The norm-referenced evaluation model (RMC Model A) for Title I project evaluation, consists of procedures whereby the expected posttest standing of a treatment group under the null condition is generated from their pretest standing. It is assumed that the treatment group is not selected on the basis of their pretest scores and can be considered representative of the population represented in the test norms. Given these assumptions, the Model A rule for estimating the expected no-treatment posttest status is that a comparison group given no special treatment will maintain its percentile across time (the equipercentile growth assumption). This paper provides a theoretical analysis of the no-treatment expectation estimation procedures for Model A. An alternative definition of the no-treatment expectation is based on a structural equation model and statistical models underlying nonequivalent control group designs. Results indicate that using a measure other than the pretest score for selection reduces but does not control regression effects. If the selection measure is more highly correlated with the pretest measure than with the posttest measure, additional regression occurs between the pretest and the posttest. Biased measures of treatment effects may results. (Author/CTM)