This study reports an attempt to measure implementation of Head Start Planned Variation, a national experiment designed to compare the relative effectiveness of 12 education models. The report first gives the background of the experiment, describes the problem of attempting to assess effectiveness without first measuring implementation, and reviews research on implementation. The process of designing instruments is outlined, and 10 of the models and their instruments are described. Methodological problems are discussed, data analyses are presented, and alternative strategies for the evaluation of implementation are suggested. The final chapter contains a summary and conclusions. It is suggested that the ideal design for the study of implementation has two components: (1) an adequate description of the intervention with input from all members of the training staff as well as the sponsor; and (2) an adequate evaluation of implementation, performing primarily a formative function. Implementation checklists and figures on interobserver agreement are included in the appendices. (GO)