Editor’s Note: Mark Mellman is Democratic pollster and strategist who has helped guide hundreds of campaigns from president to city council. Mellman is a member of the American Association of Political Consultants Hall of Fame and has taught at Yale University. The views expressed in this commentary are his own. View more opinion on CNN.

CNN  — 

We’re in the midst of a historic first: As Donald Trump’s trial winds down, a jury will decide as early as the coming week whether the former president is guilty of a crime resulting from his alleged effort to conceal the hush money paid to Stormy Daniels.

Pointing to surveys like an ABC News/Ipsos poll that found a fifth of Trump voters would either reconsider or withdraw  their support if he’s convicted of a felony, some analysts contend a conviction — itself far from a certainty — could cost Trump the White House in November.

The Mellman Group
Mark Mellman

As much as I’d like to believe it, there’s ample reason for doubt. Trump may well lose the presidential election again, but a criminal conviction in the current case is unlikely to play a significant role in that outcome.

While pollsters can ask voters questions and voters will generally oblige with answers, in some instances the answers can be meaningless mush. Part of the science and art of polling is understanding what kinds of questions are more and less likely to yield meaningful insights.

Two kinds of questions tend to be in the mush category — those that ask people directly about changing their views and those that ask individuals to transport themselves to an alternate reality and predict what their response would be in that new context.

Asking people how their votes would change if Trump were convicted falls into both categories.

Then-President Bill Clinton’s highly partisan and wholly undeserved impeachment in the late 1990s illustrates how survey research and human brains operate in these cases.

A Washington Post-ABC News poll conducted in January 1998 found 55% of voters believed Clinton should be impeached if he had lied under oath about not having an affair with Monica Lewinsky, while 40% opposed impeachment under those circumstances.

Months later, it became clear the president had lied. But instead of 55% supporting impeachment, only 30% favored impeaching him (whereas 66% opposed impeachment).

Americans had claimed they’d favor impeaching Clinton if he had lied about his relationship. However, when it was evident that he had not been forthright, they overwhelmingly opposed impeachment.

Similarly, when the Post pollsters asked whether Clinton should resign if the House voted to impeach him or fight the charges in the Senate, only 38% of Americans wanted him to battle the charges, whereas 58% preferred he resign.

Did the electorate hold onto that view after Clinton was impeached by the House?

No.

In fact, the numbers changed almost instantly. The proportion of Americans who said Clinton should resign rather than fight the charges fell to 42% within days of him being impeached.

Difficulties inherent in such poll questions also emerged in the wake of the 2020 election —when most Republicans believed Trump’s fantasy that he had actually won the election. Many of them told my polling firm that they would revise their view if the courts ruled against Trump. However, despite their previous protestations, their minds remained largely unchanged even after Trump lost some 60 cases on the integrity of the election.

Similarly, in March of 2016, an NBC/Wall Street Journal poll found half of the women who voted in Republican primaries said they could not imagine themselves voting for Trump in the general election. Nearly 90% did in November what they could not imagine in March.

Beliefs are persistent, stubborn things.

People fancy themselves open to new information and sometimes we are. But I’ve learned from 40 years of studying public opinion that mostly we cling to our preconceived views about social and political matters of import.

Get our free weekly newsletter

Moreover, people are not very adept at forecasting their reaction to changed realities. Humans struggle to transport themselves to  parallel universes and correctly predict their reactions in that alternate reality.

Is it possible that a Trump conviction will change votes? Of course it is.

The charges against Trump are more serious than those Clinton faced, and no former president has ever been convicted before.

Of course, some presidents have been deeply damaged by scandal. Richard Nixon’s approval rating sank from 67% to 31% between January and August 1973 as Watergate consumed public attention. When he resigned a year later, only 24% approved of his performance.

A Trump conviction could result in similar fate, though history would urge us to bet against it. The point, though, is that answers to poll questions today yield little insight as to how the public will ultimately react in November.

We’ll just have to wait and see what happens after the jury renders its verdict.