Are Trump supporters punking the polls?

President Trump is behind, and he knows it. How do we know? Simple. He wants more debates with Joe Biden. Whether you are a candidate for president or dogcatcher, leaders never want to debate. They want to protect their lead and avoid giving their opponents opportunities to catch up. Laggards want ever more debates — and more opportunities.

But that leaves the question: How far behind is Trump?

Over the past month, the closest national poll for Trump was the Investor’s Business Daily-TIPP poll, putting Biden’s lead within the margin of error at 45 percent to 42 percent. The biggest Biden lead was the CNN poll at 55 percent to 41 percent — a poll specifically condemned by Trump. The average over the past two weeks is just under 50 percent for Biden and just over 40 percent for Trump.

But how accurate are the polls?

Keep in mind that the 2016 national polls were fairly accurate — it’s just that the political elites could not accept that the numbers showed Trump was within the margin of error and could win. Terrible polling in key swing states was the problem.

Are people sabotaging the polls?

Any poll relies on respondents telling the truth. Since there is no way to verify if every respondent is truthful, pollsters assume they are — or at least that dishonest answers are either negligible or cancel out. But there are ways to determine if polling is roughly accurate.

Since public attitudes on major issues do not change rapidly, consistent results on issue questions would indicate honest answers. When opinions shift significantly, that shift should be associated with a critical event. Unexplained major moves in public opinion could be consistent with dishonest answers.

Looking at past polling from two prominent pollsters (YouGov and Siena College and The New York Times) that make their detailed crosstab data available, both polling groups show generally stable numbers for questions such as how important health care as an issue is or approval of Congress — and their Trump-Biden ballot test is close to the RealClearPolitics average.

Some Republican leaders claim that Trump supporters are sabotaging the polls by lying about their true intentions and thus inflating Biden’s numbers. But that does not look likely.

For one thing, every pollster has demographic quotas calculated to mimic the projected general election turnout (gender, race, age, partisanship, etc.). If a Trump voter claims to be a liberal Democrat voting for Biden, he or she is simply replacing a real pro-Biden liberal Democrat and having no actual effect on polling results.

For Trump supporters to truly skew the polls, they would have to claim to be Republicans or independents voting for Biden. And there is not much evidence that is happening. In the June 16 YouGov poll (50 percent Biden, 41 percent Trump), GOP support for Trump is 90 percent to 6 percent — an exceedingly high result. Even if you think that 100 percent of Republicans support Trump (an impossibility), that narrows the gap with Biden to only 50 percent to 44 percent. What’s more, Siena College reports mirror opposite numbers: Democrats 90 percent to 5 percent for Biden and Republicans favoring Trump by the exact same percentages.

There is more room in the numbers for Trump supporters to claim they are independents for Biden (or undecided). The Siena poll reports independents favoring Biden 47 percent to 29 percent, a very large gap and out of step with the average of other polls. The YouGov June 16 poll reports a Trump lead of 41 percent to 39 percent, while its June 9 poll reports a 42 percent to 39 percent Biden lead.

The best explanation for these differences is not a sophisticated sabotage that targets only Siena College. More likely, a high proportion of Trump-supporting independents are refusing to take the Siena survey, thus making the sample not random and leading to a skew among independents toward Biden. The YouGov poll could be biased, but the overall ballot test is in line with the average of all polls — so it seems more credible.

If there is a systematic underreporting of Trump’s numbers, it is more likely that some Trump voters are embarrassed to admit it and claiming to be undecided. Research on polling has shown that some people are reluctant to reveal their true preference if that preference is not considered socially acceptable. And Trump is just about the most vilified president ever.

The evidence from 2016 does not support the contention that there is a vast army of “shy” Trump voters. But in a close race, it takes only a percentage point or two to make a difference. It may have been in 2016 that a slice of the electorate too small to be detected by the polls but large enough to make a difference did tip the balance for Trump.

It’s all about the margins

What can we reasonably conclude from all this?

First, it is highly unlikely there is a mass sabotage of national polls by Trump voters. Millions of Trump fans would have to be prepared at a moment’s notice to take a polling call and answer most questions honestly and a couple of questions wrong.

Second, bad polling is probably the result of bad sampling, such as Siena College not getting a truly random sample of independents.

Third, there are probably some “shy” Trump voters who are inflating the undecided numbers a bit.

Overall, the national polls may well be underestimating Trump’s support — but not enough to make up for his widening deficit with Biden. Trump needs to stop complaining about the polls and start fixing his presidency if wants to win in November.

Keith Naughton, Ph.D., co-founder of Silent Majority Strategies, is a public affairs consultant who specialized in Pennsylvania judicial elections. Follow him on Twitter @KNaughton711.