Purposeful selection of variables in logistic regression.

As the sample gets larger, the variability of even borderline significant confounders gets smaller, and they get retained as significant variables, hence diminishing the retention differences between the selection methods.

also in 11,409 other papers