Is public opinion polling accurate?  The answer to that question is driven in large part by whether a specific subset of polling – i.e., preelection “horse race” polling – is perceived as accurate. On that front, the polling industry has hit some bumps recently, particularly with the 2020 election. An evaluation of preelection polls by the American Association for Public Opinion Research could not identify a clear culprit behind that year’s polling miss. The most likely explanation is differential nonresponse – i.e., that a cohort of Donald Trump supporters did not participate in the polls out of mistrust in political institutions.

Was this suspected nonresponse problem also behind the polling miss in the New Jersey gubernatorial election the following year? It turns out – via post-election voter verification – that the underlying sample in the New Jersey polling was representative of the full voter population. The real culprit was an unusually large partisan skew in who actually turned out – a skew that was not captured by likely voter models. This skew was not present in the Virginia electorate that year, which is why the polls in that state were better barometers of the actual outcome there.

Even though the 2021 analysis bolsters our confidence that polling methodology remains sound, the specter of 2020 continues to haunt pollsters.  Was that year’s nonresponse problem a one-off or will it become a recurring pattern? And if the latter, is there any way we can anticipate it and adjust our methods?

A post-election analysis by Monmouth of our 2020 polling verified the turnout of voters we polled as well as those we unsuccessfully attempted to poll. The purpose was to search for patterns in the demographics of those who did not respond to the poll versus those who did. We did not find a “silver bullet” that revealed partisan nonresponse bias. We acknowledge, however, that other researchers may be able to glean more insight.

In the interest of transparency, we are making the data from our final 2020 and 2021 polls available to the polling community. The link to those data sets, and our initial findings, can be found on our blog: If the Polls Ain’t Broke, Don’t Fix Them.