I started to use one potential workaround for "unit-wise" missing data, but
it begs the obvious question asked below.
About the data set:
2 x 7 x 4 IVs (= 56conditions) x 14 subjects (= 784 cells) x multiple
responses per subject per condition (6228 responses total).
The 18-20 DVs mentioned previously are probably only 10 or 12 variables to
be tested in repeated measures ANOVAs. Still, one issue was that in the
six missing cells, data of all dependent variables are missing. This is
essentially "unit-wise" missing data, isn't it? I arranged three IVs, the
subject number, and one DV into five columns; then, I ran Amelia to impute
six missing datum x 5. Is there a way to do the imputations on all DVs at
once, and thereby avoid splitting the dataset and repeating the procedures?
Deric Wisleder
Show replies by date