I present nonparametric methods to identify and estimate the biases associated with response errors. When applied to survey data, these methods can be used to analyze how observable and unobservable characteristics of the respondent, and characteristics of the design of the survey, affect errors in the responses. This provides a method to correct the biases that those errors generate, by using the estimated response errors to "undo" those biases. The results are useful also to design better surveys, since they point at characteristics of the design and of subpopulations of respondents that can provide identification of response errors. Several models are considered. Copyright 2007 by the Economics Department Of The University Of Pennsylvania And Osaka University Institute Of Social And Economic Research Association.
