This paper studies the identification of coefficients in generalized linear predictors where the outcome variable suffers from non-classical measurement errors. Combining a mixture model of data errors with the bounding procedure proposed by Stoye (2007), I derive bounds on the coefficient...