An empirical study was conducted in order to obtain treatment effect estimates with the Special Regression model for groups in which there was no treatment. General mathematics test scores were obtained from 730 ninth graders in city schools somewhat similar to Title I schools, but in which no special treatments were given. Hypothetical experimental and control groups were formed on the basis of pretest scores and the data were analyzed according to the RMC Model C (special regression model). Conclusions were that: (1) even minor ceiling or floor effects resulted in nonhomogeneous regression lines for treatment and comparison groups; (2) tests of statistical significance should preceed conclusions of educational significance at the local school or district level; and (3) guidelines to assist with interpreting evaluation results are needed. (Author/CTM)