A general procedure is presented for generating code values for a qualitative variable in multiple linear regression analyses that result in directly interpretable estimates of interest. The basic approach, in viewing ANOVA as a multiple regression problem, is to derive quantitative code values for the various levels of the qualitative ANOVA factors. The effect of different coding schemes is illustrated by a simple two-group problem. These notions are then extended, with accompanying directions for code value choices, to the K-group ANOVA problem, the analysis of covariance, and the aptitude-by-treatment interaction (ATI) situation. The ANCOVA model differs from the ANOVA model only in that one or more quantitative covariates are added to the regression equation. In ATI studies the estimated parameters may be directly interpreted as differences in slope values between groups by using these methods. (Author/CTM)