THE MANAGEMENT OF DATA COLLECTION REQUIRES AN INTIMATE RELATIONSHIP BETWEEN THE RESEARCH DESIGN AND THE PROGRAM BEING EVALUATED. CRITERIA FOR EFFECTIVENESS MUST BE STATED, AND A LIMITED OBJECTIVE MUST BE SPECIFIED. CHECK SYSTEMS WILL GUARANTEE THAT DATA ARE COLLECTED SYSTEMATICALLY, ACCURATELY, AND WITH A HIGH DEGREE OF COMPLETION. THESE CHECK SYSTEMS SHOULD BE INITIATED AT AN EARLY POINT IN THE PROGRAM. COST FACTORS ARE NECESSARY CONSIDERATIONS IN DESIGNING RESEARCH. INTERVIEWERS, CLERKS, AND OTHER PERSONNEL MUST HAVE SUPERVISORY COMPLETION CHECKS BUILT INTO THE SYSTEM. THE UNIT OF SAMPLING MUST BE CONSIDERED, AND IF IT IS THE AGENCY OR PROGRAM, THERE MUST BE A SUFFICIENT QUANTITY OF SUCH UNITS TO ALLOW GENERALIZATION TO A PROJECTED POPULATION. THE RESEARCH MUST DICTATE THE SAMPLING AND DATA COLLECTION PROCESS, AND IF THIS IS NOT PRACTICALLY POSSIBLE, THE RESEARCH SHOULD BE DISCOURAGED. TACTICAL PROBLEMS IN DATA COLLECTION ARE DISCUSSED, AND EXAMPLES OF SITUATIONS IN WHICH DATA COLLECTION IS DIFFICULT ARE PROVIDED. IN ANALYZING DATA, MANAGEMENT OF INCOMPLETE INFORMATION AND INFORMATION SERVING NO DEFINED PURPOSE IS A PROBLEM. THESE OCCURRENCES ARE ASSOCIATED WITH FAILINGS OF DATA COLLECTION TECHNIQUES RATHER THAN WITH THE MODEL OF ANALYSIS. DATA COLLECTED FOR WHICH PROVISION OF ANALYSIS HAS NOT BEEN ANTICIPATED IS AN ADDITIONAL DIFFICULTY. ANALYSTS OF DATA MUST DECIDE HOW DATA WILL BE TREATED, HOW SECONDARY ANALYSIS WILL BE CONDUCTED, AND HOW TO INTERPRET DATA WITHIN LIMITATIONS. (DAG)