On Thursday, we launched our updated pollster ratings here at FiveThirtyEight. While they showed that polls have not become markedly less accurate in recent years, they did have a pretty bad 2020. Our analysis also found that a longtime truism in polling — that surveys using live callers are more accurate — is no longer true. Comparing live-caller polls with online surveys, text messaging, automated calls and mixed methods, the former is not systematically likelier to reflect the final result of an election.
In this installment of the FiveThirtyEight Politics podcast, editor in chief Nate Silver talks to Galen Druke about why the gold standard of polling has changed and what this means going forward. With the benefit of hindsight (and updated pollster ratings), they also assess how polls performed in 2019 and 2020 in general. Read more