Don’t look at the polls!
by Karthika Cohen
Let’s not do what we did four years ago. Let’s not get a false sense of security from looking at election polling.
Yes, Biden is leading in the national polls. Yes, Biden’s lead over Trump nearly doubled after last Tuesday’s debate.
But at this time in 2016, Hillary was also leading in the polls. Depending on which poll you followed, her lead was anywhere between four and fourteen percentage points. And we all know what happened there.
The 2016 polls greatly underestimated the level of Trump support in key swing states, which included Pennsylvania. Pollsters said that a large number of late-deciding Trump voters, change in voter preferences during the campaign’s home stretch, and many Trump voters choosing not to reveal their support for him were mostly responsible for the polls being so wrong in 2016.
The Pew Research Center’s postmortem analysis put it down to nonresponse bias, which they described as “when certain kinds of people systematically do not respond to surveys despite equal opportunity outreach to all parts of the electorate. We know that some groups – including the less educated voters who were a key demographic for Trump on Election Day – are consistently hard for pollsters to reach. It is possible that the frustration and anti-institutional feelings that drove the Trump campaign may also have aligned with an unwillingness to respond to polls. The result would be a strongly pro-Trump segment of the population that simply did not show up in the polls in proportion to their actual share of the population.”
Nothing has changed with regard to Trump supporters in the last four years, so that can happen again.
The other reasons Pew put forth were the “shy Trumper” effect, which suggests that people were unwilling to admit their support for Trump since it was socially unattractive to be perceived as a Trump voter, and the difficulty pollsters faced in identifying likely voters, that is, models developed to predict what the electorate would look like were way off in their predictions compared to who actually showed up on Election Day.
Going by the number of Trump signs we see all over Ridley Township, it’s hard to believe that Trump voters are shy, but remember there are a lot more Trump supporters than those outright Trumpers. While pollsters have hopefully learned some lessons from 2016 and refined their polling process, it is still hard to rely on them for all the reasons mentioned above.
The fact that so many voters will be casting their ballots early and by mail is also bound to make polling difficult this year.
Pennsylvania’s well-known Franklin & Marshall College poll gave Clinton a double-digit lead in the state in its last poll prior to the 2016 election. The poll wrapped up its field work 10 days before the 2016 election and later found that a huge swath of voters—most breaking for Trump—made their decisions with the last 10 days of the campaign. The poll plans to stay in the field longer this year.
Trump ultimately won Pennsylvania by a narrow margin of less than a percentage point (fewer than 50,000 votes).
Polls have also the expanded their methodologies – for example, the Muhlenberg College Institute of Public Opinion in Allentown has added the category of educational attainment in their polling addition to the traditional categories of gender, age, region, party, and race, since the level of education really seems to matter in the case of Trump vs. anti-Trump voters.
Pollsters would argue that many of their predictions in 2016, even if not accurate, were well within the margin of error. Polls are almost never accurate, but usually voters don’t care if a victor wins by a larger or smaller percentage of the vote than what was predicted. But if the polling result swings to the other side, even if it stayed within the margin of error—as happened with Trump vs. Clinton (toward the end of the campaign, national polls showed Clinton ahead by just three points) —that’s a BIG difference.
So, basically, do not look at the polls.
As Christopher Borick, the Muhlenberg College Poll Director, said to the Guardian, “We sample. We take small groups to make inferences about big groups. And that inherently has error involved in it….. Democrats should not take anything for granted.”
VOTE! No matter what.