It sounds like you might want to use multiple regression (logistic regression if your outcome is binary; 0 = not in remediation and 1 = in remediation, which can be used to identify the "best" predictors of a given outcome. BUT - and this is a big BUT - I would not go about running this (or any kind of analysis for that matter) without an adequate knowledge of what you are doing (e.g., the assumptions of the analyses, how to interpret the results, etc. etc.)

I could think of some ways that looking for moderators and mediators might be appropriate here, but w/out any kind of guiding theory, I would shy away from it. Just stick with a basic direct effects model - i.e., What predicts whether or not someone needs remediation.

Logistic regression sounds good.

You have some other choices though, depending on the theory. It is always best to think out the psychology before you get to the stats.

Cluster analysis is one option (see SPSS). Think in terms like a biologist classifying fish. You would catch them and write down all their characteristics in columns. Then you would sort the fish into groups based on the what they had in common. (Note most psychology would sort the columns/variables. This sorts the rows/cases/people). This would produce groups that may have nothing to do with remediation! But may be interesting anyway.

You could do some basic cross-tabs. Or look at indexes. How much did behavior change over time. Sometimes exploring data with percentages tells you a lot. I've discovered a lot about how behavior occurs in particularly settings in this way and have often blow myths wide open.

There is another technique and in the minute I cannot remember its name. Imagine you have successful and unsuccessful farmers and you wanted to see if there was anything distinguishing the two groups. You would test a set of factors. The method will be in SPSS - sorry don't have a manual here.

But do begin with the psychology. Don't be afraid to address myths. And do inspect you data first with graphs and percentages. They go a long way. Psychologists tend to obscure reality with complicated models. See what basic analysis tells you first!

PS You could have given us more information so we could help you solve a puzzle rather than launch into general theory - and stats theory at that. But better than the questions I have asked so far!

Look forward to the next one!