Missing Data/Imputation Discussion > MI and questionnaire subscales
Gina,
I prefer to allow SPSS to do my pooling, as you can get results from both the "original" data and the "imputed" data, using that method.
With that said, I don't see any reason that pooling it by hand will cause a problem, it is just more work. On a side note, I typically prefer calculating subscales with the data that you have, and then imputing data on the subscale level, as opposed to at the item level, but just a suggestion.
May 30, 2012 |
Jeremy Taylor
Jeremy Taylor
Hello
I am new to SPSS missing data (v20) and have just conducted MI on raw questionnaire data. However, I am really interested in the results of anaylses on subscales calucated from this raw data.
I would like to export the imputed data to excel, which I have set up to calculate various subscales from my raw questionnaire data. If I use my imputed data to calcuate 5 (the number of imputations I have done) versions of each subscale and then do 5 analyses in SPSS (one for each imputed dataset) and pool the results by hand will this work/be methodlogically sound?
Any help is grately appreciated.
Gina