This report is a retrospective account of a single research project conducted between 1973 and 1976 which involved a field study of the administrative feasibility of vouchers for skill training in the Work Incentive Program (WIN) in Portland, Oregon. (The program was designed to change relationships among clients, WIN staff, and training vendors, by transferring responsibility for decision making and negotiations for training arrangements from staff to clients--nondirective counseling.) Focus in this report is on how evaluation research or feasibility testing (using the implementation of the WIN voucher system as an example) can be effected by staff noncooperation, problems of intra- and interagency coordination, and other variables (e.g., goal conflict, communication). A sixteen-page concluding section discusses literature concerned with program evaluation and describes elements which might go into a systematic approach to program implementation analysis. Suggestions for ways in which a two-level approach to evaluation research might increase the clarity, comprehensiveness, and sophistication of research results are also included. (SH)