Roberto QuerciaPI. Funded by the Research Triangle Institute (RTI), this project addresses what survey researchers, data users, and organizations that rely on survey data know well: Response rates for surveys have been on a steady decline. In response, survey researchers are currently developing new methods that address various aspects of this problem. One novel approach currently under development at UNC-Chapel Hill and RTI uses response propensities and “paradata” to predict response outcomes in longitudinal surveys in order to reduce response bias. Paradata describes administrative or process data such as call records, interviewer comments, and comments made by respondents during prior contacts. Prior research indicates that paradata is highly predictive of survey non-response. Standard practices in many survey organizations seek simply to maximize response rates by requiring survey managers to target cases that are most likely to end up as completed interviews. While this approach is intuitive, the now widely accepted stochastic model for survey participation indicates that it may not reduce the more important issue of non-response bias. Bias in survey estimates results from non-response and often compromises data quality. Thus, the standard practice of focusing on maximizing response rates is not ideal because it may not reduce bias in survey estimates. In contrast, the approach under development at UNC-Chapel Hill and RTI uses response propensities and paradata to target respondents with incentives in an attempt to reduce response bias. This novel approach may help survey managers address the critical issue of response bias for which traditional approaches to non-response appear inadequate.