A comparison of recent approaches to the analysis of repeated measurements

Looney & Stanley's (1989) recommendations regarding analysis strategies for repeated measures designs containing between-subjects grouping variables a nd within-subjects repeated measures variables were re-examined and compare d to recent analysis strategies. That is, corrected degrees of freedom univ ariate tests, multivariate tests, mixed model tests, and tests due to Kesel man, Carriere & Lix (1993) and to Algina (1994), Huynh (1978) and Lecoutre (1991) were compared for rates of Type I error in unbalanced non-spherical repeated measures designs having varied covariance structures and no missin g data on the within-subjects variable. Heterogeneous within-subjects and h eterogeneous within- and between-subjects structures were investigated alon g with multivariate non-normality. Results indicated that the tests due to Keselman et al. and Algina, Huynh and Lecoutre provided effective Type I er ror control whereas the default mixed model approach computed with PROC MIX ED (SAS Institute, 1995) generally did not. Based on power differences, we recommend that applied researchers adopt the Welch-James type test describe d by Keselman et al.