01-26-2013 12:27 PM
When using PROC MIANALYZE, I am encountering this error:
ERROR: Within-imputation Estimate missing for effect ... in _Imputation_= 1 in
the input PARMS= data set.
The PROC MIXED procedure I am using is running fine across all imputed data sets so I'm stumped. The effect in question is a 3-way interaction term with 1 of the IVs being a dummy-coded variable.
Thanks in advance for your input!
01-02-2015 04:11 PM
I am also getting similar from proc glimmix imputation model.
ERROR: Within-imputation StdErr missing for variable in _Imputation_= 1 in the input PARMS= data
Did you figure it out yet?
01-07-2015 04:42 PM
I am also receiving this error (though with _imputation_ = 8) when running MIANALYZE. I am not running any interactions. I would be very appreciative if you could share any resolution you found or if anyone else has any other suggestions. Thank you!
07-26-2017 11:59 AM - edited 07-26-2017 01:59 PM
You could get this warning when you have more estimates in the input file than expecated by proc mianalyze. This is likely to happen if you want to get the the differences between estimates from a model with interaction term. In a standard output from analysis proc (e.g. proc glm or mixed) you could then get a lot of comparisons but most likely you wouldn't be interested in all of them. For example you could get the LS Means Difference for say levels 1 vs 2 and 2 vs 1 but you would need only one. Please try and filter the input dataset (the one from analysis procedure) for the unique comparisons only - as defined by variables in your 'class' and 'modeleffects' statements - before reading into proc mianalyze and that should allow to get the summary out of proc mianalyze and you should get rid of the warning.
If you were interested in summarizing all comparisons you would most likely need to sort the input dataset and use 'by' statment within proc mianalyze.