Pollsters often assume respondents can remember basic political behavior. It’s not a wise assumption.
Professionals are often asking people whether they voted last year, who they voted for, whether they got a piece of mail or saw a particular ad.
We possess large, powerful brains, but we nonetheless often forget where we put our keys, where we left off in a book or TV show, a friend’s phone number or birthday, or even what we learned in this morning’s class.
One review of recent experiments found people forgot as much as 94 percent of what they had learned.
And it is not just trivial things that are forgotten. One study found some 25 percent of surgical patients forget a material fact about their treatment—whether they had surgery or not, whether a body part was removed or left inside them.
If you don’t remember whether you have an appendix, why would you expect to recall whether you voted or for whom?
And, in fact, lots of people don’t.
Often 10 percent to 20 percent more Americans report having voted than actually turned out. Similar differences exist in reports of registration.
Pollsters also love asking respondents who they voted for in the past.
Those results can be even stranger.
A recent poll asking respondents who they voted for in 2016 found Clinton and Trump tied in the national popular vote. That’s not so far off. But the fact that the toplines work out doesn’t mean that people are accurately reporting who they supported in 2016.
In other years, the gaps between recall and reality have been even bigger.
This year, voters told Morning Consult pollsters they had given Barack Obama a 24 point margin over Mitt Romney, though they had actually produced just a 4 point advantage for the Democrat.
By June 2009, the CBS/New York Times poll found people reporting they had given Barack Obama a 28 point edge over Sen. John McCain (R-Ariz.), who, in fact, only lost by seven points.
This is nothing new. George W. Bush lost the popular vote to Al Gore. Four years later, voters were telling pollsters they had given Bush a 9 point victory.
Post-election victories don’t always go to the winner. Then-Sen. John Kerry (D-Mass.) lost to George W. Bush by 2.5 points in 2004. Two years later, voters were telling pollsters they had voted for Kerry by 3 points.
Reviewing these kinds of numbers leads many to say respondents lie. Some may.
There is evidence that politically engaged voters are more likely than others to say they turned out to vote when they hadn’t. Psychologists don’t call this lying; they use the more dispassionate term “social desirability bias.”
People fudge the truth to make themselves look better. And if you are politically engaged, not voting makes you look bad.
If a president looks bad, people may be reluctant to tell pollsters they voted for him.
Social desirability bias suggests motivation — people have reasons to misstate reality.
There is also just forgetfulness. Politics is not the most important part of many peoples’ lives. People do lots of things. They vote (or not), go to the dry cleaners (or not), have an appendectomy (or not), have kids doing well or poorly in school.
People forget about these matters that, for many, are more salient than politics.
And some candidates fade from the news more quickly. Hillary Clinton has remained prominent in public discussion since the election. Mitt Romney did not. It’s easy to see how Romney may have been forgotten.
The real lesson here, though, is to beware the conclusions. When pollsters say this is how Trump or Clinton voters feel, or whether they saw an ad or received a piece of mail, take it with many grains of salt.
Mellman is president of The Mellman Group and has helped elect 30 U.S. senators, 12 governors and dozens of House members. Mellman served as pollster to Senate Democratic leaders for over 20 years and as president of the American Association of Political Consultants.