The polling industry in recent years has been under significant scrutiny dating back to the 2016 election. In 2016 and 2020, pollsters did not accurately account for the level of turnout by irregular voters in support of Donald Trump. During the 2018 midterm cycle, polling was an early indicator of what turned out to be a Democratic blue “wave” election. So we decided to find out: how did polling perform in the 2022 midterm election in NC?
US Senate
Our analysis of polling from major, reputable firms (including many released within thirty days of the election) shows a very accurate forecast of the actual statewide election results between Ted Budd and Cheri Beasley. The two best performing polls during this timeframe – Emerson College and East Carolina University – each showed Ted Budd performing at or over 50%, with Cheri Beasley performing at or within one point of her final vote share percentage. When averaged, the thirteen polls released in the final 30 days show a final vote share difference between both candidates of 3.85%. By comparison, the final voting outcome difference between both candidates was 3.63%. Polling for this race over the final thirty days was remarkably accurate given the low response rate experienced by pollsters.
US Congress and State Legislature
In midterm elections, pollsters rely on generic ballot questions to gauge party preference in congressional and state legislative races. Congressional and state legislative races rarely receive the same level of media attention that statewide offices do, so the generic ballot question attempts to quantify how voters will select lesser known candidates. In reviewing the twelve publicly released polls covering the September, October, and November timeframe, polling was fairly accurate in the generic congressional and state legislative races.
For this data, we used polling data from Cygnal, via the Art Pope-John Locke Foundation (September/October); Meredith College (September/October); ECU (September/October/November); High Point University (September); and Public Policy Polling, via Carolina Forward (October).
The polling average for the generic ballot on congressional preference was 46.88% for Republicans, with 44.25% for Democrats. The two most competitive congressional races in NC, the 13th and 1st districts, performed within the margin of error of generic polling on Election Day. Both of those races were districts President Biden carried in 2020 and Democrats won again in 2022. Republicans likely over-performed the margin of error in some of the heavily Republican leaning districts, following on the coattails of Ted Budd.
The polling average for the generic state legislative races over the final three months was 45.5% for Republicans, with 44.75% for Democrats. These numbers will need further scrutiny once all precinct-level data is available from the State Board of Elections. We saw Democratic over-performance in many suburban districts, and significant underperformance in many rural districts. There appears to have been a blue “wave” in Wake and Mecklenburg counties, but a corresponding (and smaller) “red wave” along and east of I-95.
The clear conclusion is that polling in North Carolina is still robust. The biggest challenge pollsters now struggle with is how to properly sample, and thus predict, irregular voters. There is some evidence that pollsters undercounted Gen-Z voters in several swing states, which led to a “red” mirage in polling leading up to the election. This is one reason it’s important to follow polling averages over a period of time to gauge their overall direction, since all polls are only a snapshot in time. Polling remains an invaluable, if imperfect, tool for gauging public sentiment – and it seems the pollsters are good at what they do.
