Abstract

Overestimation of turnout has long been an issue in election surveys,  with nonresponse bias or voter overrepresentation  identified as major sources of bias.However, adjusting for nonignorable nonresponse bias is substantially challenging.Based on the ANES Non-Response Follow-Up study concerning the 2020 U.S. presidential election, we investigate the role of callback data,  i.e.,  records of contact attempts in the survey course, in adjusting for nonresponse bias  in the estimation of turnout. We propose a stableness of resistance assumption to account for nonignorable missingness in the outcome, which states that the impact of the missing outcome on the response propensity is stable in the first two call attempts. Under this  assumption and by integrating with covariates information from the census data,  we establish  identifiability  and develop estimation    methods  for  turnout.Our methods produce  estimates very close to the official turnout and successfully capture the trend of declining willingness to vote as response reluctance  increases. This work highlights the importance of adjusting for nonignorable nonresponse bias and demonstrates the potential of widely available callback data for political surveys.