-35
The following comparisons were studied:
02<01 02<03 05<01
05<04 02<04
Group 02 served as a control group for Group 01. Group 05 served as a control group for Group 04. Group 05 served as a control group for Group 01. Group 02-03 was a pretest-posttest design. Group 04 was a posttest only group. This last group, 04, provided a comparison on carefully equated samples of an initial measure coming before and after, was more precise than 01-02 as far as selection was concerned and was superior to 02-03 comparison in avoiding test-retest effects.
This design controlled for the following sources of internal and external invalidity:
1. 02<01 and 05<04 -- history, testing, instrumentation,
interaction of testing and X, reactive arrangements.
2. 02<03 -- selection, mortality and reactive arrangements.
3. 02<04 -- testing, selection, interaction of testing and
X.
4. 05 provided a needed redundancy because 02 was involved in three comparisons that might appear to be confirmatory if 02 had an eccentric performance.