Dealing with missing values is tricky. It depends on the nature of the process that produces the missing values.
Proc mixed is a fair choice and will probably give a similar outcome to EM imputation in SPSS (as the make the same assumptions about missingness, ISTR). These will be suitable when missingness is correlated with some of the predictors in the model. Repeated measures ANOVA _may_ still be OK (dropping cases with missing values) if there are very few such cases (proportionately) or if missingness is determined by random factors uncorrelated with your predictors (such as hardware failure). Thom "Andreas K." wrote: > > Don't use repeated measures ANOVA, it's not a good choice. Use Proc > Mixed in SAS instead. > > On Thu, 16 May 2002 11:38:17 +0200, "Wouter Duyck" > <[EMAIL PROTECTED]> wrote: > > >Does anubody know a good method to calculate estimates for missing cells in > >repeated measures ANOVA designs which incorporates both row (case) and > >column (variables) means? . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
