MANY COVERT AND OVERT PURPOSES FOR MEASURING CHANGE CAN BE GIVEN FOR AN EVALUATION RESEARCH PROJECT, WITH SERIOUS IMPLICATIONS FOR THE MEASURES USED, THE RESULTS OBTAINED, AND THE UTILITY OF THE INFORMATION FOR MODIFYING THE SET OF INTERVENTION ACTIVITIES. THE ASSESSMENT OF PROGRAM EFFORT, PROGRAM EFFECT, AND PROGRAM EFFICIENCY ARE THE THREE MAJOR FORMS WHICH AN EVALUATION MAY TAKE. SOME OF THE COMMON PROBLEMS IN MEASURING CHANGE EFFECTED BY SOCIAL PROGRAMS ARE DISCUSSED. OFTEN THE PERIOD OF INTERVENTION OF A PROGRAM IS TOO SHORT TO ALLOW FOR CHANGE, AND A MIXED COLLECTION OF PROGRAM COMPONENTS PRODUCES TOO MANY VARIABLES FOR CALCULATING CHANGE. ADMINISTRATIVE REQUIREMENTS FOR FLEXIBILITY IN MODIFYING THE PROGRAM IN RESPONSE TO THE CHANGING NATURE OF STAFF OR DIFFERENT TREATMENT PREFERENCES OF STAFF USUALLY INTERFERE WITH MEASURING CHANGE IN THE PROGRAM. AN INADEQUACY IN THE MODEL AND METHODLOGY OFTEN EXISTS SINCE A MODEL OF CAUSATION COMMONLY APPLIED ASSUMES A LINEAR RELATIONSHIP BETWEEN EXPOSURE AND OUTCOMES. THE OCCURRENCE OF REACTIVE EFFORTS, PARTICULARLY ASSOCIATED WITH INTERVIEWING OR QUESTIONNARE METHODS, CAN ALSO CAUSE PROBLEMS OF INTERNAL VALIDITY. FOR EXAMPLE, DIFFICULTIES IN SPECIFYING A CLEAR, MEANINGFUL, AND MEASUREABLE GOAL CAUSE PROBLEMS IN IDENTIFYING PROGRESS TOWARDS THE GOAL. RELIABILITY AND VALIDITY OF MEASURES IS A MAJOR CONCERN IN DEFINING SUCCESS IN A PROGRAM. THIS IS ESPECIALLY TRUE IN THE CASE OF CRIMINAL JUSTICE EVALUATION IN WHICH RECIDIVISM IS A COMMON MEASUREMENT OF CHANGE. THE CONCEPT OF RECIDIVISM, HOWEVER, IS AMBIGUOUS AND IS USED TO DEFINE MANY DIFFERENT OCCURRENCES, SUCH AS REARRESTS, RECONVICTIONS, AND REIMPRISONMENT. THE WIDESPREAD DISCRETION EXERCISED BY THE POLICE TO ARREST IS A FURTHER SOURCE OF INVALIDITY. IN SEPTEMBER 1971, THE MINNESOTA RESTITUTION CENTER WAS DEVELOPED AS A COMMUNITY-BASED RESIDENTIAL CORRECTIONS PROGRAM FOR ADULT MALES. AN EVALUATION RESEARCH DESIGN FOR THE PROGRAM DURING THE FIRST 24 MONTHS OF OPERATION TOOK THE FORM OF A BEFORE-AFTER EXPERIMENT WITH THE RANDOM ASSIGNMENT OF MEN FROM WITHIN A SPECIFIED POPULATION OF RECENT PRISON ADMISSIONS. THE PROBLEMS DEFINED IN MEASURING CHANGE PRODUCED BY A SOCIAL PROGRAM ARE APPLIED TO THIS EVALUATION PROJECT, AND THE RESULTS ARE DISCUSSED. GENERALLY, MEASUREMENT PROBLEMS APPEAR TO FOLLOW FROM A LACK OF CONGRUENCE BETWEEN THE PURPOSE OF THE EVALUATION AND THE DEVELOPMENTAL LIFE OF THE PROGRAM, ESPECIALLY IN RELATION TO THE NATURE OF THE PROGRAM INTERVENTIONS, THE ARTICULATED GOAL STATEMENTS, AND THE VALIDITY AND RELIABIITY OF MEASURES. (DAG)