Introduction

There is a peculiar phenomenon in President Trump’s approval ratings in Utah: a certain number of voters report disapproving of the President, but still plan to vote for him in November. For them, any misgivings they have about the President are not strong enough to change their vote to another candidate. We might call these disapproving Trump voters.

As part of the Utah Policy/KUTV2/Y2 Analytics poll, we often call consenting respondents to get more clarification on their thoughts and pattern of responses to particular questions. In calls we made to such respondents, they confirmed that they answered in the manner described above (as disapproving Trump voters) and that it was not a mistake. They answered in one part of the poll that they’re ‘somewhat’ disapproving of the President and still plan to vote for him in the Presidential General Election later this year. 

What describes this phenomenon? Are voters of this kind not saying they are voting against their interests by voting for someone of whom they disapprove? Is their disapproval not of the magnitude to make them want to support the candidate best poised to defeat the President in an election? If not, why? More importantly, what other relative factors are less desirable than voting for a President they disapprove of?

We set out to investigate this question, both with quantitative and qualitative research, as we saw in it an opportunity to better understand Utah’s unique electoral landscape. The research also provides additional insights into broader political constructs that might be at play here, such as party loyalty, polarization, and tribalism, among others.

Overview

At face value, Trump’s approval rating is solid in Utah. While he is certainly a polarizing figure, his approval is well-consolidated among his base, if more in practice than in principle. He also maintains a slim but positive net approval (total approval - total disapproval) among Utah voters. His supporters vary in dedication but are more numerous, while those who disapprove of him are more vehement in their opposition but relatively fewer in number.

Trends over time are not shown here, but it is important to point out that President Trump’s approval rating has remained fairly consistent among his supporters in Utah in our analysis.  When elected, his approval rating was slightly over 45%. While it varied slightly, it stayed mostly consistent through many of the significant public relations challenges weathered by the Trump Administration in the past year, such as Robert Mueller’s testimony to Congress, the Articles of Impeachment presented by the House of Representatives, and the controversial federal response to the COVID-19 pandemic. This context is important for understanding the attitudes of his support base. Indeed, we called several respondents and each of them affirmed that their approval of President Trump stayed steady during and after these events.

Below, we rendered an interactive diagram to show the flow of respondents in our panel. This chart shows the movement between the Trump Approval and November election vote choice questions.

Note that the diagram is interactive, and the mouse can be hovered over specific strands to highlight their sizes and show them in context of the other choice movements. You can also hover over specific nodes to show their outflows or inflows.

Our areas of interest are the two strands of respondents that either ‘somewhat’, or ‘strongly’ disapprove of Trump who are nonetheless planning to vote for him in November according to our survey.

While sample size is not shown, it is not hard to see that these strands are relatively small. When analyzed using weights, these groups collectively comprise only about 1% of our sample -- not enough to be significant in analysis, nor enough to be predictably meaningful in winning an election. Quantitatively, approval and vote choice are still linked. That is, most people who disapprove of Trump will vote for Joe Biden (19% of respondents who disapproved of Trump said they would not) and vice-versa. Qualitatively, however, their reasoning might give us insight into the power of party loyalty and political tribalism in polarizing elections, however large or small these effects might be. For this reason, we decided to further analyze this group demographically, and also see if we couldn’t ask them about their reasoning directly.

Demographic Profile and Scope

In this section, we analyze subsets of our response data to get a sense of the demographic makeup of respondents who are disapproving Trump voters and why they place themselves there. We also show here, among the multiple factors associated with being a disapproving Trump supporter, which factor or factors mattered the most.

We analyzed the demographic profiles of two sub-groups of the type of respondents in question: those who noted disapproving of President Trump (either ‘somewhat’ or ‘strongly’ so), and those who approved of the President but only somewhat. We were interested in the second sub-group because while they represent a practical base of support for the President, their responses reflect some misgivings, however insignificant, about the President’s performance. In a similar but not identical vein as the first sub-group, then, this group could likewise give us insight into the determinants for being unhappy with President Trump’s performance. While the two sub-groups had slight differences, they do not seem to be substantively different, so we describe both sub-groups of disapproving Trump voters here together.

Demographically, people who are dismissive of President Trump yet still plan to vote for him are essentially the same as Trump supporters. They are generally older, more conservative, and significantly more religious. This similarity comes from them both still planning to vote for President Trump’s second term.

Separating respondents who said they would vote for Trump into disapprovers and somewhat approvers did not yield any interesting differences in demographic makeup. Both of the groups were of the same demographic contour described above, with the exception of the ‘somewhat’ approving Trump voters being slightly more likely to be men.

We also analyzed the effects of various factors on casting a disapproving vote for Trump together. This type of analysis allowed us to control for other factors that we were simultaneously analyzing. For instance, we were able to analyze the effect of party loyalty while controlling for geographical differences. This step was important as we could not know the relative importance of factors in a simple tabular analysis because some demographic factors might be correlated. This helped us to parse out which, among many factors, were most important and the results were telling.

In this phase of analysis, we effectively analyzed the likelihood of fitting into the disapproving Trump voter category. For our purposes, the criteria for this category is planning to vote for President Trump in November even if one disapproves of him or only ‘somewhat’ approves of him. This takes the assumption that like reporting disapproval of the President, only ‘somewhat’ approving of him likewise connotes some apprehensions about the President that are similar to concerns inherent in disapproval. When controlling for party identity, county, religious affiliation, age group, and education levels, the variable that we found to be most significant was the party identification variable. Namely, Republicans are significantly more likely to be dismissive of President Trump, while Democrats are significantly less likely to be so. We note that we analyzed  Substantively, this means that, all things being equal, a Democrat is more likely to vote against Trump if that person disapproves of him, and a Republican is more likely to vote for Trump even if that person disapproves of him. Notably, the significance of this trait persists even when controlling for the other factors mentioned. All things being equal, party identity is the most important factor in self-reported voting preference.

Research shows that party identification remains the most powerful predictor of the attitudes and behavior of voters. Indeed, it overrides ideology and other considerations people often associate with vote choice. Therefore, Republicans will generally want to vote for President Trump regardless of other considerations.

Reasoning

The interviews we conducted with individuals about their support at the polls for President Trump highlighted some of the ways in which these other considerations enter into the considerations of voters. They also show the ways in which these other considerations rarely rise to the importance of overriding partisan considerations.

When we asked one of our respondents named Daniel about his answer patterns, he was careful to clarify, or perhaps qualify his own response by saying “I don’t disapprove of [President Trump], I just don’t like some of the things he’s done or some of the ways he’s handled things. I like his policies in general and the results that he’s gotten -- that’s all.” He continued “...I think his Defense [policies are good], and taking out terrorists, and not being bullied by foreign powers. He’s his own man.” (emphasis added) This was a common theme for all of the respondents we contacted. A respondent named Will -- who, in his own words, is “... as ultraconservative a Republican as they come” -- admitted to mostly agreeing with the President on policy goals, but also to hating watching or listening to the President for longer than two or three minutes. Another respondent, Nancy, admitted that she “...really struggles with him as a person” but mostly supports his political platform. Each of these planned to vote for President Trump again in November despite their apprehensions. In Will’s case, however, he will do so “...begrudgingly.”

These answers once again point to the importance of other considerations when assessing the performance of a president. And yet the considerations do not often move a partisan away from supporting a president of the same party. In this case, our respondents’ answers spoke to unique preferences for a certain ‘style’ of leadership that a President might exemplify. However, when push comes to shove, these considerations do not undermine support for the ultimate form of approval: a vote at the ballot box. 

This conclusion suggests some caution when interpreting approval ratings. It is hard not to conclude that responses to approval rating questions are idiosyncratic and somewhat random, given that a person who approves of an official’s political stance might disapprove of stylistic features of the official’s behavior. In other words, disapproval is not always disapproval of the magnitude that would move a person away from partisan loyalties.

Our respondents were certainly not shy to explain what they don’t like about the President. When I asked Daniel to elaborate on those stylistic features of the Trump presidency, he explained “ ...he’s too sarcastic and… he ridicules people… He ridicules people to their face and I think that’s inappropriate. He could do it more anonymously so that he doesn’t have to name them by name and sit and ridicule them on and on.” One respondent named Glenn stated that he thinks President Trump is “not refined”. Similarly, Nancy said that she doesn’t “...believe [President] Trump is a moral person” and another respondent, Beth, said she doesn’t “...appreciate his candor and the way he presents things.”  Will even went so far as to say he thinks “...[Trump] is a buffoon” and “a disgrace”.

These responses reveal countervailing considerations that voters face when assessing a president. With regard to style, they express multiple problems with the President. They see his sarcastic tone as unnecessary, find his ridiculing inappropriate, do not appreciate his confrontational style, and believe that he could be more tactful with how he addresses disagreements with different people. 

Political choices are not made in a vacuum. Respondents to surveys have to judge the quality of a candidate and the probability of voting for that candidate against the qualities and characteristics of opponents. For many of our respondents, President Trump’s faults were not enough to make them overlook the perceived faults of his opponents. Both Will and Glenn spoke directly to this line of reasoning. Will admitted that, compared with the former vice president, he sees President Trump’s tone and “...[willingness] to take action” endearing. He sees the other candidates like VP Biden as having “... some credibility issues.” He went on to express his assessment of his age by saying. “I’m concerned with Crazy Uncle Joe… and that he won’t make it through a first term.” He continued, “I also think he’s losing his memory. He’s not as sharp as he once was.” 

While partisan loyalties matter most, other considerations seem to make the vote choice more difficult for some than for others. For example, Will admitted he would consider Vice President Biden, but it would depend heavily on his pick for vice president. “If he picks somebody more [centrist] in the Democratic Party… such as Amy Klobuchar, then I don’t know” he offered. “If he picks [Elizabeth Warren], that’s not gonna win me over, nor is that congresswoman from California, [Kamala Harris].” Nancy drew a sharp distinction between the president and his policies. She sees “...the office [as separate] from the man.” She continued “I’ve always been pro-life. Also a balanced economy, a strong economy [are important to me]. I like it when states can have more control. Those are the kinds of things in the Republican platform that matter to our family. Education. Those kinds of things. Y’know, national security, those things that I know the platforms are split on.” Both of these respondents leave open the possibility of voting for somebody else if conditions are right. 

Conclusion

Much of what we see today in modern politics baffles both the average citizen and the seasoned political observer. The data and the interviews reveal that some factors, such as party identification and loyalty, continue to exercise a significant influence in the ways in which voters think and behave. However, citizens now confront a political landscape in which the expectations and norms for politics have changed. They are required to consider situations and practices that have been relatively rare in modern American politics. Such situations and practices inevitably reveal tensions between principles and partisanship. How the citizen navigates these tensions, resolves them, and then maps those decisions on to a public opinion question makes survey research more exciting and demanding than ever.

For this reason, we will continue to dig deeper into the distributions and trends that we see in our survey work and supplement those with the insights from the people who are called upon to answer the questions devised by researchers. We hope that the results provide a richer understanding of modern politics and the various elements that affect public opinion. In this way, we seek to contribute to the maintenance of this democratic republic.