Statistical procedures of the regression discontinuity and regression projection designs were compared to determine which design was more able to detect treatment effect in remedial or gifted programs. Neither test was more precise in all situations. Tables list the more precise test usder specific conditions. The regression projection model produced a more precise test when the sample sizes of the two groups were equal or nearly equal. The regression discontinuity model was more precise when the sample size of the treatment group was much smaller than that of the comparison group. Both models led to erroneous conclusions about program impact when a treatment-selection variable interaction existed. For this situation, a simultaneous region procedure based on Scheffe intervals is suggested. This provides a test of the treatment effect at all levels of the selection variable, allowing the user to infer for which individuals--none, some, or all--an effect exists. A derivation of the simultaneous region procedure and a practical example demonstrating its use are presented. (Author/RL)