Significance of S

the Difference UNIT 3 SIGNIFICANCE OF THE di

of Frequency
DIFFERENCES BETWEEN MEANS
(T-VALUE)
Structure
3.0 Introduction
3.1 Objectives
3.2 Need and Importance of the Significance of the Difference between
Means
3.3 Fundamental Concepts in Determining the Significance of the Difference
between Means
3.3.1 Null Hypothesis
3.3.2 Standard Error
3.3.3 Degrees of Freedom
3.3.4 Level of Significance
3.3.5 Two Tailed and One Tailed Tests of Significance
3.3.6 Errors in Making Inferences
3.4 Methods to Test the Significance of Difference between the Means of
Two Independent Groups t-test
3.4.1 Testing Significance of Difference between Uncorrelated or Independent Means
3.5 Significance of the Difference Between two Correlated Means
3.5.1 The Single Group Method
3.5.2 Difference Method
3.5.3 The Method of Equivalent Groups
3.5.4 Matching by Pairs
3.5.5 Groups Matched for Mean and SD
3.6 Let Us Sum Up
3.7 Unit End Questions
3.8 Glossary
3.9 List of Formula
3.10 Suggested Readings

3.0 INTRODUCTION
In psychology some times we are interested in research questions like Do the
AIDS patients who are given the drug AZT show higher T-4 blood cell counts
than patients who are not given that drug? Is the error rate of typist the same
when work is done in a noisy environment as in a quiet one? Whether lecture
method of teaching is more effective than lecture cum discussion method?

Consider the question whether lecture method of teaching is more effective

than discussion method. For this investigator divides the class in to two
groups. One group is taught by lecture method and other by discussion
method. After a few months researchers administer an achievement test for
34 both the group and find out the mean achievement scores of the two groups
 f say M1 and M2. The difference between these two mean is then calculated. Significance of the
e Now the questions is whether the difference is a valid difference or it is differences between
Means (T-Value)
because of sampling fluctuation or error of sampling. Whether this difference
is significant or not significant. Whether on the basis of this difference, could
we conclude that one method of teaching is more effective than the other
method.

These question can be answered by the statistical measures which we are

going to discuss in this unit. To test the significance of difference between
mean we can use either the t-test or Z test. When the sample size is large, we
employ Z test and when sample is small, then we use the t test. In this unit we
are concerned with t test. We will get acquainted with the various concepts
related, to computation and description of t test.

3.1 OBJECTIVES
After reading this unit, you will be able to:
 Understand the need and importance of the significance of the difference
between means;
 Know about what is null hypothesis, standard error, degrees of freedom,
level of significance, two tailed and one tailed test of significance, type I
error and type II error; and
 Calculate the significance of difference between mean ( t-test) when
groups are independent, when there are correlated groups, groups matched
by pair and groups matched by mean and standard deviation.

3.2 NEED AND IMPORTANCE OF THE

SIGNIFICANCE OF THE DIFFERENCE
BETWEEN MEANS
In psychology sometimes we are interested in knowing about the significance
of the differences between populations. For example we are interested to
discover whether ten year old boys and ten year old girls differ in their
linguistic ability. Or we want to find out if children from high SES perform
and score better academically than children from low SES. We may also try
to find out at times, if two groups of persons coming from different
background differ in their agility factor. Thus many questions are asked and
to be answered in psychology for which one of the measures we use is the
Mean.

Let us take the first question on linguistic ability of boys and girls. First we
randomly select a large sample of boys and girls (large sample means the
group comprises of 30 or more than 30 persons.). Then we administer a
battery of verbal test to measure the linguistic ability of the two groups,
compute the mean scores on linguistic ability test of the two groups. Let us
say the obtained mean scores for boys and girls are M1 and M2. Now we try to
find the difference between the two means. If we get a large difference(M1 –
M2) in favour of the girls then we can confidently say that girls of 10 years of
age are significantly more able linguistically than 10 years old boys. On the
contrary if we find small difference between two means then we would
conclude that ten years old girls and boys do not differ in linguistic ability.
35
 Significance of An obtained mean is influenced by sampling fluctuation or error of sampling S
the Difference and whatever differences are obtained in the means, it may be due to this di
of Frequency sampling error. Even mean of population 1 and mean of the population 2 may
be the same but because of sampling error we may find the difference in the
range of 2 samples drawn from two populations. In order to test the
significance of an obtained difference we must first have a standard error(SE)
of the difference. Then from the difference between the sample mean and
standard error of difference we can say whether the difference is significant or
not. Now the question arises what do we mean by significant difference?
According to Garrett (1981) a difference is called significant when the
probability is high and that it cannot be attributed to chance that is
(Temporary and accidental factors) and hence represent a true difference
between population mean.

A difference is non significant when it appears reasonably certain that it could

easily have arisen from sampling fluctuation and hence imply no real or true
differences between the population means.

3.3 FUNDAMENTAL CONCEPTS IN

DETERMINING THE SIGNIFICANCE OF
THE DIFFERENCE BETWEEN MEANS
3.3.1 Null Hypothesis
This is a useful tool in testing the significance of differences. Null hypothesis
asserts that there is no true difference between the two population means, and
the difference found between the sample mean is therefore, accidental or
unimportant ( Garreet 1981). In the course of a study or an experiment, the
null hypothesis is stated so that it can be tested for possible rejection. For
example to study the significant difference in linguistic ability of 8 years old
girls and boys we select random sample of girls and boys and administer a
battery of verbal test, compute the means of the two groups. In this study the
null hypothesis may be stated thus: there exists no significant difference
between the linguistic ability of boys and girls. If this null hypothesis is
rejected then we can say one group is superior to the other.
3.3.2 Standard Error
The primary objective of statistical inference is to make generalisation from a
sample to some population of which the sample is part. Standard error
measures (1) error of sampling and (2) error of measurement. Suppose we
have knowledge of the true mean, means of the population, we randomly
select 100 representative sample from the population and compute their mean
and standard deviations. The standard deviation obtained from this
representative sample is known as standard error of the mean. The standard
error of the mean can be calculated by the following formula:
SEm or σm = σ/√N
Where
σ= The standard deviation of the sample mean
N= The number of cases in the sample.
If the standard error of measurement is large it shows considerable sampling
36 error.
 f 3.3.3 Degrees of Freedom Significance of the
e differences between
When a statistics is used to estimate a parameter the number of degrees of Means (T-Value)
freedom (d.f) available depends upon the restriction placed upon the
observations. One d.f. is lost for each restriction imposed. For example we
have five scores as 5,6,7,8 and 9 the mean is 7 and deviation of our scores
from 7 are -2, -1, 0, 1 and 2. The sum of these deviations is zero. In
consequence if any four deviations are known the remaining deviations may
be automatically determined. In this way, out of the five deviations, only four
(N-1) are free to vary as, the condition that “ the sum equals to “Zero” impose
restriction upon the independence of the 5 th deviation. Originally there were
5(N=5) degrees of freedom in computing the mean because all the observation
or scores were independent. But as we made use of the mean for computing
standard deviation we lost one degree of freedom.
Degrees of freedom varies with the nature of the population and the restriction
imposed. For example in the case of value calculated between means of two
independent variables, where we need to compute deviation from two means,
the number of restrictions imposed goes up to two consequently d.f. is (N-
1+N-2).

3.3.4 Level of Significance

Whether a difference between the means is to be taken as statistically
significant or not depends upon the probability that the given difference could
have arisen “by chance”. The researcher has to take a decision about the level
of significance at which he will test his hypothesis.

In social sciences .05 and .01 level of significance are most often used. When
we decide to use .05 or 5% level of significance for rejecting a null hypothesis
it means that the chances are 95 out of 100 that is not true and only 5 chances
out of 100 the difference is a true difference.

In certain types of data, the researcher may prefer to make it more exact and
use .01 or 1% level of significance. If hypothesis is rejected at this level it
shows the chances are 99 out of 100 that the hypothesis is not true and only 1
chance out of 100 it is true. The level of significance which the researcher will
accept should be set by researcher before collecting the data.

3.3.5 Two Tailed and One Tailed Tests of Significance

In many situations we are interested in finding the difference between
obtained mean and the population mean. Our null hypothesis states that the
M1 amd M2 do not vary and the difference between them is zero. (i,e. Ho:M1-
M2=0). Whether this difference is positive or negative we are not interested in
the direction of such a difference. All that we are interested is whether there is
a difference. For example we hypothesised that two groups will differ from
each other we don‟t know which group will have higher mean scores and
which group lower. This a non directional hypothesis and it gives rise to a
two-tailed hypotheses test. In other words the difference may be in either
direction and thus is said to be non directional.

In many experiments our primary concern is with the direction of the

difference rather than with its existence in absolute term. For example if we
are interested to determine the gain in vocabulary resulting from additions by
weekly reading assignment. Here we are interested in finding out the gain in
37
vocabulary. To take another example, if we say that training in yoga will
 Significance of reduce the degree of tension in persons, then we are clearly stating that there S
the Difference will be a reduction in the tension. In cases like this we make use of the one di
of Frequency tailed or non directional test to test the significance of difference between the
means.

3.3.6 Errors in Making Inferences

If the null hypothesis is true and we retain it or if it is false we reject it, we had
made a correct decision. But sometimes we make errors. There are two types
of errors Type I error and Type II error.

A Type I error, also known as alpha error, is committed when null hypothesis
(Ho) is rejected when in fact it is true.

A Type II error, also known as beta error, is committed when null hypothesis
is retained and in fact it is not true. For example suppose that the difference
between two population means (µ1- µ2) is actually zero, and if our test of
significance when applied to the sample mean shows that the difference in
population mean is significant we make a Type I error. On the other hand if
there is true difference between two population mean, and our test of
significance based on the sample mean shows that the difference in population
mean is “ not significant‟ we commit a type II error.

3.4 METHODS TO TEST THE SIGNIFICANCE

OF DIFFERENCE BETWEEN MEANS OF
TWO INDEPENDENT GROUPS (t-test)
3.4.1 Testing Significance of Difference Between
Uncorrelated or Independent Means
The step used to find out significance of differences between independent
mean are as below:
Step 1. Computation of the mean
Step 2. Computation of the combined standard deviation by using the
following formula:
x1=X1-M1 (deviation of scores of first sample from its mean)
x2= X2-M2 (deviation of scores of second sample from its mean)
SD = √ (Σx12+ Σx22)/(N1-1)+(N2-1)
Step 3. Computation of the Standard error of the difference between two
means by using following formula:
SED= SD / √(1/N1+1/N2)
Step 4. To Compute the t value for the different in two independent sample
mean. The following formula is used to determine t value is
t = (M1-M2)-0/SED
Step 5. Find out the degree to freedom. The (df) degree of freedom is
calculated using the following formula
df=(N1-1)+(N2-1)
Step 6. We then refer to table of t (can be found in any statistic‟s book)
38 distributions with the calculated degree of freedom df and read the t value
 f given under column .05 and .01of two tailed test. If our computed t value is Significance of the
e equal or greater than the critical t value given in table then we can say that t is differences between
Means (T-Value)
significant. If the computed value is lesser than given value then we will say
that it is non-significant.
Let us illustrate the whole process with the help of example
Example: An interest test was administered to 6 boys and 10 girls. They
obtained following scores is the mean difference between two groups
significant ?
Table: Scores of boys and girls and the t value calculation

Scores of Scores of
boys girls
X1 x1 x² X2 x2 x2²
28 -2 4 20 -4 16

35 5 25 16 -8 64
32 2 4 25 1 1

24 -6 36 34 10 100

36 -4 16 20 -4 16
35 5 25 28 4 16

31 7 49
24 0 0

27 3 9
15 -9 81

∑X1=180 ∑x1²= 110 ∑X2=240 ∑x2² =352

Calculation
M1=(∑X1/N1)
M1=180/6 =30
M2=(∑X2 / N2)
M1=240/10 =24
SD= √ [( ∑x21+∑x22)/(N1-1)+(N2-1)]
SD = √ [( 110+352)/(6-1)+(10-1)] = 5.74
SEd= SD [√( (N1+N2)/(N1N2)]
= 5.74 / √ [ (16)/(60)]
= 5.74 × .5164 = 2.96
t= (M1-M2)-0/SED
= (30-20)-0/2.96 = 2.03
df=(N1-1)+(N2-1)
df=(6-1)+(10-1) =14 39
 Significance of Entering the value in table on 14 d.f. We get 2.14 at the .05 and 2.98 at the.01 S
the Difference level, since our t is less than 2.14, therefore we will say that the mean di
of Frequency difference between boys and girls is non significant.
Let us take another example
Example
On an academic achievement test 31 girls and 42 boys obtained the following
scores.
Table: Mean scores and Standard deviation
Mean SD N df
Boys 40.39 8.69 31 30
Girls 35.81 8.33 42 41
Is the mean difference in favour of boys and significant at the .05 level.
First we will compute the pooled standard deviation by the following formula
SD = √ [(SD1)2 × (N1-1) + (SD2)2 × (N2-1)]/(N1+N2)
SD1= Standard Deviation of group 1 i.e. 8.69
N1= Number of subject is group 1 i.e. 31
SD2= Standard devotion of groups 2 i.e. 8.33
N2= Number of subject in group 2 i.e. 42
SD = [(8.69)2 × (31-1)+(8.33)2 × (42-1)/(N1-1)+(N2-1)]
SED= SD [( (N1+N2)/(N1N2)]
SED= 8.48 [(31+42)/(31X42)]
t= (M1-M2)-0/SED
= (40.39-35.81)-0/2.01 = 2.28
df=(N1-1)+(N2-1)
df=(31-1)+(42-1) =71
Entering Table with 71 df we find t entries of 2.00 at the .05 level and 2.65 at
the .01 level. The obtained t of 2.28 is significant at .05 level but not at the .01
level. We may say boys academic achievement is better in comparison to that
girls.

3.5 SIGNIFICANCE OF THE DIFFERENCE

BETWEEN TWO CORRELATED MEANS
3.5.1 The Single Group Method
In the previous section we discussed the problem of determining the
significance of difference between mean obtained by two independent groups
of boys and girls.
Some time we have single group and we administer the same test twice. For
example if we are intending to find out the effect of training on the students‟
educational achievement, first we take the measures of subject‟s educational
achievement before training , then we introduce the training programme, and
again we take the measure of educational achievement. In this we have single
group and administer educational achievement test twice. Such type of design
40
is known as single group design. In order to get the significance of the
 f difference between the means obtained in the before training and after training Significance of the
e we use the following method. differences between
Means (T-Value)

SED or σD = [σ2M1+ σ2M2- 2r σM1 σM2]

Where
σM1= Standard error of the initials test
σM2=Standard error of the finals test
r= coefficient of correlation between scores on initial
test and final test
t=(M1-M2)-0/ σD
Let us illustrate the above formula will the help of
following example.
Example
At the beginning of the session an educational achievement test in maths was
given to 100 IX grade students. Their mean was 55.4 and SD was 7.2. After
six months an equivalent from of the same test was given and the mean was
56.9 and SD was 8.0. The correlation between scores made on the first testing
and second testing was .64. Has the class made significant progress in maths
during six month? We may tabulate our data in the following manner
Table: Scores in the initial and final test of students
Initial Test Final Test
No. of students 100 100
Mean 55.4 56.9
SD 7.2. 8.0
Standard error of the 0.72 0.80
mean
r12 .64
Calculation
SED or σD = [σ2M1+ σ2M2- 2r σM1 σM2]
SEd= [( (.72)2+(.80)2-2x.64x.72x.80]
= [.5184+.6400-.7373] = .649
t= (M1-M2)-0/SED
= 1.5-0/.649 = 2.31
df=(N1-1)=(100-1) , df=99
From Table we look at the df = 99, and find that the value at .05 level is 1.98
and at .01 level is 2.63. Our obtained value is 2.31 therefore this value is
significant at .05 level and not on .01 level. Here we can say that class made
substantial improvement in mathematical achievement in six months.
3.5.2 Difference Method
When groups are small then we must prefer the difference method to that
given above.
Let us illustrate the use of this method with the help of following example. 41
 Significance of Example S
the Difference di
of Frequency
Ten subjects were tested on an attitude scale. Then they were made to read
some literature in order to change their attitude. Their attitude were again
measured by the same scale. The results of the initials and final testing are as
under.
Table: Results of initial and final testing of attitude
Initial Final Difference X = D-M x²
condition condition Cond 2- cond 1
(Mean = 8)
50 62 12 4 (12-8) 16
42 40 -2 -10(-2--8) 100
35 30 -5 -13(-5-8) 169
51 61 10 2 (10-8) 4
42 52 10 2 (10-8) 4
26 35 9 -1 (9-8) 1
41 51 10 2 (10-8) 4
42 52 10 2 (10-8) 4
60 68 8 0 (8-8) 0
70 84 14 -6 (14-8) 36
55 63 8 0 (8-8) 0
38 50 12 4 (12-8)) 16
Total ∑D = 96 ∑x² =
354
Note: The sum of scores of final condition is more than sum of initial
condition therefore we subtract scores of initial condition from scores of final
condition (Final condition –Initial condition) and add the score to find ΣD.
Mean = (∑D)/N
Mean = 96/12=8
SDD= √[(∑x2)/N-1]
SDD = √[354/11] = 5.67
SEMD= SDD/ (N)
SEMD= 5.67/ (12)
t= MD-0/SEMD
t=8-0/1.64
t=4.88
d.f.=12-1
Entering in table with 11 df, we find t entries of 2.20 and 3.11 at the .05 and at
the .01 levels. Our t of 4.88 is far above the .01 level. We can conclude that
subjects attitude changed significantly from initial to final condition.

42
 f 3.5.3 The Method of Equivalent Groups Significance of the
e differences between
In experiments when we want to compare the relative effect of one method of Means (T-Value)
treatment over another we generally take two groups, one is known as
experimental group and the other is known as control group. Here we have
two groups not a single group. For the desired results these two groups need to
be made equivalent. This can be done by (i) Matched pair technique or (ii)
Matched groups technique. These are explained below
i) Matched pair technique: In this techniques matching is done by pair.
Matching is done on variables which are going to affect the results of the
study like age, intelligence, interest, socio-economic status.
ii) Matched groups technique: In this technique instead of person to person
matching, matching of groups is carried out in terms of Mean and S.D.
3.5.4 Matching by Pair
Formula for calculation of standard error of difference between mean is:
SED or σD = [σ2M1+ σ2M2- 2r σM1 σM2] . Here
σ2M1= Standard error of mean 1
σ2M2= Standard error of mean 2
r=correlation between the two groups scores
t = ( M1-M2 )-0 /SED
Example
Two groups X and Y of Children 72 in each group are paired child to child for
age and intelligence. Both groups were given group intelligence scale and scores
were obtained. After three weeks experimental group subjects were praised for
their performance and urged to try better. The control groups did not get the
incentive. Group intelligence scale again was administered on the groups. The
data obtained were as follows. Did the praise affect the performance of the group
or is there a significant difference between the two groups.
Table:
Experimental Control group
group
No. of children in each group 72 72
Mean score of final test 88.63 83.24
SD of final test 24.36 21.62
Standard error of the mean of 2.89 2.57
final test
Correlation between experimental .65
and control group scores
SED or σD = [σ2M1+ σ2M2- 2r σM1 σM2]
SED or σD = [( (2.89)2+(2.57)2-2x.65x 2.89x2.57]
= 2.30
t= (88.63-83.24)-0/2.30=2.34
d.f. = 72-1= 71
43
 Significance of If we see the table we find that at 71 d.f. the value at .05 is 2.00 and at .01 is S
the Difference 2.65. The obtained t is 2.34 therefore this value is significant at .05 level and di
of Frequency not at .01 level. On the basis of the results is can be said that praise did have
significant effect in stimulating the performance of children.

3.5.5 Groups Matched for Mean and SD

When groups matched in terms of mean and S.D., the following formula is
used to calculate „t‟.
SED = √[σ2M1+ σ2M2 (1-r2]
t = (M1-M2)-0/ SED
SED= Standard error of difference
σ2M1 = Standard error of mean 1
σ2M2 = Standard error of mean 2
r= Correlation between final scores of two tests
The above formula can be illustrated by the following example
Example
The 58 students of academic college and 72 students of technical college were
matched for mean and SD upon general intelligence test. Then the
achievement on a mechanical aptitude test was compared. The Questions is do
the two groups enrolled in different courses differ in mechanical ability?
Academic Technical
No. of children in each group 58 72
Mean on Intelligence GTest (Y) 102.50 102.80
SD on Intelligence Test Y 33.65 31.62
Mean on Mechanical Aptitude (X) 48.52 53.51
SD on Mechanical Aptitude X 10.60 15.36
r 12 .50
SED = [σ2M1 / N1] + [σ2M2 / N2 ]-(1-r2 )
Therefore
SED or σD = [(10.60)2/58 +( 15.35)2/72]-[1-.25] = 1.97
t = (53.61-48.52)-0/1.97 = 2.58
d.f.=(N1-1)+(N2-1)=(58-1)+(72-1)=128
Entering the value in table we find that on 125 df (which is near to 128) the
value are 1.98 at .05 level and 2.62 at .01 our obtained value is 2.58. This is
significant at .05 level. We may say that two group differ in mechanical
aptitude.

3.6 LET US SUM UP

In the field of psychology some time we are interested in testing the
significance of difference between two sample means.
The sample may comprise of two independent groups and single groups tested
twice. Some time we have two groups matched by pair or matched for means
44
and S.D. The process of determining the significance of difference between
 f the Means is different in different conditions. We may broadly summarise the Significance of the
e procedure of calculating significance of differences between Means as under. differences between
Means (T-Value)
 Establish a null hypothesis.
 Decide a suitable level of significance .05 or .01
 Determine the standard error of the difference between means of two
samples.
 Compute the value of „t‟
 Find out the degrees of freedom.
 Determine the critical value of t from the „t‟ table.
 If the computed value is same or more than the value given in the table
then it is taken to be significant if the computed valued is less than the
given value it is considered as non significant.
 When the t-value is significant we reject the null hypothesis and when „t‟
value is not significant we retain the null hypothesis.

3.7 UNIT END QUESTIONS

1) Given below are some statements. Indicate whether the statement is true
or false
i) We commit a Type I error when we reject a null hypothesis
when it is really true. (T/F)
ii) In testing a hypothesis, one can make three types of error. (T/F)
iii) An exercise in hypothesis testing enables us to draw
conclusions about the estimated parameters. (T/F)
iv) For a given level of significance, we find that as the sample
size increases, the critical values of t get closer to zero. (T/F)
v) If the standardised sample mean exceeds the critical clue,
we should accept Ho. (T/F)
2) Differentiate between
1) Null hypothesis and alternative hypothesis.
2) One tailed test and two tailed test.
3) Type I error and Type II error.
3) Write short notes on the following
Concept of Standard error
Level of significance
4) The marks obtained in math‟s by 10 boys and 10 girls are given in the
following table. Find out whether their is any difference between the
mean of boys and girls.
Scores of boys Scores of girls
7 7
5 6
6 5
5 8
6 9
6 8
7 8 45
 Significance of 5) One group of boys (N=20) and one groups of girls (N=22) were tested S
the Difference on verbal ability test. They got following scores: di
of Frequency
Boys Girls
Mean 34.56 30.56
SD 5.68 6.98
Do the groups differ on the verbal ability test?
6) Ten persons are tested before and after the experimental procedure, their
scores are given below. Test the hypothesis that there is no change.
Before After
60 72
52 50
61 71
36 45
45 40
52 62
70 78
51 61
80 94
65 73
72 82
7) We take two groups one from technical classes and other from non
technical class and each group is compared on numerical learning test.
Both groups have been matched on in terms of means and standard
deviation on the basis of scores on general Intelligence test. Do the
groups differ in terms of mean numerical ability?.
The data are as under:
Non Technical
technical
N 58 72
Mean of Numerical reasoning test 48.52 53.61
SD on Numerical reasoning test 10.61 15.35
r between Intelligence and Numerical .50
reasoning test
8) A Vocabulary test was administered to a random sample of 8 students of
section A and and 7 students of section B of Class IX of a school . The
scores are:
Scores on 16 14 12 12 10 8 6 4
Section A
Scores on 14 8 7 6 4 4 1 2
Section B

46
 f Significance of the
e differences between
Is the difference between the means of two groups significant at o.05 level? Means (T-Value)

9) In the first trial of a practice period, 25 twelve-year-olds have a mean score

of 80.00 and a SD of 8.00 upon a digit-symbol learning test. On the tenth
trial, the mean is 84.00 and the SD is 10.00. The r between scores on the
first and tenth trials is .40. Our hypothesis is that practice leads to gain.
Answers 1
i) True (ii) False, (iii) True, (iv) True, (v) False
Answers 4 t= 1.66 non significant
Answers 5 t= 1.99 non significant
Answers 6 t= 4.88 significant at .01 level
Answers 7 t= 2.57 significant at .05 level
Answers 8 t= .80 not significant
Answers 9 t= 2.00 significant at .05 level

3.8 GLOSSARY
Null Hypothesis : A zero difference hypothesis. A statement about
the status Quo about a population parameter that
is being tested.
Alternative Hypotheses: A hypothesis that takes a value of population
parameter different from that used in the null
hypothesis. It states that there is a difference in
the groups on a certain characteristic that is being
tested.
Type I error : An error caused by rejecting a null hypothesis
when it is true.
Type II : An error caused by failing to reject a null
hypotheses when it is not true.
One tail test : A statistical hypothesis test in which the
alternative hypothesis is specified such that only
one direction of the possible distribution of
values is considered. It would state there will be
an increase in the performance of students after
training.
Two tailed Test : A statistical hypothesis test in which the
alternative hypothesis is stated in such way that it
included both the higher and the lower values of a
parameter than the value specified in the null
hypothesis, It would state that there will be a
difference (can be an increase or decrease) in the
group that undergoes training.
Significance level : The probability that a given difference arises
because of a chance factor or it is a true
difference.
Standard error : The standard deviation of a sampling
distributions. 47
 Significance of S
the Difference
3.9 LIST OF FORMULA di
of Frequency t test for two independent group
t= (M1-M2)-0/SED
SED= SD/√ (1/N1+1/N2)
SD = √(Σx12+ Σx22)/(N1-1)+(N2-1)
t-test for single group (difference method)
Mean = (∑D)/N
SDD= √ [(∑x2)/N-1]
SEMD= SDD/ √(N)
t= MD-0/SEMD
t –test for groups matched for mean and S.D
t = (M1-M2)/ SED
SED or σD = [(σ2M1+ σ2M2 )( 1-r2)]
t –test for groups matched for paired
SED or σD = [σ2M1+ σ2M2- 2r σM1 σM2]
t= (M1-M2)-0/SED

3.10 SUGGESTED READINGS

Garrett ,H.E. (1981) Statistics in Psychology and Education, Bombay, Vakils,
Feffer and Simons Ltd.

48