The CLASS statement names the classification variable that classifies the data set into
two groups; in this example there are two age groups. The VAR statement specifies the
response variable that is analysed.
T-TEST PROCEDURE
Variable: SCORE
AGE N Mean Std Dev Std Error
6 10 13.20000000 5.07280330 1.60416126
7 10 19.50000000 4.16999867 1.31866936
Variances T DF Prob>|T|
Unequal −3.0338 17.4 0.0074
Equal −3.0338 18.0 0.0071
For H0: Variances are equal, F′=1.48 DF=(9.9) Prob>F′=0.5686
Figure 8.11: Independent t-test
comparing ages 6 and 7 on total score
correct for 30 subtraction test items
InterpretationLook first at the bottom of the output where the result of performing the folded F′ test for
homogeneity of variance (equal variances) is shown. The p-value indicates that F′
statistic is not significant. This means that the variances are not significantly different
Therefore, the results of the equal variance t-ratio are used (pooled variance estimate).
Here the p-value associated with a t-value of −3.0338 is 0.0071, which is significant at
the 5 per cent level. The results are significant at the 1 per cent level but α (the
significance level or probability of incorrectly rejecting the null hypothesis) was set
initially to 0.05 so this should be reported; statistical significance should never be based
on the results of the test. The null hypothesis of no difference is rejected, and we can
conclude that there is a difference in means.
8.5 Paired t-test (related)When to UseWhen two group means are to be compared which are from the same sample, that is,
paired measurements for each subject in the sample, the paired t-test (sometimes called
dependent t-test) should be considered. Measurements would also be related if two
independent samples were matched and then the means of the two samples compared. A
paired t-test is often used to analyze the results of a ‘before’ and ‘after’ research design.
For example, a researcher may want to know whether there is any improvement in
vocabulary acquisition scores which could be attributed to the effect of a reading
programme. Subjects’ vocabulary score would be determined before commencement of
the reading programme, and they would be measured again after the programme had been
completed. Each subject would have paired vocabulary scores one measure before the
programme and one after. The purpose of a repeated measures analysis using the paired t-
Inferences involving continuous data 303