I have an SPSS file with longitudinal data, stretching back 18 months or so (monthly survey with around n=1000), so the survey month is one of the variables but it's all in the same data set.

Now, I've been asked to see if I can run a regression using survey data and actual market share for the category. We have a few questions pertaining to "favorite brand" and "intended next purchase", so we have frequencies for those questions for every month. And we have official/publicized monthly market share data.

I just can't figure out if I should include the market share data as an additional variable in the current data set, or if I need to create a whole new data set, using monthly data for each of the variables (both dependent and independent) as individual cases. Of course then you won't have sample sizes and distribution etc from the original data, so it's an "estimate of an estimate" which doesn't sound too good...

Any good suggestions? Would appreciate any help I can get!

T.A.