Use goodness-of-fit tests to determine if there is evidence of poor fit (calibration) in logistic regression models.These tests are best used as indicators of fit problems rather than accept/reject criteria.