The thing everyone fears about this election is the potential impact of the “shy” Trump voter. At first glance, this seems to be a reasonable fear — there is obviously a sizeable number of GOP voters that may have a problem with Trump but will hold their noses and vote for him regardless because voting GOP is what they do. And it’s not unreasonable that this internal conflict would result in these voters being unwilling to express their intention to vote for Trump to others, particularly to strangers.
The questions are, how sizeable is this group of people, and what impact will this have on the election? Speaking to Politico, pollsters Arie Kapteyn of USC and Robert Cahaly of the Trafalgar Group, the latter of whom predicted the Trump victories in the key swing states in 2016, believe that not only is this group significant, the potential for them to swing the election to Trump is not only possible but likely:
For Cahaly, those votes are likely to make the difference again. “There’s a lot of hidden Trump votes out there,” he says. “Will Biden win the popular vote? Probably. I’m not even debating that. But I think Trump is likely to have an Electoral College victory.”
Kapteyn backs this idea up with other research showing that while voters may not be willing to admit support directly for Trump, the impression to others in their “social circle” indicates that there are far more Trump supporters out there than polling and research can identify:
“We actually get a 10-point lead, nationally, for Biden over Trump” when asking voters who they personally plan to support, says Kapteyn. “But if you look at the ‘social-circle’ question, Biden only gets like a 5- or 6-point lead. … In general—and certainly on the phone—people may still be a little hesitant to say to that they’re Trump voters.”
This is a completely reasonable conclusion, backed up with evidence more solid than any poll — the result of 2016.
Here’s why I think it’s not likely that “shy” Trump voters will have the same impact this year.
1) The state polls WEREN’T wrong in 2016
At least not wrong in the same way that everyone keeps assuming they were.
Let’s look at the big three of Wisconsin, Pennsylvania, and Michigan:
|
Trump AVG |
Trump result |
Clinton AVG |
Clinton result |
Wisconsin |
40.3 |
47.2 |
46.8 |
46.5 |
Pennsylvania |
44.3 |
48.2 |
46.2 |
47.5 |
Michigan |
42 |
47.3 |
45.4 |
47 |
Notice that in only one case — Wisconsin — was a candidate’s share of the vote LOWER than polling indicated (and in that case only by a fraction). This was also the case for the vast majority of states in 2016 (for the top two candidates — third-party candidates actually frequently polled higher than the results, but not enough to account for the differences in the top two).
What does this mean? That while the state polls were wrong in predicting the winner, they WEREN’T wrong in predicting the base level of support. The problem wasn’t overcounting the Clinton voters — it was undercounting the Trump voters.
In other words, there wasn’t a significant block of voters getting attributed to Clinton that actually voted for Trump. It’s far more likely that the “shy” Trump voters were hiding in the undecideds or third parties. Which is further supported by...
2) Trump is OVERPERFORMING his Senate counterparts in many states
In Iowa, Ernst is at 48.4, while Trump is at 49.6. In NC, Tillis is at 47.7 and Trump at 48.7. In Michigan, James and Trump are both at 45.2. In Georgia, Perdue and Trump are both at 49.2.
Why does this matter? Well, no one is claiming that there are large numbers of “shy” Ernst, Tillis, James, or Perdue voters. If there ARE significant numbers of “shy” Trump voters out there, some will still answer the surveys and even if they’re not willing to support Trump, there’s nothing stopping them from supporting other GOP candidates.
In other words, unless EVERY “shy” Trump voter is avoiding ALL polls, you’d expect to see the Senate GOP candidates generally outperforming Trump. Instead, we’re seeing the opposite. Which likely means...
3) If the state polls ARE wrong, they have to be wrong in a different way
For the “shy” Trump voter to be a problem in this election, they’d need to be represented in the polls in a much different way than in 2016. It’s obvious from the results that the “shy” Trump voter wasn’t falsely representing as a Clinton supporter, which means that even if the polls are wrong in the same way as 2016, Biden’s absolute lowest level would be his current polling average — which is 50+ or close to it in many states.
No, for the “shy” Trump voter to be a problem, they’d need to be reacting in a different way than in 2016. Is this possible? Sure. Likely? Cahaly seems to think so, although his opinion seems more anecdotal than data-driven:
I’m finding that people are very hesitant [to share their preference for Trump], because now it’s not just being called “deplorable.” It’s people getting beat up for wearing the wrong hat, people getting harassed for having a sticker on their car. People just do not want to say anything
Setting aside the veracity of the claims of these supposedly “beaten-up” Trump supporters (this is from a Republican pollster, after all), it’s hard to imagine that there’s such a sizeable number of people that 1) are more embarrassed now than in 2016 about their support of Trump, and 2) are so embarrassed by it that they’re not even willing to answer an anonymous poll at all.
While this is possible, at the end of the day, most pollsters are professionals that want to be as accurate as possible in order to stay competitive, and they’ve adjusted quite a bit from the lessons of 2016. I couldn’t find comparisons in response rates between 2016 and 2020 polls, but with most polls being fairly consistent, it’s hard to believe that there’s a larger than usual problem with response rates that the pollsters aren’t attempting to account for.
Does the “shy” Trump voter exist? Undoubtedly. I’d like to believe that there are at least SOME GOP voters out there with some level of shame, no matter how partisan they are. But we’d need to see a problem with them larger than virtually any problem seen in the history of election polling, including in 2016, for it to dramatically alter the trajectory in 2020.