The authors develop a method to correct for nonrandom measurement error in a binary indicator of illicit drugs. Their results suggest that estimates of the effect of self-reported prenatal drug use on birth weight are biased upwards by measurement error--a finding contrary to predictions of a...