NCJ Number
59484
Journal
Social Work Research and Abstracts Dated: (1979) Pages: 27-32
Date Published
1979
Length
6 pages
Annotation
A POSITIVE RATIONALE FOR THE USE OF TRUE EXPERIMENTAL DESIGNS IN THE EVALUATION OF SOCIAL PROGRAMS IS PRESENTED; MAJOR OBJECTIONS TO RANDOMIZATION ARE REVIEWED AND ARGUMENTS TO COUNTER THEM HIGHLIGHTED.
Abstract
INCREASING PRESSURE TO EVALUATE SOCIAL SERVICE PROGRAMS HAS LED TO A DEBATE REGARDING WHICH RESEARCH DESIGNS FORM THE BASIS OF SOUND ASSESSMENTS. SOCIAL SCIENTISTS ARE DIRECTLY INVOLVED, WITH INPUT FROM THE POLITICAL SPHERE. DESPITE THE NUMEROUS ESSAYS DETAILING THE CLEAR METHODOLOGICAL AND STATISTICAL SUPERIORITY OF THE SO-CALLED TRUE EXPERIMENT, WHICH MAKES USE OF THE RANDOM ASSIGNMENT OF SUBJECTS TO EXPERIMENTAL AND CONTROL GROUPS, THIS APPROACH IS NOT USED IN ALL APPROPRIATE CASES. SOME OF THE COMMON OBJECTIONS RAISED ABOUT RANDOMIZATION INCLUDE THE ETHICAL PROBLEM OF WITHHOLDING TREATMENT FROM THE CONTROL GROUP, THE PROBLEM OF CONVENIENCE, AND THE QUESTION OF COST. BUT THE GREATEST RESISTANCE STEMS FROM A MORE GENERAL FEAR AND RESISTANCE OF EVALUATION ON THE PART OF ADMINISTRATORS AND STAFF; EVALUATION MAKES THEM FEEL THREATENED. THE FACT IS, HOWEVER, THAT A RANDOMIZED DESIGN REVEALS THE TRUE POSITIVE EFFECTS OF EXPERIMENTAL TREATMENT. A POSITIVE COUNTERARGUMENT TO ADMINISTRATORS' FEARS OF NEGATIVE EVALUATION RESULTS IS THAT NOT USING RANDOMIZATION AND A TRUE EXPERIMENTAL DESIGN IS LIKELY TO HAMPER FINDING DEMONSTRABLE POSITIVE RESULTS. THIS ARGUMENT IS ILLUSTRATED IN ONE SPECIFIC CASE IN WHICH THE USE OF A SINGLE-GROUP DESIGN EVALUATION WOULD HAVE LED TO THE CONCLUSION THAT THE PROGRAM WAS INEFFECTIVE; ONLY THROUGH USE OF A RANDOMLY DETERMINED CONTROL GROUP WERE THE PROGRAM'S POSITIVE EFFECTS OBVIOUS. NOTES, REFERENCES, AND FIGURES ARE INCLUDED IN THE ARTICLE.