EVALUATION OF SOCIAL PROGRAMS (FROM READINGS IN EVALUATION RESEARCH, 2ND ED., 1977, BY FRANCIS G CARO SEE NCJ-50468)

• NCJ Number: 50474
• Author(s): B G GREENBERG
• Date Published: 1977
• Length: 22 pages
• Grant Number(s): 1-F3-GM37,750
• Sponsoring Agency: US Dept of Health, Education, and Welfare
• Corporate Author: Russell Sage Foundation
• Publication Format: Document
• Publication Type: Program/Project Evaluation
• Language: English
• Country: United States of America
• Note: REPRINTED FROM REVIEW OF THE INTERNATIONAL STATISTICAL INSTITUTE, V 36, N 3 (1968), P 260-277

Annotation

STATISTICAL METHODS IN PROGRAMMING, STATISTICAL DESIGN OF EVALUATIVE STUDIES, AND ASSESSMENT PROCEDURES WHEN EVALUATION IS NOT CONTROLLED ARE THE MAJOR TOPICS DISCUSSED.

Abstract

STATISTICAL METHODS IN PROGRAMMING ARE DISCUSSED UNDER THE FOLLOWING FIVE STAGES: 1) MEASUREMENT OF NEED THROUGH COMMUNITY DIAGNOSIS; 2) PROGRAM DESIGN AND SETTING OF GOALS; 3) MEASUREMENT OF SERVICE; 4) EVALUATION OF GOAL FULFILLMENT; AND 5) COST-BENEFIT ANALYSES AND OTHER INPUT-OUTPUT STUDIES. THE BASIC STATISTICAL DESIGN OF EVALUATIVE STUDIES IS CONSIDERED TO CONSIST OF THE ESTABLISHMENT OF ONE PORTION OF THE SAMPLED POPULATION AS THE EXPERIMENTAL GROUP (THOSE RECEIVING THE SERVICES OF THE EVALUATED SOCIAL PROGRAM) AND THE REMAINING PORTION AS THE CONTROL GROUP (A GROUP LIKE THE EXPERIMENTAL GROUP IN ALL RESPECTS EXCEPT SUBJECTS DO NOT RECEIVE THE SERVICES OF THE PROGRAM). DIFFERENCES IN RESPONSE BETWEEN THE TWO GROUPS ARE TESTED TO DETERMINE THEIR STATISTICAL SIGNIFICANCE BEFORE GENERALIZING THE OBSERVATIONS FOR THE LARGER POPULATION. SOME OF THE BROAD CLASSES OF STATISTICAL PROBLEMS ARISING FROM THIS DESIGN ARE DISCUSSED. IF RANDOM ASSIGNMENT IS IMPOSSIBLE IN A SPECIFIC PROGRAM EVALUATION, IT IS SUGGESTED THAT THE INVESTIGATOR TAKE EVERY PRECAUTION TO MEASURE HOW PAIRED COMMUNITIES DIFFER AND TRY TO CORRECT FOR ANY IMBALANCE. TECHNIQUES SUCH AS ANALYSIS OF COVARIANCE, AGE ADJUSTED AND STANDARDIZED RATES, AND SIMILAR TOOLS ARE CONSIDERED HELPFUL IN SUCH CIRCUMSTANCES. IN THE EVENT THAT CONCURRENT CONTROL OR COMPARISON COMMUNITIES WERE NOT INCLUDED IN THE EVALUATION DESIGN AT THE TIME SERVICES WERE IMPLEMENTED, A FREQUENTLY USED METHOD IS TO CONSIDER EACH COMMUNITY AS ITS OWN CONTROL OVER TIME. SEVERAL PAST POINTS IN TIME ARE SELECTED TO JUDGE HOW THE INTRODUCTION OF THE SOCIAL PROGRAM MIGHT HAVE AFFECTED INDICATED TRENDS ASCERTAINED FROM DATA CHARACTERIZING THE COMMUNITY PRIOR TO THE INTRODUCTION OF THE PROGRAM. ANOTHER APPROACH TO THE USE OF A COMMUNITY AS ITS OWN CONTROL IS TO CLASSIFY THE SERVICES RENDERED ACCORDING TO CHARACTERISTICS OF ITS RECIPIENTS. THEN, CHANGES WITHIN THE STUDY COMMUNITY SHOULD BE CONCENTRATED AMONG THOSE GROUPS RECEIVING THE MOST SERVICES. AN ILLUSTRATION OF THE USE OF THIS METHOD IS PROVIDED. (RCB)