Researchers encountering wrong signs on regression coefficients are inclined to blame their variable list rather than measurement error. It is generally assumed that ...