Beth, I don't remember much about SPSS from school, but I do use SAS every day, and it is 'normal' for stats procedures to run on a de-normalized dataset, where each observation (row) contains variables (columns) that would be stored in many related tables in a relational database. There are really two important steps in doing a statistical analysis: 1) create the dataset you need, and 2) perform the stats procedures on that dataset. I have found that some statisticians do not get involved with creating the dataset, they only put in a request to someone able to pull the data to create the dataset and make it available for analysis. However, there are many stats folks able to do both. Any stats package (including SPSS) has the ability to merge several normalized datasets into one 'flat' dataset for analysis. I would suggest you find a SPSS programmer able to do this, so you would only have to worry about importing the normalized datasets, then do the merging in SPSS. Just my two cents.