Affiliation:
1. Department of Political Science, Columbia University, USA
Abstract
Survey researchers are sometimes tasked with sifting through possible ways of phrasing a question in order to discover the wording that maximizes the proportion of respondents who answer in a particular way. One example occurs when a political interest group seeks to phrase a ballot measure on a given topic in a way that will garner as much voter support as possible. The present study illustrates how adaptive sample allocation can be used quickly and efficiently for this purpose. Using the example of a ballot measure on ranked choice voting, we show that adaptive design was better suited to identify the most promising ballot wording than a more conventional static design. We provide an open-source software R package and instructions for survey researchers seeking to implement adaptive experiments.
Reference23 articles.
1. Athey S, Byambadalai U, Hadad V, et al. (2022) Contextual bandits in a survey experiment on charitable giving: within-experiment outcomes versus policy learning. ArXiv Preprint arXiv:2211.12004.
2. Status Quo Bias in Ballot Wording
3. A Single-Sample Multiple Decision Procedure for Ranking Means of Normal Populations with known Variances
4. p-Hacking and False Discovery in A/B Testing