Good evening,
During the past few months, I have been working with the SPSS/PASW Statistics 18 software and more recently the Complex Samples component. I am using the Medical Expenditure Panel Survey (MEPS) for analyses, specifically a pooled dataset from 2002-2007. I completed the weighting for the pooled dataset with the SPSS Complex Samples during the past few days, and have run the initial frequencies. However, when I added the PERWTF, VARSTR and VARPSU variables, my original sample size decreased from 75,532 to 71,692 for the weighted dataset (a loss of 3840 participants).
I can find no reasonable explanation for this loss in cases, and would like to ask if this has happened with other researchers and with other SPSS program users. When I reviewed the unweighted counts for each variable, most variables unweighted counts decreased by a few numbers from the original dataset, with the exception of the values I coded as missing (99) where many stayed the same. What have other researchers used or experienced that has been similar to this and how was it corrected?
Regards,
K. M. Lida
Locked
