How We Conducted Our ‘Shy Trumper’ Study

Are there “shy Trump” voters? It’s a question that many have asked since Morning Consult released a blockbuster study during the primaries that showed GOP voters were less likely to say they support Donald Trump to a live interviewer than via a self-administered interview online. But will “shy Trumpers” play a large factor in the general election?

In a follow-up to the study from the primaries, Morning Consult, in partnership with POLITICO, devised a definitive study to test this claim. The results, summarized below, show that while shy Trumpers do exist, they are not significant enough to affect the vote on Nov. 8.

Related: Yes, There Are Shy Trump Voters. No, They Won’t Swing the Election.

Study Design

Morning Consult conducted a survey among 2,075 likely voters from Oct. 27 to Oct. 30, 2016. Respondents entered the survey online and were asked a few background questions. Once completed, they were randomly assigned to complete the rest of the interview either by:

  • Brief live telephone interview containing vote choice questions by calling an inbound call center (n = 1,249)
  • A short, identical online interview (n = 825).

Once assigned to a live telephone interview or online interview, individuals answered a set of vote choice questions about the 2016 presidential election and other potential public actions related to support for a candidate (e.g., talking to friends in support of, putting out a yard sign, attending a rally). Finally, after completing the interviews, the leaned version of online and phone interviews was utilized because online typically has higher levels of “don’t knows” than live telephone.

Key Findings

There is no overall social desirability mode effect

Overall, there is no mode effect among likely voters. Clinton leads by 5 (52-47) by phone and by 3 (51-48) online. This is an effect in the expected direction (Clinton performing better on the phone), but it’s not statistically significant.

hiddentrumpvote-overall

There is a social desirability effect with well-educated and higher income voters

Trump does perform worse on the phone with a live interviewer if the respondent has a college degree (Bachelor’s or post-grad). Trump performs 14 percentage points worse among college grads in live phone interviews.

hiddentrumpvote-educ

Trump also performs worse on the phone among households earning more than $50,000 annually. He trails Clinton among households earning more than $50,000 by 10 points on the phone, but is tied in such households online.

hiddentrumpvote-income

Brief Analysis

After analyzing the results, it would appear that social desirability bias is not playing as large a role in the general election as it did in the primaries for two main reasons:

The primary and general electorates are vastly different

When looking at who turns out to vote in the primaries versus the general election, there a few key items that could downplay the “shy Trump” voter effect:

  • More Voters in General Election than During the Primaries — Far fewer voters turn out in the primaries than during the general election. Additionally, the primaries are a test of party support. In the general election, it is not just a test of GOP voters, but also Democrats and independents.
  • Primary Electorate Are Higher-Educated and Higher Income — “Shy Trumpers” tend to be concentrated among adults with higher levels of education and higher incomes. These types of voters make up a more significant chunk of the electorate in the primaries than they do in the general election.

Different choices in primary than general election

This matters in terms of the “Shy Trump” vote for a couple reasons:

  • Primaries Tested Donald Trump Against Fellow Republicans — During the primary, voters were asked to make a decision about who bests represents their party. Social desirability is more likely to play a role among your colleagues and friends.
  • Both Clinton and Trump are Highly Controversial Candidates — Voters are not just deciding on Trump, they are choosing between two candidates who face very high unfavorable ratings. They are both socially undesirable to broad swaths of the electorate.

Full Results: Toplines, Crosstabs

Morning Consult