You can see that the narratives public leaders share with their following have a massive impact in determining consequential behaviour. In the case of MAGA crusaders — it’s important to understand how earlier tactics of describing polls as a form of suppression led to a disgruntled fan base who started to believe that even the media are lying.
Trump spoke at the White House on the 6th of November, accusing “fake polls” of attempting to keep his voters at home. He called these polls “suppression polls”. There is no denying that polls in the last 2 US elections have underestimated the voter turnout and support for Trump.
Yet, the semantics Trump employs paint a picture of the world vs Trump, with big tech, big media and pollsters all joining hands to fight off Trump’s re-election campaign. I posit that the reality of why polls were wrong again in 2020 is in fact much more nuanced than Trump suggests.
Trump is not employing new tactics to discredit the polls. Previous tweets in 2017 said “Any negative polls are fake news” and “Don’t believe the Fake News Suppression Polls”!
While his tactics to discredit polls play into Trump’s general narrative about fraud and intentional media deception, there are well-researched reasons Trump loyalists are difficult to poll.
Reason One: The social desirability bias
Research shows that social desirability bias led to systemic polling errors in Clinton’s favour in 2016. I considered this when I was talking to a Trump supporter earlier last year. I asked what he did and didn’t like about Trump. He said that he liked his economic plans and how savvy he was in bringing manufacturing back to the USA. He didn’t like how uncouth Trump was when he speaks, and how unpolished and offensive he is.
These reasons make sense and hint at a larger problem with polling an unpolished, offensive candidate: the social desirability bias.
Social desirability bias is a bias which affects how people answer surveys. As a social scientist, I am all too aware that people will answer questions to conceal preferences which aren’t perceived to be socially desirable. Research has found socially desirable responding makes people conceal political preferences when they are socially inappropriate.
A study which used real polling data from the US 2016 election showed that due to social desirability bias, polling overstated agreement with Clinton relative to Trump.
Ultimately, socially desirable responding often biases unincentivized surveys. But media, campaigns, and markets all employ unincentivized polls to make predictions about electoral outcomes. I believe this bias is uniquely important in any election with Trump as a candidate, giving how extremely polarising he is.
Reason Two: Measurement Errors
Are bad polls fake? What is the point of polling? Political polling is designed to take the pulse of an electorate, yet it isn’t a dynamic measurement. It is fixed at a certain time — e.g. when the respondent answers the poll.
This is at odds with how our opinions change over time. Elections are dynamic, voter’s opinions change over time in response to campaigns, information related to campaigns, and the happenings of the world around them (e.g. a pandemic, a recession, need for health care).
So, embedded in the nature of polling is a static measurement, trying to capture a dynamic opinion, this alone will create a margin of error in polling. That is not even considering the importance of representative samples, e.g. are the people who take the polls representative of the people who vote? How are they recruited, online & through landlines? What about voters who don’t have internet or a landline etc.
Reason Three: An inherent distrust is institutional systems
How do you get a representative sample of people who believe in conspiracies — e.g. QAnon? The demographic who believes in QAnon is the exact opposite of the type of person who trusts institutional systems enough to participate in a survey.
Those who believe in QAnon and other Trump loyalists are extremely unlikely to participate in traditional polling due to a high level of distrust in traditional systems.
Reason Four: Trump loyalists believe in suppression polling
It’s hard to persuade MAGA supporters that polling is just inaccurate and not a suppression tool when it was so inaccurate in the 2016 election. But, do negative polls act as a suppression tactic?
Polling results influence political candidates. Candidates generally adjust their campaign strategies if they are shown to be behind, or carry on as they were if they are doing well. But do the polls influence voter behaviour?
Trump argued that the polls showing him as trailing mean that his voters won’t engage with the election because they feel it wouldn’t make a difference. Yet, research has found a weak effect of polling on influencing voter turnout.
Researchers define suppression polls as polls “which slander a candidate with the specific objective of discouraging turnout”. Yet, considering the turnout for the 2020 election was over 66.4%, the highest since 1900, and considering how much of the popular vote was for Trump, it seems the polls did not discourage turnout, but conversely, encouraged it.
Another worry of suppression polling is that by inflating a candidate’s lead, there are ‘bandwagon voters’. That is voters who jump on the winner's bandwagon. But again, research shows voters don’t turn out to ‘bandwagon’. Rather they most frequently turn out to affect the outcome of the election.
In this case, it seems the polling which showed Biden in the lead may have emboldened Trump supporters to vote — it may have mobilised his base to turn out to shape election outcome.
Understanding methodological bias and polling errors is so important, mostly to help people realise there are well understood, well-researched explanations for why polling can be wrong.
These explanations go beyond Trump vs the world. Explanations that hopefully reduce polarisation and don’t lead to an increase in violence when frustrated voters feel like the media has intentionally misreported results to change the election. Hopefully, polling firms can address these concerns if Donald Trump attempts a 2024 rerun.