Inferring health interview response bias from imperfect record checks

Results of incomplete criterion validity checks can be misinterpreted. The kind of design incompleteness potentially determines the sign of the response bias estimate. The amount of random error can determine the magnitude of the bias estimate. Publi...