When I tried to do multiple imputation, I got the message:

The imputation model for question1 contains more than 100 parameters. No missing values will be imputed. Reducing the number of effects in the imputation model, by merging sparse categories of categorical variables, changing the measurement level of ordinal variables to scale, removing two-way interactions, or specifying constraints on the roles of some variables, may resolve the problem. Alternatively increase the maximum number of parameters allowed on the MAXMODELPARAM keyword of the IMPUTE subcommand.

Execution of this command stops.

Is this because I set the question variables as ordinal? I remember someone said changing ordinal to scale may lead to bias in imputation.

any suggestions? ]]>

I'm a PhD student and for my most recent study I conducted a 4 (experimental condition) x 2 (emotion type) mixed design ANOVA with repeated measures on the last variable. I followed the steps in Andy Field's textbook and companion website, conducting and interpreting my results accordingly. My supervisor is not pleased with the way I have reported a few of my results, however and I am a bit confused as to why (I followed the book, like many a stats teacher has said!).

According to the guide on his website (http://www.statisticshell.com/docs/mixed.pdf) to explore a significant main effect I report the significance values in the pairwise comparisons table of the output. For example, on the link I have here it says "Bonferroni corrected post hoc tests showed that ratings of Santa #1 and Santa #3 did not significantly differ (p = .45), but ratings of Santa #2 were significantly higher than both Santa #1 and Santa #3 (both ps < .001)."

In the write up of my study, I reported similar results for my significant main effect, putting only the p-values. My supervisor said that I was not reporting this correct and needed to report the values along with the p-value. I would be more than happy to do this, but the pairwise comparison table only has the mean difference, standard error, and confidence intervals with the p-value.

I would really appreciate any help or suggestions anyone could give me. Should I be looking at a different table? Is there a totally separate test I should be conducting? Help! ]]>