What's the best method to handle a sparse-data/separation problem in a matched case-control study with 600 cases and 600 controls?

The sparse data is from multiple interaction terms in the conditional logit model.

01 January 2014 8,677 3 View