Using SAS® Software to Check Assumptions for Analysis

Analysis of covariance (ANCOVA) is a powerful statistical tool for adjusting an analysis to acoount for the effects of concomitant variables. The technique may be applied to completely randomized designs (CRO) as well as repeated measures designs. In addition to normality assumptions, ANCOVA depends on assumptions about variances and slopes. Although not usually provided directly, most statistical software has the capability of testing the equal slopes assumption in a CRO ANCOVA; however, the slope assumptions of repeated measures ANCOVA are more complex and more difficult to check. The SYSLIN procedure in SAS/ETS® software is an elegant tool to assist with checking the slope assumptions of repeated measures ANCOVA. The assumptions of ANCOVA are discussed and SAS programs for checking these assumptions are presented. A method for comparing groups when the equal slopes assumption is violated is also presented.