Poll Position
American’s view of polls in the current era
Political pundits love polls, but if the 2020 election proved anything, it’s a reminder that even the best polling is not perfect. Case in point: the popular presidential vote. According to polling site FiveThirtyEight’s average of polls, Joe Biden was projected to win the national vote by a margin of 8.5 points. Once all the votes are counted, though, he’s expected to finish at about 4.7 points ahead, a polling error of about 3.8 points. This follows the trend of, on average, 3.7 point errors in recent elections, with pollsters overestimating the Democratic share in five of the eight presidential elections from 1988 to 2016. State level polling in swing states this year also underestimated Republican vote share, though not enough for Trump to win an upset victory in the electoral college.
Fat Finger also predicted a national result that was off by slightly more than our stated +/- 3 pt error. Our final election poll had Biden winning the popular vote by 8 points. Again, the final actual margin is likely to be about 4.7 pts, for a Fat Finger miss of 3.3 pts, or 0.3 outside our 95% confidence level.
Now comes the polling post-mortems. Pundits are likely to suggest there are flaws in how modern surveys are conducted and who is getting polled, and people are more cynical and distrustful of polls than ever — though not with good reason given 2016 and 2020 polls were in line with historic norms in terms of magnitude of misses. However, the misses these days are magnified as the electorate is so evenly divided that even a miss of small amounts can mean predicting the wrong winner, which was rarely the case in the past. For more insight, we asked Americans to share their views on how the polls turned out, and tell us who they would vote for if the presidential election was held today.
Here’s what we learned:
The Partisan Gap
Despite continued court battles from Trump’s legal team seeking to overturn the result of the election in several states that flipped for Biden — mainly tossing out mail-in ballots — a majority (56%/42%) of Americans trust that the final vote counts are accurate. When broken down my party affiliation, though, the trust divide is stark. Nearly every (94%) Democrat respondents said they believe the results while only 13% of Republicans agreed. As for the all-important Independents, a majority (63%/38%) sided with Dems this time that the counts are correct.
Trust Me, I’m a Pollster
Hindsight truly is 2020. With the final vote counts mostly in, it’s clear in retrospect that all the stock put in polls leading up to the election was a tad misguided. Americans agree, with a majority (55%) giving both the popular vote and state polls a failing grade. Fat Finger fared about the same with 54% saying we missed the mark. A plurality (48%) of Dems did defend the popular vote polls by FiveThirtyEight and Fat Finger, but a strong majority of Independents (56%/38%) and Republicans (72%/7%) were not impressed.
Different Time, Same Result
Even with all the talk of bad polling and claims from the Trump team of voter fraud, if the election was redone today, Americans would still give Biden a clear victory over Trump (53%/40%). With Dem and Republican respondents firmly in their camps — less than 5% of each affiliation voted for the candidate of the opposing party — it comes down to Independents who say they would back Biden over Trump by a 12-point margin.
Related Polls:
Fat Finger had Clinton +11 in the national vote. The Real Clear Politics average had her +3.2, the actual national vote was Clinton +2.1. After the election, we asked Americans what they thought contributed to FF being so removed from the actual result. Nearly half of respondents said sample sizes were too small, closely followed by “not enough rural residents in the sample” (43%) and “not enough Republicans in the sample” (41%).
In June, we asked Fat Finger panelists why they thought the state poll averages consistently underestimated Trump’s results in key states. Almost half chalked it up to Trump supporters being disinclined to tell pollsters their true intentions. Another 21% blamed pollsters for intentionally putting out false results
Just before election day, we asked Americans what they thought of the accuracy of several 2020 presidential election polling averages. A plurality (46%) answered that the +7.8% for Biden popular vote averages would overstate the final margin for Biden, a prediction that turned out to be spot on. Respondents also gave Biden the edge in who they planned to vote for by an eight point margin.
by Landry Harlan
