INFERENTIAL STATISTICS

QUALITIES OF GOOD RESEARCH INSTRUMENT

Research –made instruments such as tests, questionnaires, rating scales, interviews, observation schedule, and
others should meet the qualities of a good research instrument.

1. Validity
2. Reliability
3. Usability

Validity

The degree to which a test or measuring instrument measures what it intends to measure. This measuring instrument has
something to do with soundness, what the test or questionnaire measures in terms of its effectiveness and how could it be
applied.

Types of Validity

1. Content validity
-the extent to which the content or topic of the test is truly representative of the content of the course. It is essentially
involved in the systematic examination of the test content to determine whether it covers a representative sample of
the behavior domain to be measured. It is important that the behavior domain to be tested must be systematically
analyzed to make certain that all major aspects are covered by the test items and in correct proportions.
- is described by the relevance of the test to different types of criteria such as thorough judgments of subject matter
experts, statements of behavioral objectives, analysis of research- made test questions, among others. The content
validity depends on the relevance of the individual ‘s responses to the behavior area under consideration rather
than an apparent relevance of the item content.
- is commonly used in evaluating achievement test. A well – constructed achievement test should cover the
instruction not just the subject matter.

2. Concurrent validity
- Degree to which the test agrees or correlates with a criterion set up as an acceptable measure. The criterion is
always at the time of testing. It is applicable to tests employed for the diagnosis of existing status rather that for
the prediction of future outcome.

3. Construct validity
- the extent to which the test measures a theoretical construct or trait. This involves tests such as those of
understanding, appreciation, and interpretation of data. Examples are intelligence and mechanical aptitude tests.

Reliability
- the extent to which a test is dependable , self- consistent and stable. It means a test agrees itself. It is concerned
with the consistency of responses from moment to moment. Even if the person takes the test twice, the test yields
the same results. However, a reliable test may not always be valid.

METHODS OF TESTING RELIABILITY TEST

1. Test – Retest method

The same research instrument is administered twice to the same group of subjects and the correlation
coefficient is determined. The limitations of this method are:
1). When the same time interval is short, the subject may recall his previous responses and this tends to make
the correlation coefficient high.
2). When the time interval is long , such factors as unlearning, forgetting, among others may occur and may
result in low correlation of the test.
 3) Regardless of the time interval separating the two administrations, others vary environmental conditions
such as noise, temperature, lighting, and other factors may affect the correlation coefficient of the research
instrument.

2. Spearman –Rank Correlation of Coefficient or Spearman Rho.

- is a statistic used to measure the relationship between paired ranks assigned to individual scores on two variables.
Thus, this is used to correlate the scores in a test-retest method for a non- parametric statistic.

3. Pearson Product Moment Correlation Coefficient ( PPM)

PPM or Pearson r is used if two variables to be correlated are both in interval or ratio measurement and have a linear
relationship ( for a parametric statistic)

4. Split –Half Technique

5. Cronbach Alpha
6. Kuder Richardson 21

Testing the Significance of a Product – Moment Coefficient of Correlation

 The null hypothesis for a product –moment coefficient of correlation maybe tested ( Yamane, 1973) with the
use of t- statistic.

r = n ∑XY - (∑X) (∑Y)___

[n∑X2-(∑X)2] [n∑Y2-(∑Y)2 ]

______
t = r √n–2
1 – r2

A t- statistic has a t-distribution with n-2 degrees of freedom . In the given t- statistic , r is the sample
product moment coefficient of correlation, while n is the number of paired observations. The null hypothesis
of no significant correlation between the two given variables is rejected if the computed t falls in the areas of
rejection at the appropriate level of significance. With a two-tailed test, if the computed value is positive and
is greater than the absolute of the tabled t, H0: p = 0 is rejected. Otherwise it is accepted. Acceptance of
H0 would mean that the two variables being correlated have no significant relationship ( i.e. They are
independent of each other).

Note: Use table for a student’s t distribution. If the computed value is greater than the tabular value, the
null hypothesis is rejected.
GR ≥ TV Decision Reject the null hypothesis.
When p  0.05 ,there is a significant relationship.

CORRELATIONAL ANALYSIS
Many research studies in the field of psychology, sociology, industry and business, education and even in
medicine deal with relationships between two or more variables. Statistics provide a wide range of tools and
techniques necessary in making studies. One of these is correlational analysis, which specifically used to measure
the degree of linear relationship or association between two variables. To determine the nature of the correlation, a
graph of the data is plotted on a scatter diagram or scattergram .
 CORRELATION COEFFICIENT may be positive or negative. A positive correlation is present when high
values in one variable are associated with high values of other variables or vice- versa. On the other hand, when
high value on one variable is associated with low values of the other variables or vice - versa , a negative
correlation is present. A perfect positive correlation is presented by + 1.00 value with a perfect negative correlation
is presented by a – 1.00. Correlation coefficients are computed and the most widely used measures of correlation is
the Pearson Product Moment Correlation Coefficient or simply Pearson r. There are two basic assumptions
underlying the use of the Pearson r i. e. linear relationship is present, and the level of the measurement of the data
for two variables are either in interval or ratio scale.

Range of Value Interpretation

0.90 to 1.00 ( -0.90 to – 1.00 ) Very high positive ( negative) correlation

0.70 to 0.89 ( -0.70 to – 0.89) High positive ( negative) correlation

0.50 to 0. 70 ( -0.50 to -0. 69) Moderate positive ( negative) correlation

0. 30 to 0. 49 ( -0. 30 to -0.49) Low positive ( negative) correlation

0.00 to 0. 29 ( 0.00 to -0.29) Little , if any correlation

TESTING HYPOTHESES
A hypothesis test is a statistical procedure designed to test a claim about a population.

Steps in testing hypothesis:

1. Set up the null and alternative hypotheses. ( H0 and Ha)

2. Calculate test statistics ( Determine the statistic tool and solve)

3. Determine the critical values( determined by the significance level or alpha level, the most common one to use is 5% or
referred to as 95%confidence level for a confidence interval)

4. Drawing conclusions and making interpretations.( Reject H0 at a level ¶, or Fail to reject H0 (or accept H0).

Every hypothesis test contains two hypotheses. The null hypothesis, denoted by H0. The null hypothesis states that
the population parameter is equal to the claimed value.

Along with every null hypothesis is an alternative hypothesis, denoted by Ha or H1. If H0 turns out to be false, you
conclude that Ha is true.

Type I Error - Rejection of the null hypothesis when it is true.

Type II Error - Acceptance of the null hypothesis when it is false.

The probability of committing a type I error is also called the level of significance of the test and is denoted by ¶ .

Sometimes the level of significance is called the size of the critical region.

FIGURING OUT p - values

( Review Normal Probability Curve)
Example:

Suppose you want to test H0 : µ = 0 versus Ha : µ > 0 , and your test statistic is Z = 1.96 ( based on a large sample ),
What is your p-value?
 Solution: Because the alternative hypothesis is “ > “ , you will need to find the area above Z = 1.96 using the NPC
table.The corresponding percentile is 97.73 % or .9773, which means the area beyond ( above) 1.96 which is

1.0000 - .9773 = .0227.

Therefore. p = .0227 an d this is less than .05 ( level of significance). This is the p-value that appears at the lower part
of the table when you present the significant difference stating * significant when p  0.05

This is the p-value. When you work with p-values , use them in decimal form , not percents.

Interpreting p-values
* For small p-values ( generally less than .05) , reject H0 ( meaning significant , data does not support H0)

* For large p - values (generally greater than .05 ), fail to reject H0 ,in our common term, accept H0 ( meaning not
significant , sample data does not contain enough evidence to refute it.

* For p- values on or close to the borderline between accepting and rejecting , results produced are marginal ( it can go
either way)

Spearman –Rank Correlation of Coefficient or Spearman Rho

Example:
Respondents x y Rx Ry d d2

1 90 70 2.0 7.5 -5.5 30.25

2 43 31 13.0 12.5 0.5 0.25

3 84 79 6.5 3.0 3.5 12.25

4 86 70 4.5 7.5 -3.0 9.00

5 55 43 11.0 10.5 0.5 0. 25

6 77 70 8.5 7.5 1.0 1.00

7 84 75 6.5 4.5 2.0 4.00

8 91 88 1.0 1.0 0.0 0.00

9 40 31 14.0 12.5 1.5 2.25

10 75 70 10.0 7.5 2.5 6.25

11 86 80 4.5 2.0 2.5 6.25

12 89 75 3.0 4.5 -1.5 2.25

13 48 30 12.0 14.0 -2.0 4.00

14 77 43 8.5 10.5 -2.0

4.00
∑ d2 = 82
Spearman –Rank Correlation of coefficient or Spearman Rho

Formula r = 1 - 6 ∑ d2

N( N2 – 1)
 Substitute to the formula

r = 1 - 6 (82)
14( 142 – 1)

r = 1 - 6 (82)
14( 196– 1)

r = 1 - 492
14( 195)

= 1 - 492
2730
= 1 - 0.1802
= 0.8198

round to the nearest two decimal places

r = 0.82 high positive

PPM or Pearson Product Moment Correlation Coefficient

Pearson Product Moment (PPM) or Pearson r is used if two variables to be correlated are both in interval or ratio measurements and have a
linear relationship.

r = n ∑XY - (∑X) (∑Y)___

[n∑X2-(∑X)2] [n∑Y2-(∑Y)2 ]

Example: Data on two sets of scores for calculating Pearson r between Science X and Math Y test scores

Pupil X Y XY X2 Y2

A 5
B 4
C 2
D 1
E 3
4 20 25 16
5 20 16 25
3 6 4 9
1 1 1 1
4 12 9 16

n=5 ∑X=15 ∑Y=17 ∑XY=59 ∑X2=55 ∑Y2=67

(∑X)2 = (15)2 = 225 (∑Y)2= (17)2 = 289

r xy = 5(59) – 15(17)____________
√ [ 5 (55 )−( 15 ) ( 15 ) ] ¿ ¿]
= 295-255
√[275−225¿][335−289 ]¿
= 40
√(50)( 46)
= 40 40
√ 2300 = 47.9583
r = 0.834 r = 0.83
There is a high relationship between the Science and Math test scores.

Testing the Significance of a Product –Moment Coefficient of Correlation

The null hypothesis for a product- moment coefficient of correlation maybe tested ( Yamane, 1973) with the use of t- statistic.
t = r n-2
1 – r2

A t statistic has a t distribution with n – 2 degrees of freedom. In the given t statistic, r is the sample product moment coefficient of
correlation, while n is the number of paired observations. The null hypothesis of no significant correlation between the two given variables is
rejected if the computed t falls in the areas of rejection at the appropriate level of significance. With a two –tailed test, if the computed t is
positive and is greater than the tabled value of t, null hypothesis H 0, is rejected. If the absolute value of the computed t is greater than
the absolute value of the tabled t , H 0 :p = 0 is rejected. Otherwise, it is accepted. Acceptance of H 0 would mean that the two variables
being correlated have no significant relationship ( ie. ;they are independent of each other)
As in the example, to test if the r is significant we use t.

t = 0.83 5- 2_____

1- 0.832

= 0.83 3

1- 0.6889

= 0.83 (3.10535)

= 2.577

Computed Value CV t = 2.58

Using the table for student’s t distribution, the tabled value TV at 5% level is 2.353. Since the computed value of 2.58 is greater than the
tabled value, the null hypothesis is rejected. The obtained r of 0.83 is highly significant. GR CV > TV Decision: Reject Null
Hypothesis

This also means that the p value is less than 0.05

Interpretation: There is significant relationship between Math and Science scores of the respondents.

CORRELATIONAL ANALYSIS
Many research studies in the field of psychology, sociology, industry and business, education and even in medicine deal with
relationships between two or more variables. Statistics provide a wide range of tools and techniques necessary in making studies. One of these is
correlation analysis, which is specifically used to measure the degree of linear relationship or association between two variables. To determine
the nature of the correlation, a graph of the data is plotted on a scatter diagram or scattergram as shown below.

5
4
3
2
1
0
0 0.5 1 1.5 2 2.5 33 3.5
3.5 44 4.5
4.5

High Positive Correlation Moderate Positive Correlation

67
56
45
4
3
3
22
11
0
0 0.5
0.5 11 1.5
1.5 22 2.5
2.5 3 3 3.53.5 4 4 4.54.5

High Negative Correlation Zero Correlation

Correlation coefficient may be positive or negative. A positive correlation is present when high values in one variable are
associated with high values of other variables or vice versa. On the other hand, when high value on one variable is
associated with low values of the other variables or vice versa, a negative correlation is present. A perfect positive
correlation is presented by +1.00 value with a perfect negative correlation is presented by a -1.00 value. Correlation
coefficients are computed, and the most widely used measures of correlation is the Pearson Product Moment Correlation
Coefficient or simply Pearson r. There are two basic assumptions underlying the use of the Pearson r, i.e., linear
relationship is present, and the level of the measurement of the data for the two variables are either in interval or ratio scale.

Range of value:*
0.90 to 1.00 (-0.90 to -1.00) Very high positive (negative) correlation
0.70 to 0.90 (-0.70 to -0.90) High positive (negative) correlation
0.50 to 0.70 (-0.50 to -0.70) Moderate positive (negative) correlation
0.30 to 0.50 (-0.30 to -0.50) Low positive (negative) correlation
0.00 to 0.30 (0.00 to -0.30) little, if any correlation

Sample for test of Normality Kolmogorov- Smirnov

Tests of Normality
Kolmogorov-Smirnova Shapiro-Wilk
sex Statistic df Sig. Statistic df Sig.
availmentdiscount male .109 108 .003 .967 108 .009
female .104 230 .000 .964 230 .000
availmentexemption male .118 108 .001 .960 108 .002
female .119 230 .000 .963 230 .000
availmentGovernment male .122 108 .000 .950 108 .001
female .083 230 .001 .967 230 .000
availemnt_overall male .123 108 .000 .967 108 .008
female .097 230 .000 .972 230 .000
a. Lilliefors Significance Correction

EXERCISES

Test the significance of relationship.

1. Calculate Pearson’s r for the following set of data. Interpret the obtained result.

Test Scores 20 30 45 15 30 33 20
in Algebra X
Test Scores 25 40 48 20 25 38 26
in Science Y
2. Use Spearman’s Rho to calculate the significant relationship for the data on leadership and maturity of 8 students.

Student Strength of Degree of

Leadership (Rank) maturity( Rank)
A 1 1 3. Test if there is a significant relationship between Math and English
B 4 3 grades of 6 students selected at random.( Assume that the data is normally
C 2 4 distributed)
D 5 2
E 8 6 Math 75 92 86 78 89 95
F 7 8
G 6 7 English 78 89 87 85 78 93
H 3 5
 STATISTICAL TOOLS

STEPS IN COMPUTING THE Z-TEST

1. Compute the mean of each group

___
X = ∑X
N

Where: __
X = mean
∑X = sum of scores
N = number of cases
2. Compute the standard deviation with the
S= standard deviation with the use of formula

S= N (∑X2) – (∑X)2
N ( N-1)

Where:
S = standard deviation
N = no. of cases
(∑X2) = sum of the squares of scores
(∑X)2 = square of the sum of scores
___ __
Z = X1 - X2

S12 +S22
n1 n2

where Z = critical ratio

__
X1 = mean score of 1st group
__
X2 = mean score of 2nd group
S1 = standard deviation of 1st group
S2 = standard deviation of 2nd group
n1 = total no. of cases of 1st group
n2 = total no. of cases of 2nd group

Example : Z – test

The National Education Association survey on starting salaries of college graduates. Independent samples of n1 = 32 accounting
graduates and n2 = 35 liberal arts yield the following starting salaries. Data are in thousands of pesos rounded to the nearest thousand pesos

Accounting Liberal Arts

24 20 23 20 20 17 21 21 22
23 25 26 22 19 24 18 17 18
24 22 21 23 19 21 20 19 19
26 26 24 25 21 20 17 20
25 20 22 24 19 19 19 17
24 25 25 24 18 23 20 19
22 22 21 24 21 19 21 20
21 22 25 27 21 16 16 21

At 5% significant level , does it appear that accounting graduates have a higher mean starting than liberal art graduates?
H0 There is no significant difference between the mean starting salaries of the accounting and liberal art graduates.

S1 = N (∑X2) – (∑X)2
N (N – 1)

= 32 ( )–( )2
32 (31) ** Supply the blanks with the correct figures

S1 = ___________

S2 = 35 ( )–( )
35 (34)

S2 = _______________

__
X 1 = ∑X = = _______ = __________
N 32
__
X2= ∑X = ________ = ___________
N 35

__ __
Z = X 1 - X2
S12 + S22
N1 N2

Substitute to the formula.

Z =
at 5% level of significance, the tabular value is 1.645

Decision: Interpretation:

Interpretation: There (is a significant difference) or (no significant difference) in the mean starting salaries of the two groups
The mean starting salaries of accounting graduates is __________ than that of the liberal graduates.

T-Test
The owner of a chain washes needs to decide between two brands of hot waxes. One of the brands, Sureglow, cost less than the other brand,
Clean Sheen. Therefore, unless there is a strong evidence that the second brand outlasts the first, the owner will purchase the first brand. With
the cooperation of several local car dealers, 30 cars are randomly selected to take part in the test. Fifteen of the 30 cars are waxed with Sureglow
and 15 with Clean Sheen. The cars are then exposed to same environmental conditions. The test results shown below give the effectiveness
time, in days.

Sureglow Clean Sheen

87 90 92 92 91
88
93 90 91 92 91
92
91 89 93 93 92
93
88 87 92 93 94
89
91 91 93 94 91
90

At the 1% level significance, does Clean Sheen seem to have longer effectiveness time, on the average, than Sureglow.
A.) Ho: (There is no significance difference……………….)
B.) α = 1% or 0.01
S1 = 15 (121, 373) – (1349)2 = 3.78
15 (14)
 S2 = 15 (127, 712) – (1384)2 = 1.07
15 (14)

(3.78)2 + (1.07)2 df = (n1 + n2 ) -2

15 15

= 0.0732 + 0.0763
= 0.1495

___
X1 = 1349 = 89.93
15
___
X2 = 1384 = 92.27
___ __ 15
t = X1 - X2 = 89.93 – 92.27 = - 2.34_______ = - 2.34 = - 2.28
S12 + S22 3.782 + 1.072 0.95256 + 0.07633 1.028
N1 N2 15 15

CV TV
/-2.28 / 2.467
Tabular value @ 1 % level of significance is 2.467
The absolute value of the computed value is less than the tabular value.
Conclusion: Accept the null hypothesis
Interpretation: There is no significant difference between the effectiveness of using Clean
Sheen and Sure Glow. Clean Sheen has higher mean than Sure Glow.

Partial Table for critical values of the t- distribution

(Taken from Table IV of R.A. Fisher, Statistical Methods for Research Workers)

df ¶ @ 0.05 0.01

1 6.314 31.821
2 2.920 6.965
3 2.353 4.541
4 2.132 3.747
5 2.015 3.365
6 1.943 3.143
7 1.895 2.998
8 1.860 2.896
9 1.833 2.821
10 1.812 2.764
11 1.796 2.718
12 1.782 2.681
13 1.771 2.650
14 1.761 2.624
15 1.753 2.602
16 1.746 2.583
17 1.740 2.567
18 1.734 2.552
19 1.729 2.539
20 1.725 2.528
21 1.721 2.518
22 1.717 2.508
23 1.714 2.500
24 1.711 2.492
25 1.708 2.485
26 1.706 2.479
27 1.703 2.473
28 1.701 2.467
29 1.699 2.462
Inf. 1.645 2.326

Sample Titles:

*The Level of Behavior and Its Relationship to Performance Task of Department Heads:Basis for Intervention Program

*Level of English Proficiency of UNO-R College Freshmen

 *Effects of Multiple Intelligences Approach in the Communication Skills of Education Students

*Communicative Competence of the La Consolacion College Senior College Students

*The Trimedia Competence of the Junior and Senior Mass Communication Students

*Problems Encountered by Employees of the Provincial Government Office and Their Job Performance

Problems
Examples:
Descriptive
1. What is the level of behavior of exceptional children when categorized as task behavior, social behavior, and general behavior?

2.What is the degree of classroom climate of a public school in Antique during the school year 2019-2020 in terms of discipline,

student interaction , learning assessment, and attitude and culture as assessed by teachers and students when they are taken as a whole and
grouped according to designation?
3 .What is the effect of multiple intelligences approach in the communication skills of education students when taken all together and
when grouped according to variables?
4.What is the level of English proficiency of UNO – R college freshmen when taken as a whole and when grouped according to
variables? (age, gender, school of origin.)
5.What is the level of awareness and extent of practice on natural disaster preparedness before, during, and after disaster of High school
students of a Southern Negros school in Negros Occidental during the the school year 2019-2020 when they are taken as awhole and when
grouped according to grade level?

6. What is the extent of effectiveness of social development initiatives in areas affected by armed conflict in Negros Occidental in terms of
a ) Health b) Housing c) Education d) ) Peace and Security ?
8. What are the challenges encountered by implementers in extending development initiatives in areas affected by armed conflict in Negros
Occidental?
9. What are the lived experiences of residents in the areas affected by armed conflict in Negros Occidental?

Inferential
1.. Is there a significant relationship in the level of behavior of exceptional children when they are categorized as task, social, and general

behavior?

2. Is there a significant relationship between the level of behavior and level of school performance of exceptional children in ….?

3. Is there a significant difference in the level of proficiency of UNO – R college freshmen when they are grouped according to variables?

4. . Is there a significant relationship in the level of behavior of exceptional children when they are categorized as task, social, and general
behavior?

5..Do multiple intelligences have significant difference in the communicative skills of education students when they are grouped
according to variables?

6..Is there a significant difference in the level of communicative competence of CSC Senior College students when they are grouped
according to variables?

7. Is there a significant difference in the extent of effectiveness of social development initiatives in areas affected by armed conflict in
Negros Occidental as assessed by the residents when they are grouped according to variables?
8. Is there a significant difference in the extent of effectiveness of social development initiatives in areas affected by armed conflict in
Negros Occidental as assessed by the implementers when they are grouped according to variables?

Hypotheses
Examples:
1. There is no significant relationship in the level of behavior of exceptional children when they are categorized as task, social, and
general behavior.

2. There is no significant relationship in the level of behavior and the level of school performance of exceptional children.
3. There is no significant difference in the level of proficiency of UNO-R College Freshmen when they are grouped according to
variables.
 4. There is no significant difference in the level of communicative competence of CSC Senior College students when they are
grouped according to variables.
5. There is no significant difference in the extent of effectiveness of social development initiatives in areas affected by armed conflict
in Negros Occidental as assessed by the residents when grouped according to variables.
6. There is no significant difference in the extent of effectiveness of economic development initiatives in areas affected by armed
conflict in Negros Occidental as assessed by the implementers when grouped according to variables.

Interpretation of Scores / Mean Range

Lickert scale
* This interpretation is used when the problem is descriptive.
Mean Range Interpretation
Level Extent
4.21 - 5.00 - very high level very great extent
3.41 – 4.20 - high level great extent
2.61 – 3.40 - average ( moderate) average/moderate
1.81 – 2.60 - low (or poor) level poor extent
1.00 – 1.80 - very low( or very poor) very poor extent

* This interpretation is based on scores . For example , the perfect score in a test is 80 and the lowest possible score is 0.

To make the range of scores , follow the steps.

First, find the Range by subtracting the highest possible score to the lowest possible score.
80 - 0 = 80
Second, divide the difference by the number of interpretation.
If the interpretation covers 5 as very poor, poor, average, high , very high
80 / 5 = 16.
Third, make the range, by starting with 0 to 16.00. , then add the difference to the next range. As :
Scores Interpretation
65 - 80 very high
49 - 64 high
33 - 48 average / moderate
17 - 32 poor
0 - 16 very poor

Example:
The mean of the item/variable was computed and resulted to 2.68. Look at the mean range
2.61– 3.40. The interpretation of the item/problem is average.

ANOVA OR F-TEST
Analysis of variance is a technique designed to test whether or not more than two samples (or groups), are significantly different from
each other.

While the t-test, together with the z-test is used to test non-significance of difference between a simple pair of samples, taking them one
pair at a time, ANOVA is used as a simultaneous test taking the samples all at a single time and applied only once.

Method A Method B Method C

No. X1 X2 X3 X12 X22 X32
1 84 70 90 7,056 4,900 8,100
2 90 75 95 8,100 5,625 9,025
3 92 90 100 8,464 8,100 10,000
4 96 80 98 9,216 6,400 9,604
5 84 75 88 7,056 5,625 7,744
6 88 75 90 7,744 5,625 8,100
TOTAL 534 465 561 47,363 36,275 52,573

∑X = 1,560
∑X2 = 136,211
TSS – Total variation or the total sum of squares
(∑X)2
2
TSS = ∑X - ---------
N

(1,560)2
= 136,211 - ---------
18

2,433,600

= 136,211 - ----------------------- TSS = 136,211 – 135,200

18
= 1,011

Between – column variance or between – column sum of squares

(∑X)2 ∑(sum of each column)2

SSb = ------------------------ - ---------------------------------
No. of rows N

1 15602
= -------------- (5342 + 4652 + 5612) - -------------------
6 18

1 2,433,600
= ----------------- ( 285,156 + 216,225 + 314,721) - ---------------------
6 18

816,102 2 ,433,600
= ------------------------ - ------------------
6 18

= 136,017 – 135,200
= 817

The within-column variance or within-column sum of squares

SSw = TSS - SSb

= 1, 011 – 817 SSw = 194

ANOVA TABLE
Source of variation Sum of Squares df MSS = SS/df

Between – column SSb 817 2 408.5

Within – column SSw 194 15 12.93

TOTAL TSS 1, 011 17 F = 31.59

Total df = total number of items – 1

=N-1
= 18 – 1
= 17

Or of total df = rk -1
= (3x6) – 1
= 18 – 1 = 17
Between – column df = no. of columns – 1
dfb =K–1
=3–1
=2
Within – column df = total df – between – column df
dfw = dft - dfb
= 17 – 2
= 15

F – test Formula

MSSb
F = -----------------------
MSSw

408.5
= -----------------------
12.93

F = 31.59

Using the F-test table for the tabular value, the corresponding degree of freedom of the greater MSS is designated as n1 and the degree of
freedom of the smaller MSS is designated n2.

At 5% significance level, the tabular value is 3.68

Thus, 31.59 > 3.68

Decision : Reject the null hypothesis
Interpretation : There is a significant difference among the samples.

RESEARCH INSTRUMENTS EVALUATION FORM

Direction: Kindly rate or value the rating scale attached herewith according to the following criteria set forth by Carter V. Good and Douglas F. Scates.
4 = Very Good
3 = Good
2 = Fair
1 = Poor

Criteria for Evaluation Rating

1 2 3 4

1. The questionnaire is short enough such that the

respondent will not reject it because it will not use
much of his precious time.

2. The questionnaire has face appeal much that the

respondent will be inclined to accomplish it fully.

3. The questionnaire can obtain some depth to the

responses and avoid superficial answer or information.

4. The items and their alternative responses are not too

suggestive nor too unstimulating.

5. The questionnaire can elicit responses, which are

definite but not mechanically forced.

6. The items were stated in such a way that the responses

will not be embarrassing to the person concerned.

7. Items are framed in such a manner as to allay suspicion

on the part of the respondent concerning hidden
purposes in the questionnaire.

8. The questionnaire is not so narrow, not restrictive or

limited in its scope or philosophy.

9. The responses to the questionnaire, when taken as a

whole could answer the basic purpose for which the
questionnaire is designed and, therefore are considered
valid.

Rating
Overall Rating =
 RESEARCH INSTRUMENTS EVALUATION FORM

Direction: Kindly rate or value the rating scale attached herewith according to the following criteria set forth by Carter V. Goo d and Douglas F. Scates.

4 = Very Good
3 = Good
2 = Fair
1 = Poor

Criteria for Evaluation Rating

J1 J2 J3 J4 J5

1. The questionnaire is short enough such that the

respondent will not reject it because it will not use
much of his precious time.

2. The questionnaire has face appeal much that the

respondent will be inclined to accomplish it fully.

3. The questionnaire can obtain some depth to the

responses and avoid superficial answer or information.

4. The items and their alternative responses are not too

suggestive nor too unstimulating.

5. The questionnaire can elicit responses, which are

definite but not mechanically forced.

6. The items were stated in such a way that the responses

will not be embarrassing to the person concerned.

7. Items are frames in such a manner as to allay suspicion

on the part of the respondent concerning hidden
purposes in the questionnaire.

8. The questionnaire is not so narrow, not restrictive or

limited in its scope or philosophy.

9. The responses to the questionnaire, when taken as a

whole could answer the basic purpose for which the
questionnaire is designed and, therefore are considered
valid.

Rating
Overall Rating =
 NONPARAMETRIC STATISTICS

Statistical tests, such as the z, t, and F tests are called parametric tests. Parametric tests are statistical tests for population
parameters such as means, variances, and proportions that involve assumptions about the populations from which the samples were
selected. One assumption is that these populations are normally distributed. But if the population is not normally distributed a
branch of statistics known as nonparametric statistics or distribution free statistics is used. This is when the population from
which the samples are selected is not normally distributed. Nonparametric statistics can also be used to test hypotheses that do not
involve specific population parameters.

For example, a sportswriter may wish to know whether there is a relationship between the rankings of two judges on the diving abilities of
10 varsity swimming players. Or in another situation, a sociologist may wish to determine whether men and women enroll at random for a
specific drug rehabilitation program. These type of problems are tested by nonparametric tests or distribution –free tests. The non-
parametric tests are the sign test, the Wilcoxon rank sum test, the Wilcoxon signed- rank test, the Kruskal-Wallis test, and the runs test. The
Spearman rank correlation coefficient which was discussed earlier, a statistic for determining the relationship between ranks is also non-
parametric.

Advantages of Nonparametric Method

1. They can be used to test population parameters when the variable is not normally distributed.
2.They can be used when the data are nominal or ordinal.

3. They can be used to test hypotheses that do not involve population parameters.
4. In most cases, the computations are easier and faster than those for the parametric counterparts.
5. They are easier to understand.
6. Data need not be quantitative measurements but could be in the form of qualitative responses such as “ defective”versus “non-defective”,
“yes” versus “no”,etc or frequently are values of the ordinal scale to which ranks are assigned.

Disadvantages of Nonparametric Method

1. They are less sensitive than their parametric counterparts when the assumptions of the parametric methods are met. Therefore, larger
differences are needed before the null hypothesis can be rejected.
2. They do not utilize all the information provided by the sample.
3. They tend to use less information than the parametric tests. For example, the sign test requires the researcher to determine only whether
the data values are above or below the median, not how much above or below the median each value is.
4.They are less efficient than their parametric counterparts when the assumptions of the parametric methods are met. That is, larger sample
sizes are needed to overcome the loss of information. For example, the non-parametric sign test is about 60% as efficient as its parametric
counterpart, the z test. Thus, a sample size of 100 is needed for use of the sign test, compared with a sample size of 60 for use of the z-test
to obtain the same result.
Since there are both advantages and disadvantages to the nonparametric methods, the researcher should use caution in selecting these
methods. If the assumption cannot be met , the nonparametric methods are a valuable tool for analyzing the data.

** Note :If a parametric and a nonparametric are both applicable to the same set of data , one should probably
avoid the “quick and easy” nonparametric test and carry out the more efficient parametric technique.

THE KRUSKAL-WALLIS TEST

The analysis of variance uses the F test to compare the means of three or more populations. The assumptions for the ANOVA test are
that the populations are normally distributed and that the population variances are equal. When these assumptions cannot be met, the
nonparametric Kruskal-Wallis, sometimes called the H test, can be used to compare three or more means. This test is always a right
tailed t-test; Formula:

H= 12 ( R21 + R22 + …. + R2k ) - 3 ( n + 1)

n( n-1) (n1 n2 nk )

Where;
R1 = sum of the ranks of sample 1
n1 = size of sample 1
R2 = sum of the ranks of sample 2
n2 = size of sample 2

Rk = sum of the ranks of sample k

nk = size of sample
N = n1 + n2 + …+ nk
 K = number of samples

Note: If H falls in the critical region H > x2nwith v = k-1 degrees of freedom, reject H0 at the level of significance, otherwise , accept H0

Sample problem.

1. A researcher tests three different brands of breakfast drinks to see how many milliequivalents of potassium per quart each contains. The
following data are obtained.

Brand A Brand B Brand C

4.7 5.3 6.3

3.2 6.4 8.2
5.1 7.3 6.2
5.2 6.8 7.1
5.0 7.2 6.6

At α = 0.05, is there enough evidence to reject the hypothesis that all brands contain the same amount of potassium?

SOLUTION

Step 1 State the hypothesis and identify the claim.

Ho : There is no difference in the amount of potassium contained in the brands (claim).

H1 : There is a difference in the amount of potassium contained in the brands.

Step 2 Find the critical value. Use the chi-square table, Table G, with df = k – 1 (k = number of
groups). With α = 0.05 and df = 3 – 1 = 2, the critical value is 5.991.
Step 3 Compute the test value.
a. Arrange all the data from the lowest to highest, and rank each value.
Amount Brand Rank
3.2 A 1
4.7 A 2
5.0 A 3
5.1 A 4
5.2 A 5
5.3 B 6
6.2 C 7
6.3 C 8
6.4 B 9
6.6 C 10
6.8 B 11
7.1 C 12
7.2 B 13
7.3 B 14
8.2 C 15

b. Find the sum of the ranks of each brand.

Brand A 1+2+3+4+5 = 15
Brand B 6 + 9 + 11 + 13 + 14 = 53
Brand C 7 + 8 + 10+ 12 + 15 = 52

c. Substitute in the formula .

Where
 N = 15 R1 = 15 R2 = 53 R3 = 52

n1 = n2 = n3 = 5

H= 12 ( R21 + R22 + …. + R2k ) - 3 ( n + 1)

n( n-1) (n1 n2 nk

Therefore,

H=
12
15(15+1) 5 5
+ (
152 532 ... 522
+ +
5 )-3(15+1)=9.38
Step 4 Make the decision. Since the test value, 9.38, is greater than the critical value, 5.991,
the decision is to reject the null hypothesis.

Step 5 Summarize the results. There is enough evidence to reject the claim that there is no difference in the amount of potassium
contained in the three brands. Hence, all brands do not contain the same amount of potassium.

CHI – SQUARE TEST For Goodness of Fit

Suppose a market analyst wished to see whether consumers have any preference among five flavors of a new fruit juice. A sample of 100 people
provided the following data:

Cherry Strawberry Orange Lime Grape

32 28 16 14 10

If there were no preference, one would expect that each flavor would be selected with equal frequency. In this case, the equal frequency is 100/5
= 20. Approximately 20 people would select each flavor.

Frequency Cherry Strawberry Orange Lime Grape Total

Observed 32 28 16 14 10 100
Expected 20 20 20 20 20 100

H0: Consumers show no preference for flavors of the fruit soda.

Ha: Consumers show a preference.

Degree of freedom = number of categories - 1

df = 5 – 1 = 4

χ2 = Σ (O–E)2
E
O- Observed frequency
E – Expected frequency

Assumptions for the Chi- Square goodness of fit- Test

1. The data are obtained from a random sample.
2. The expected frequency for each category must be 5 or more.

Step 1: State the null hypothesis

Step 2: Find the critical value

Step 3: Compute the test value.

χ2 = Σ (O–E)2
E
 = ( 32 – 20)2 + ( 28 – 20)2 + ( 16- 20)2 + ( 14-20)2 + ( 10- 20)2
20 20 20 20 20

χ2 = 18

df = 5-1 = 4 at 5% level of significance , the tabular value is 9.488( Use chi-square table)

Step 4: Decision: 18 > 9.488 Reject the null hypothesis.

Step 5; Interpretation: There is enough evidence to reject the null hypothesis. Consumers showed significant preference for flavors of fruit
soda.

Another formula for n ( ad - bc)2

χ2 = ____________________________

( a+ b) (c+d)(a+c)(b+d)

a b a+b
c d c+d
a+ c b+d n =a+b+c+d
Where a,b,c,d are similarly defined as the cells in a contingency table for
calculating the phi-coefficient.
Example. 1. A sample of resident’s opinions regarding the curfew hour at 10:00 pm -4:00 am.
Here is a contingency table showing the responses. Test if there is a significant relationship between a resident’s opinion
regarding the curfew hour at 10:00 pm -4:00 am.

Sex Opinion Total

Favor Not favor
Male a 200 b 250 a+b
450
Femal c 120 d 100 c +d
e 220
a+c b+d a+b+c+d
Total 320 350 670
H0 . There is no significant relationship between a resident’s sex and opinion regarding the curfew hour from 10;00 pm to 4;00 am.
Ha: There is a significant relationship between a resident’s sex and opinion regarding the curfew hour from 10;00 pm to 4:00 am.
Using the formula:

670 ( 200 x 100 - 250 x 120)2

χ2 = ________________________________________

( 450) (220)(320)(350)

Using your scientific calculator, press the keys 670 x ( 200x100-250x120)2 ÷ (450x220x320x350) = 6.04

Your final answer is χ2 = 6.04

Degree of freedom df = ( c-1) (r-1)
c is number of columns, r is number of rows
With 1 degree of freedom , the value of χ2 at .05 level of significance is 3.84.
H0 is therefore untenable and we conclude that sex is dependent of resident’s opinion.

Partial Table Critical values of Chi Square

 Probability under H0 that χ2  chi square
df .05 .01
1 3.84 6.64
2 5.99 9.21
3 7.82 11.34
4 9.49 13.28
5 11.07 15.09

COMMON STATISTICAL TOOLS AND ITS USES

A) Descriptive
DATA STATISTICAL TOOL USES
1) All types of data to be presented in a Frequency Count
table or graph. (gender, age, income, To organize data using tables or graphs.
scores, etc.)

 2) All types of data to be presented in a
table or graph. (gender, age, income,
scores, etc.)
 3) Ratio and interval data (scores,
weights, scales, etc.)
 4) Ratio and interval data (scores,
weights, scales, etc.)
B) Inferential – Parametric
[Data derive from sample, data are assumed to be normally distributed, data are ratio and interval.]
This is used to test a Hypothesis. to make a generalization of the population from sample data.
DATA
 1) Ratio and interval data (scores,
weights, scales, grades, IQ,
occurrence, etc.)
[Sample size more than 30]
2) Ratio and interval data (scores,
weights, scales, grades, IQ,
occurrence, etc.)
[Sample size less than 30]
 3) Ratio and interval data (scores,
weights, scales, grades, IQ,
occurrence, etc.)
 4) Ratio and interval data (scores,
weights, scales, grades, IQ,
occurrence, etc.)
Percentage To measure the weight/value of each item with
respect to the whole group.
Mean or Median
To measure the center of distribution.
Standard Deviation To measure the spread or dispersion of the data.
STATISTICAL TOOL USES
Critical Ratio or Z - Test  To test the significant difference between two
for means sets of data.
 To test the significant difference between two
t-test sets of data.
for means
t-test To test the significant relationship (correlation)
for correlation coefficient between two sets of data.
Pearson Product Moment To describe quantitatively the relationship
Correlation (PPM) (correlation) between two sets of data. Use to test
the reliability of the research instrument.

 5) Ratio and interval data (enrolment,
oil, price hike, population data, etc.)
 6) Ratio and interval data (scores,
weights, scales, grades, IQ,
occurrence, etc.)
 7) Three or more sets of ratio or interval
data. (scores, weights, scales, grades,
IQ, occurrence, etc.)
Linear Regression Analysis To forecast the value of variables given to current
values.
One-way ANOVA( Analysis To test for hypothesis when comparing
of Variance) (difference) among three or more groups or sets of
data.
Post Hoc Test Conducted after a significant result is obtained
[Shceffe or Tukey’s] from ANOVA or Kruskal-Wallis Test to identify
which two sets of means are significant.

C) Inferential – Non-Parametric
[Data derived from sample; data cannot be assumed to be normally distributed, any types of data]
This is used to test a Hypothesis, to make a generalization of the population from sample data.
 DATA STATISTICAL TOOL USES
Chi Square

χ2 Test dependence (one variable is affected or not

 1) Discrete data (not measured data) related by the other). It is not a measure of degree
of relationship, it is merely used to estimate the
likelihood that some factor/s other than chance
(sampling error) accounts for the apparent
relationship.

 2) Data are ordinal or ranks. (scales)
 3) Ratio and interval data (scores,
weights, scales, etc.)
 4) All types of data. (gender, age,
income, scores, scales, etc.)
Spearman’s Rank To measure the relationship between two
Correlation variables. (Equivalent to PPM)
To test for hypothesis, significant difference
between three or more variables. (Equivalent to
Kruskal-Wallis Test One-way ANOVA)
Wilcoxon Signed –Rank To test for hypothesis, difference between two
Test sets of data. (Equivalent to t-test).

OTHER TESTS FOR DISTRIBUTION –FREE OR NONPARAMETRIC TESTS

TESTS FOR UNCORRELATED DATA

The Median Test

The median test is applied to see if two groups come from populations which have the same median. In using this test the size of the two
samples need not be the same.

Mann –Whitney U Test

A more powerful test than the median test for uncorrelated data .This is used with independently drawn random samples . the sizes of which
need not be the same. When the sample sizes are very small, that is when N1 and N2 are made up of eight or less measures, Siegel ( 1956) for a
method and tables that apply to such sample sizes.
When samples are bigger than 8, this is the formula to be used.

U1 = N1N2 + N1 (N1 + 1 ) - ∑ Rx
2
U2 = N1N2 + N2 (N2 + 1 ) - ∑ Ry
2
∑ Rx + ∑ Ry = N ( N + 1)
2
z = U1 - (N1 N2/ 2)
[N1 N2 ( N1 + N2 + 1) ] 12

A z value of this size leads to the rejection of the null hypothesis at level of significance.

Sample Problem:

X ( N = 7) Y ( N = 9) Rx Ry

14 18 11.5 15.0
 12 16 9.0 14.0

13 15 10.0 13.0

10 14 8.0 11.5

7 19 5.5 16.0

6 7 3.5 5.5

4 8 2.0 7.0

6 3.5

3 1.0

Nx Ny  Rx  Ry

Substitute to the formula:

* Finish the problem on board.

PROBLEMS.
1. Organize the data in a relative frequency table and make a pie chart. Answer the percent to two decimal places.

665 individuals were asked what type of vehicle they owned and here are the results. Category /Frequency /relative
frequency or Percent. SUV = 160 ; Truck = 145 ; Car = 135 ; Jeepney = 225.

2. The following table shows 10 brands of cigarettes showing milligrams of tar content, nicotene content. Calculate the
correlation coefficient to measure the degree of relationship between tar and nicotene content in cigarettes.

Cigarette Brand Tar Content Nicotene Content

A 14 0.9
B 17 1.1
C 28 1.6
D 17 1.2
E
16 1.0
F
13 0.8
G
24 1.5
H
25 1.4
I
18 1.2
J
31 2.0

3. Is there a significant difference in the level of level of performance of employees of when they are grouped according to
sex? Use 5 % level of significance .

Male Female

85 88
83 79
 78 80
81 83
95 90
93 96
88 93
96 85
92 82
94
89

4. Three types of models of computers were investigated as to the defects when they were returned to the company. They
were compared for defects as recorded. These computers were manufactured by the same company Test the hypothesis at
5% level of significance if there is a significant difference in the average number of defects for the three models.

Model

A B C

1) 4 5 8

2) 7 1 6

3) 6 3 8

4) 6 5 9

5) 3 5

6) 4
Total 23 21 36 80

5. Twenty college freshmen were divided into ten pairs , each member of the pair having approximately the same IQ.
One of each pair was randomly selected and assigned using modular learning and the other using on line learning.At the
end of the semester,each group was given the same examination and the results are tabled below. Test if there is a
significant difference in the two learning modalities. Data assumed to be normally distributed.

Pair On Line Modular

 Learning
1 76 81

2 80 82

3 87 78

4 90 85

5 87 90

6 92 92

7 78 80

8 82 77

9 79 76

10 93 84

Prepared by.Crispina Vedra- Diego, PhD

Professor