The report examines the potential of Social Security Administration data for manpower training evaluation. It assesses the utility of SSA data for comparing the economic gains of on-the-job training and institutional trainees before and after training. The methodology developed incorporates multiple correlation analysis, analysis of variance, and F-ratios and proportions of explained variation as measures of significance. The report notes general information about the nature and limitations of these data. Some of the problems of benefit measurement are discussed and a description of the data sources is provided. (Author)