Paper Title
Nonresponse Propensity Weighting With Unknown Eligibility

Abstract
Propensity weighting, which is based on the estimation of the response propensity score, has been widely used in many surveys, to address survey nonresponse, especially in longitudinal surveys. The idea behind this weighting is to inflate the weights of responded sampled units so that these units can be used to represent both responded and non-responding units which are eligible for the survey. Sampled units that are ineligible should not be included in the weighting. However, in many surveys, the eligibility of some sampled units is unknown and it is not clear in the literature how these units should be treated in the weighting. Should they be included or excluded entirely in the weighting? This study proposes a method to include them in the weighting based on discounted factors which are the probabilities of eligibility. A simulation is used to show that this method is preferable to other alternative treatments of these units in the weighting. Keywords: propensity weighting, nonresponse, survey.