Bayesian variable selection in sample selection models using spike-and-slab priors.

If close to (or more than) half the variables may be significant, the prior is a more sensible choice, at the cost of potentially shrinking small effects.

also in 13,097 other papers