Evaluators often utilize analysis of covariance (ANCOVA) techniques to compensate statistically for the lack of experimental control when assessing the effects of innovative programs implemented in naturalistic settings. In this paper design, analysis, and reporting considerations important to the application of ANCOVA-type techniques in educational settings are described. Problems that arise from the use of complex data analysis techniques are identified, based on a review and critique of the evaluation of the national Follow Through program. Specific information that should be included in an evaluation report when ANCOVA-type techniques are used is described. Examples of the kinds of problems that appear when collecting data in school settings are provided in order to illustrate the need for this information. Alternative ways of presenting the needed information in an evaluation report are discussed. The overall perspective is that evaluation reports must be more precise and indicate the limitations as well as strengths of the methodology used for the specific setting. (Author/APM)