Regression Model

[fhaquassho]

wefwefwfweff

[Dason]

My opinion:
What I'm thinking is that I have to make a model comparing everything to my dependent variable attendance. Then, seeing if there's a relationship between the variables?

Wait... you want to fit a model where ALL of the possible predictors are in the model? I'm thinking you would end up with a SEVERLY overfit model that won't do a good job. You might want to look into different model selection procedures. Here's a place to start: http://en.wikipedia.org/wiki/Model_selection

If I'm misunderstanding you just let me know but it's not a good idea to throw ALL of the variables in there because the model will most likely not be satisfactory.

[fhaquassho]

I figured i'd have to do them all individually, then choose the ones which are related and have a high attendance level? I don't know though, do you have any other suggestions?

[Dason]

Well my suggestion was to look into model selection procedures. I personally prefer a stepwise algorithm over forward or backward selection if you're going to do that. But if whatever software you use supports a "fit all subsets" procedure then by all means try that.

You don't need to fit all the models by hand yourself. Whatever software you're using should be able to do these procedures for you. After that it's up to you to verify the model assumptions and run the diagnostics.