T Test PDF

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

T-test Example

Rosenthal and Jacobson (1968) informed classroom teachers that some of their students showed
unusual potential for intellectual gains. Eight months later the students identified to teachers as
having potentional for unusual intellectual gains showed significiantly greater gains performance
on a test said to measure IQ than did children who were not so identified. Below are the data for
the students in the first grade:
Table 1: Scores for First Graders

F:\rsm\y520\y520_spr04\week_11\ttest_exp.fm

Experimental

Comparison

35

40

27

12

38

15

31

21

14

19

46

10

34

28

48

16

30

32

48

31

22

12

39

29

19

37

25

Mean = 27.15

11.95

SD = 12.51

14.62

Formula for T-test for indepentdent groups

X 1 X2
t = -------------------------------var1 var 2
----------- + ----------n
n

Substituting our values:

27.15 11.95t = -----------------------------------2


2
14.6
12.5
------------- + ------------20
20

15.2
15.2
15.2
15.2
27.15 11.95
t = ------------------------------------- = -------------------------------------------- = ---------------------------------- = ----------------- = ------------- = 3.54
4.298
2
2
156.25
213.45
7.81
+
10.67
18.48
14.6
12.5
---------------- + ---------------------------- + ------------20
20
20
20

Our obtained, or calculated t value is 3.54. Our degrees of freedom equals the total group size (40)
minus 2, or 38. Entering a t table with 38 degrees of freedom, we see that for alpha = .05 the
tabled value is 2.03 and for alpha = .01, the tabled value is 2.72.

Our calculated value is larger than the tabled value at alpha = .01, so we reject the null hypothesis
and accept the alternative hypothesis, namely, that the difference in gain scores is likely the result
of the experimental treatment and not the result of chance variation.
The comparison group is considered to be the standard for comparison. This may be the reason
the older adjective, control has been replaced by comparison when referring to the group that
does not receive the treatment. When the null hypothesis is rejected, an effect size is calculated.
For the t-test, subtract the mean of the comparison group from the mean of the treatment group
and divide the difference by the standard deviation of the comparison group. For this example, the
effect size is calculated as:

X 1 X2
27.15 11.95
t = ----------------- = --------------------------------- = 1.04
14.62
sd 2

which is considered to be a large effect.

/* Spss Code for T-test of indpendent groups */


data list / group 1-1 scr 3-4 .
begin data.
1 35
1 40
1 12
1 15
1 21
1 14
1 46
1 10
1 28

F:\rsm\y520\y520_spr04\week_11\ttest_exp.fm

1 48
1 16
1 30
1 32
1 48
1 31
1 22
1 12
1 39
1 19
1 25
2 2
2 27
2 38
2 31
2 1
2 19
2 1
2 34
2 3
2 1
2 2
2 3
2 2
2 1
2 2
2 1
2 3
2 29
2 37
2 2
end data.
value labels 1 'Treatment'
t-test groups = group(1,2)
/ variables = scr .

F:\rsm\y520\y520_spr04\week_11\ttest_exp.fm

2 'Comparison' .

T-Test
Group Statistics

group
scr

Mean

Std. Deviation

Std. Error
Mean

20

27.15

12.508

2.797

20

11.95

14.612

3.267

Independent Samples Test


scr
Equal variances assumed
Levene's Test for
Equality of Variances

t-test for Equality of


Means

1.877

Sig.

.179

df
Sig. (2-tailed)
Mean Difference
Std. Error Difference
95% Confidence Interval
of the Difference

F:\rsm\y520\y520_spr04\week_11\ttest_exp.fm

Equal variances
not assumed

3.534

3.534

38

37.117

.001

.001

15.200

15.200

4.301

4.301

Lower

6.493

6.486

Upper

23.907

23.914

You might also like