How Memory Influences Hopes and Anxieties about the Election
Availability and vividness of specific memories tell us what to believe.
Posted Oct 27, 2020
Despite consistent political polling that shows Joe Biden leading Donald Trump nationally and in key battleground states, many supporters of Joe Biden remain anxious and even pessimistic about their candidate’s chances, while many supporters of Donald Trump maintain confidence and optimism. Why the disconnect between representative polling and beliefs?
One answer is the availability bias. We use the availability of specific examples in memory to make judgments about the likelihood of worldly events in general. People who watch a lot of local news, for example, overestimate the level of danger in their community because they can readily think of specific examples of crime from the news. In joint projects, people consistently overestimate their own contributions relative to others—because instances of their own contributions are far more available in memory. The same applies to those in romantic relationships. People magnify their own contributions while underestimating their partner’s.
By far, the most available and salient memory for thinking about the outcome of the upcoming election is the 2016 presidential election. It projects itself over other elections since then and more importantly, over available data in the form of opinion polls.
For those who support Donald Trump, the memory of the 2016 election is salient and readily available, likely giving hope for the upcoming election, despite the current political polls. This same memory effect may even occur for the candidate himself, with the added influence of recent memories of cheering crowds at large rallies.
For supporters of Hillary Clinton (many of whom now support Joe Biden), the results of the 2016 presidential election came as a shock—and then, for many, as a kind of trauma. So the memory of the 2016 election is not only salient and available, it may also be traumatic, with all the characteristics of traumatic memory: vividness, strong emotionality, and the tendency to be relived. The memory then helps maintain a wary and pessimistic perspective.
According to stable patterns in current political polls, neither perspective is accurate. The memories themselves are accurate—the remembered victory and joy for Trump supporters and the remembered loss and shock for Clinton supporters. But the extrapolations are not. Of course, political polls have different sampling strategies and margins of error (MoE), but many measured differences between the two candidates are now outside the MoEs. Moreover, the actual percentages listed for each candidate show the best estimates for each poll. (See Note 1 below and this earlier blog post.)
Even our general suspicion of political polls is likely based on the availability bias. Given the sophistication and transparency of highly regarded polling organizations, most political polls are accurate within their MoE. Polling has come a long way since the infamous error of 1948 when Thomas Dewey was predicted to defeat Harry Truman. More recently, polls from several states that overcounted Clinton voters in 2016 have made corrections. But examples of polls being demonstrably wrong are far more newsworthy than examples of accuracy—and thus more available in memory. So we exaggerate the inaccuracy of polling and diminish its value.
This judgment is reinforced by our bias about the lack of available instances in memory—a kind of reverse availability effect. If we cannot recall instances of a particular outcome, we then judge it as unlikely.
What can we do about the influence of our specific memories? In this case, it may be more prudent to set the memories aside and pay attention to the political polls. Polling results supply the most accurate available information about an upcoming election. It is these results that should be used to anticipate the winner of the upcoming presidential election and to inform how we feel about its likely outcome.
Note 1. Even within the margin of error (MoE), the polling results are noteworthy. If Candidate B leads Candidate T by 52% to 48% in a poll with a MoE of 4%, that means the poll’s best estimate is that Candidate B leads by 4 percentage points. Simply being within the MoE does not mean the candidates are “virtually tied” or in a “dead heat.” Indeed, it’s just as likely that Candidate B leads Candidate T by 8 percentage points as it is that the candidates are tied. Measurement error should be noted but not misinterpreted.