A confession: in the runup to the 2020 election, I adopted the practice of adding two points to all of Donald Trump’s polling numbers and subtracting two points from all of Joe Biden’s polling numbers. As an example, if I read that Biden was leading a state by one point, (48–47), I flipped that in my mind to Trump leading by three points, (49–46). Two more points for Trump; two fewer points for Biden. There were two reasons for this:
1 — The ashamed Trump voter. I firmly believe that there are many voters in this country who are ashamed of their support for Donald Trump but they cast a ballot for him each chance they get because...they like the guy.
In polls they may say they are voting for “the other guy” or they are undecided, but when they have that ballot in hand, voting for Trump becomes the guilty pleasure that they know is wrong but makes them feel so good.
They could care less that in five years he’s built no wall, delivered no healthcare plan, told 200,000 lies, caged women and children, lost 545 sets of parents, failed to manage a once-in-a-century pandemic, been impeached for high crimes, used the presidency as a cash cow, promoted White supremacy, disrespected the military, fired (or worse) his political enemies, withheld his tax returns, cozied up to dictators, fomented violence and division, and so much more.
All they care about is how he makes them feel. He’s funny, he’s entertaining, he gets me, he sticks it to the elites, he’s one of us, he owns the libs, he understands, he’s tough, he’s macho.
And, so….they slip into the ballot box and vote for their hero, ashamedly though, so they tell pollsters that they still haven’t made up their minds the day before election day or that they are going to vote to restore unity and dignity to the office of the presidency.
Pollsters have not yet figured out how to account for this dissonance.
2 — Polling is subjective. As a qualitative researcher (think stories and lived experiences), I’ve endured years of (friendly) condescension from quantitative researchers (think numbers and statistics) regarding the subjectivity and the lack of extrapolate-ability of qualitative research. I happily concede this. The subjectivity of and inability to extrapolate from qualitative research is what makes it so poignant and powerful and personal and important. It is important to collect numbers, to read statistics, to analyze trends, to predict outcomes. It is also important to hear and tell unique stories, even if what an individual story tells you about the whole is limited and anecdotal.
The subjectivity of quantitative research is baked into the inputs. How the question is asked and who answers the question are two points of subjectivity in polling. When it comes to politics, external polling seems to favor left-leaning candidates and issues, likely because the respondents “know” which answer is politically correct or socially expected and they try to provide it. External polling relies heavily on people self-reporting who they are, (e.g. evangelical or college graduate) and what they believe in (e.g. pro-life or pro-marriage equality). External polling over samples White people. External polling misses nuance; it misses the way that people process information from an intellectual place but then make decisions and take actions based on their “gut.”
It’s for these reasons that I adjust polls in my own head to account for what I believe polling — particularly political polling — misses.
The challenge though is that many of us rely heavily on polling — whether consciously or subconsciously — to shape our behaviors. When polls consistently suggest narratives or opinions, pundits begin framing their commentary around these narratives and opinions, then the electorate begins to act on these narratives and opinions.
If the polls say Democrats can flip the US Senate, the pundits begin to say it, the voters begin to believe it.
If the polls say Biden will win in a landslide, the pundits begin to say it, the voters begin to believe it.
On Election Day — or even days after — few if any of these polling opinions prove true. The electorate is surprised. The polls were wrong. Their favorite pundits were wrong. The harm is done.
Who will trust presidential polls for the next generation? No one who lived through 2016 and 2020.
How will Democrats ever mobilize the kind of turnout they had in Texas in 2020? That turnout was driven — at least in part — by the belief that Texas could be turned blue-ish. What will Democrats say to get young Texans to the polls two or four years from now?
How will folks convince Georgia or Florida or North Carolina voters that record-breaking turnout can overcome surgical and sophisticated voter suppression? How many years of long voting lines will busy people put up with in losing efforts?
We have to rethink the way we poll and the emphasis placed on polls in future elections. Polling is an important part of the campaigning and electoral process, but consistent inaccuracy erodes confidence in that process, an erosion that free and fair elections can’t afford.