Hi there,
I have conducted multiple imputation using PASW, and I have generated 10 imputed datasets. When I conduct analyses to compare 2 groups on some variable, I have decided to use the pooled t-statistic (indicating the difference between the 2 groups) that PASW output provides. Is this practice considered acceptable? I understand that F-statistics are more "sound" than t-statistics, but it is highly convenient that PASW produces the t-stat information, and I have limited time before I submit my results.
With all of that said, I have some repeated measures ANOVAs to conduct with my imputed datasets. I realize that I have no choice but to pool the F-statistic for this type of analysis. I have looked at Dr. Ginkel's macro syntax, which seems a bit involved given my current timeline. Another suggestion made by Dr. Ginkel was to "use a multi-level model in mixed models where repeated measures are nested within subjects". This seems like an attractive option, however I am not entirely sure how to conduct said analysis. Any guidance would be appreciated. FYI, I have pre-, post- and follow-up measurements for a particular variable, and I want to see how this variable changes across the 3 time waves. Many thanks in advance.
Best,
Teresa
Teresa Grimbos, PhD candidate
Developmental Psychology & Education Program
OISE/University of Toronto
Locked
