I'm no expert on this, but my first thought would be to run a principal component analysis on the predictor variables, then use the resulting principal components to predict your dependent variable.
Mike Lawrence
- 14k
- 9
- 47
- 72