Adjusting for unmeasured confounding due to either of two crossed factors with a logistic regression model.

Adjusting for unmeasured confounding due to either of two crossed factors with a logistic regression model.