NCJ Number
52180
Journal
Psychological Bulletin Volume: 82 Issue: 3 Dated: (MAY 1975) Pages: 345-362
Date Published
1975
Length
18 pages
Annotation
AFTER A DISCUSSION OF THE POSSIBILITIES FOR SAMPLING ERROR AND BIASED SELECTION WHEN USING NONEQUIVALENT CONTROL GROUP EXPERIMENTAL DESIGN, FOUR STATISTICAL TESTS OF TREATMENT EFFECT ARE EVALUATED.
Abstract
NONEQUIVALENT CONTROL GROUP DESIGN CONSISTS OF PRETREATMENT AND POSTTREATMENT MEASURES OF A DEPENDENT VARIABLE WITH BIASED ASSIGNMENT TO TREATMENT GROUPS. DIFFERENT STATISTICAL TECHNIQUES USED TO ADJUST FOR THIS NONRANDOM ASSIGNMENT ARE THE ANALYSIS OF COVARIANCE, ANALYSIS OF COVARIANCE WITH RELIABILITY CORRECTION, RAW CHANGE SCORE ANALYSIS, AND STANDARDIZED CHANGE SCORE ANALYSIS. THE STRENGTHS AND WEAKNESSES OF EACH TECHNIQUE ARE ASSESSED, AND APPROPRIATE APPLICATION IS ILLUSTRATED. IT IS CONCLUDED THAT IF ASSIGNMENT TO TREATMENT GROUPS IS BASED ON PRETEST SCORES, ANALYSIS OF COVARIANCE IS APPROPRIATE. SELECTION BASED ON STABLE GROUP DIFFERENCES AND SELECTION THAT OCCURS MIDWAY BETWEEN PRETEST AND POSTTEST NECESSITATES CHANGE SCORE ANALYSIS. PROBLEMS WHICH OCCUR WITH RETROSPECTIVE MEASUREMENT OF THE TREATMENT AT POSTTEST AND POSSIBLE RIVAL HYPOTHESES OF CHANGES IN THE TREATMENT-EFFECT CORRELATIONS ARE DISCUSSED. VARIOUS SIGNIFICANCE TESTS PROPOSED BY RESEARCHERS ARE REVIEWED. IT IS WARNED THAT COMPENSATORY EDUCATION PROGRAMS CAN BE MADE TO LOOK MISTAKENLY INEFFECTIVE IF THEY ARE ANALYZED BY AN INAPPROPRIATE STATISTICAL TECHNIQUE RATHER THAN BY STANDARDIZED CHANGE SCORE ANALYSIS. THE USE OF QUASI-EXPERIMENTAL DESIGN AND THE IMPORTANCE OF GOOD STATISTICAL ANALYSIS TO THE EVALUATION OF REAL-WORLD SITUATIONS IS EXAMINED. REFERENCES ARE PROVIDED. TABLES AND FIGURES ILLUSTRATE THE MATHEMATICAL DISCUSSION AND AN APPENDIX DETAILS THE CALCULATIONS SUPPORTING DATA IN ONE OF THE TABLES. (GLR)