index number for the base period is always taken as:
(c) Price in the given year × 100 (d) Price in the base year × 100
(a) Sum of the price relatives (b) Average of the price relatives
34. The value at the base time period serves as the standard point of comparison:
(a) 2 (b) 3
(c) 4 (d) 5
p p
(c) P = n ´100 (d) P = 0 ´100
p0 pn
Sp0 qn
(c) ´100 (d) None of these
Spn qn
94
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:
(c) Both (a) and (b) (d) Neither (a) nor (b)
44. If the relation P01 ´ P10 = 1, it means that the price indices satisfy:
(a) Time reversal test (b) Factor reversal test
46. The first and foremost step in the construction of index number is :
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]
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]
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.
A NSWERS
Long Answer Type Questions
14. P01 = 11127
. 15. (i) 118.96 (ii) 119.79
21. 209.01
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)
True or False
1. T 2. F 3. T 4. T 5. F
6. T 7. T 8. F 9. F 10. T
16. F 17. F
❍❍❍
UNIT
II
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.
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.
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.
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’.
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
r eg eh = the correlation between error scores from different testing occasions (for two
parallel tests, g and h).
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.
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
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.
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 .
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
difference whether we use test g or h. So in order that g and h may be parallel tests, we
assume that,
x i = ti + ei and since m e = 0,
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 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
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.
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.
3.9.2 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.
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.
B1 a b a+b
B2 c d c+d
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
Long Answer Type Questions
1. Discuss different types of Scaling giving its purpose.
11. What do you understand by test reliability? Define parallel test and discuss it.
(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
A NSWERS
Objective Type Questions
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.
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.
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.
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.
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:
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".
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.
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.
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
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
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.
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.
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
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
Specific death rates are standardised in a number of ways. We shall consider here only
two methods:
1. Direct Method 2. Indirect Method
(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.
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
m xA m xB PxS
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:
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
PxS
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,
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,
PxS mS
x PxS mS
x PxA m xA PxA m xA
^ 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
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.
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
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
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:
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
(in thousands)
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.,
46. 2
= ´ 2255 = 1041.81 per thousand.
100
133
1. The number of births and number of deaths are exactly equal and distributed
uniformly throughout the year.
Thus, population with constant size (because number of births and deaths are equal) and
constant age and / sex composition over time is called stationary.
2. If constant mortality and fertility rates are experienced at each age and
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:
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.
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.
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
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 + ...
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.
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
Life Table
x lx dx qx Lx Tx e0x
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 — —
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.
x 11 12 13 14
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
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
Age in lx dx px qx Lx Tx e0
x
years X
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))
0 T 27296732
e30 = 30 = = 35.8098
l30 762270
0 T 26536307
e31 = 31 = = 34.98
l31 758580
Age x lx dx px qx Lx Tx e0
x
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
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
4 60 30 0.50 45 60 1.0
5 30 30 1.00 15 15 0.5
6 0 — — — — —
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
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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'.
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.
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.
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.
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
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.
E XERCISE
Long Answer Type Questions
1. What do you mean by vital statistics ? Mention the various sources from where the
vital statistics data can be obtained.
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]
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.
(per 1000)
11. Compute the standarised death rate for the following data:
12. Find out the values of a, b, c, d, e and f of the following life table:
21 690673 — — — — e f
5 ? 400 ? ? ? ? ?
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:
lx : 97 59 32 75 6 2 0
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
(c) There is no mortality in female birth till their child bearing age
(a) Tx (b) Lx
32. If Tx = 3 and l x = 4 then what will be the value of e0x in life table:
33. The standard number of births 10,000 originating a life table is known as:
34. The death rate belonging to the age group 0-1 years is called as:
f Px f Bx
(c) fix = ´1000 (d) fix = ´100
f Bx f Px
38. If the value of net reproduction rate greater than unity then:
39. The probability of dying of a person of age between x and (x + 1) years is known as:
(a) 0 to ¥ (b) 0 to 1
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
(a) Tx / l x (b) d x / l x
(c) 2 q x / 2 – q x (d) q x / 2 – q x
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
Long Answer Type Questions
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)
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
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.
It has been established that if the variations are only due to chance causes, the
observations will follow "normal curve" described later on.
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
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
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)
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
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
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.
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
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
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
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.
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
1 k
E ( x) = S E( x i) = m P
k i =1
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.
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
= 0.5262 cm = 0.5198 cm
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
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
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.
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
Thus, k = 20, n = 5
S X 1263.4
\ X= = = 6317
.
k 20
SR 215
and R= = = 10.75
k 20
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
U.C.L. = D4 R = 211
. ´10.75 (for n = 5, table value of D4 = 211
. )
= 22.68
=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
Solution:
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
8 18 20 27 42 60 167 33.4 42
9 15 30 39 62 84 230 46.0 69
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
U.C.L. = X + A2 R
= 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
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.
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
Solution:
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
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
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.
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
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
Compute 3-s limits for three charts respectively. You may use the following control factors
for n = 6:
A2 B3 B4 D3 D4
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
U.C.L. = X + A2 R L.C.L = X - A2 R
= 25.87 = 19.73
U.C.L = D4 R L.C.L = D3 R
= 12.83 =0
U.C.L = B4 s L.C.L = B3 s
= 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. 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.
d
p=
n
where, p = sample proportion defective
d = number of defectives in a sample
n = size of the sample
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
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.
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
S.No. d p S.No. d p
Spi 3 .5095
1 k 3.5095
Thus, p= S p = = 0 .1595
k i =1 i 22
q = 1 - 01595
. = 0.8405
pq
U.C.L. = p + 3
n
01595
. ´ 0.8405
= 01595
. +3
2000
= 01595
. + 3 0.000067
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
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.
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
= 016
. + 3 ´ 0.0082 = 016
. - 3 ´ 0.0082
= 0.1846 =0.1354
Example 10: Draw a suitable control chart from the following and comment on your
findings
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
\ q = 1 - p = 1 - 0.03 = 0.97
n d d pq pq pq U.C.L. L.C.L.
p= 3´
n n n n
1900 59
195
Y
p-chart for variable sample size
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.
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
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
.
. + 3 192
= 192 . ´ 0.904 = 192 . ´ 0.904
. – 3 192
= 1.92 + 3 173568
. = 1.92 – 3 173568
.
= 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.
\ q = 1 - p = 1 - 01
. = 0.9
= 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
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
= l + 3 l ; = l -3 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
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:
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
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
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
3. The data is collected purposively 3. Data collected for other purposes may
for the use of variable charts. also be used.
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.
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.
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
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).
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
C ( np) x
- np
L( p) ~ å e x!
x =0
206
Procedure
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.
E XERCISE
Long Answer Type Questions
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.
[C.C.S. Meerut 2004]
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.
[C.C.S. Meerut 2002]
(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]
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
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
Defectives found: 0 1 2 2 3 5 1 2 3 1
Sample No.: 21 22 23 24 25 26 27 28 29 30
Defectives found: 1 1 4 2 2 4 1 3 3 2
Sample No.: 31 32 33 34 35 36 37 38 39 40
Defectives found: 3 6 2 3 2 3 1 0 2 0
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
(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
(a) A2 (b) B2
(a) 9 (b) 10
5. If one sample point falls out side the control limits then it means:
(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
(a) c - 3 c (b) c + 3 c
(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
(a) X + A2 R (b) X + A1 R
(c) x = D2 R (d) x + D1 R
[B.R.A. Agra 2010]
14. Which chart is used to control the number of defects in each unit of product ?
15. For constructing a p-chart in case of observed data, if n = 9, p = 0.36, then U.C.L. is :
(a) 9 (b) 27
(a) Which can not be detected (b) Which can not be controlled
(c) Neither (a) nor (b) (d) (a) and (b) both
(a) Uncontrollable
(b) Tolerable
20. Variation in the quality of a product due to assignable causes arises due to :
(c) Both (a) and (b) (d) Neither (a) nor (b)
(c) Both (a) and (b) (d) Neither (a) nor (b)
30. The expected sample size required to arrive at a decision about the lot is called:
(c) Both (a) and (b) (d) Neither (a) nor (b)
(a) B4 . s (b) B3 . s
(c) s + B4 (d) s + B3
215
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
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
38. The upper control limits of x and R charts when standards are not given are
40. If the lower control limit is found negative in case of control charts for attributed, it
is taken as:
(a) 0 (b) 1
42. The relation between expected value of R with S.D. s with usual constant factors is:
(c) Both (a) and (b) (d) Neither (a) nor (b)
(a) c ± 3 c (b) c + 3 c
(c) c ± 3 1 / c (d) c ± 3 1 / c
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:
50. When the value of population process range R ¢ is not known, then with usual
notation, the upper control limit for x-chart is:
(c) x + D2 R (d) x + D1 R
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
Objective Type Questions
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)
True or False
1. F 2. T 3. T 4. F 5. F
6. T 7. F 8. F 9. T 10. F
❍❍❍