We try to fix election polling (update)

Planet Money examines election polling’s participation problem: people are not answering calls or agreeing to speak with pollsters, putting confidence in polling under pressure. The team visits Marist College, home of the Marist Poll, learns how polling works and conducts calls itself. Its central experiment is to take an idea from economics and test whether it can help polling. The show notes do not identify the hosts, guests or the specific economic idea.

The episode combines polling history, current polling science and hands-on experience with the techniques pollsters use to keep respondents engaged through a complete survey. Working with a polling firm gives the team a setting in which to test its proposed solution rather than merely discuss it. However, the notes provide no sample sizes, response rates, experimental design or findings, so they do not establish how effective the intervention is or whether any improvement in participation translates into better election estimates.

Originally published before the 2022 midterm elections, the episode now revisits the experiment while Marist continues testing the idea. The update explores what the work might reveal about the 2026 midterms, connecting an ongoing operational challenge to the next election cycle. The unresolved question is whether the approach can meaningfully strengthen polling; the notes establish continued testing, not a demonstrated fix. They also leave open how broadly the findings might apply beyond the setting in which Marist is testing them.