Skip to main content

Authoritarianism Among Pro- and Anti-Trump Voters (Part 2)

by Patrick Murray

As discussed in a prior post, Monmouth University’s Polling Institute assisted with a survey that formed a central part of Authoritarian Nightmare, a new book by John Dean and Bob Altemeyer. In another post, I examined measurements for Right Wing Authoritarianism (RWA) and Social Dominance Orientation (SDO) and how they relate to support for President Donald Trump. I take that analysis a step further to examine variations within the electorate on both sides of the partisan divide and a key question about constitutional norms.

While the analysis in the book, as well as in my last post, show a clear correlation between RWA and SDO inclinations, that is not to say that all Trump supporters are authoritarians. Or indeed that all Trump opponents are not authoritarians.  While 32% of strong Trump approvers score in the highest quartile on both the RWA and the SDO scales, 23% actually score below the top quartile on both dimensions (Table 1). Among voters who somewhat approve of the president’s job performance, 51% fall below the top quartile on either scale. Conversely, while the vast majority of Trump opponents have a moderate or low score on these scales, 21% of those who somewhat disapprove of Trump and 5% of those who strongly disapprove of Trump exhibit high RWA and/or SDO tendencies.

Table 1.RWA/SDO Inclination by Trump Job
Rating
 Strongly
disapprove
(n=463)
Somewhat
disapprove
(n=43)
Somewhat
approve
(n=119)
Strongly
approve
(n=346)
Double High RWA & SDO1%0%9%32%
High RWA only2%14%24%20%
High SDO only2%7%15%26%
Moderate RWA/SDO21%49%41%19%
Low on both74%30%10%4%

Some Trump supporters aren’t particularly enamored of his authoritarian tendencies and at the same time some Democrats prefer authoritarian leadership, just not Trump’s.  Since opinion of Trump’s job performance correlates highly with partisan identity, we can ask which types of Republicans are more likely to be on the authoritarian train. The data do not show a lot of variation demographically (Table 2). Republicans without a college degree tend to score highest on the RWA scale, while younger Republicans are somewhat less authoritarian than older partisans. But that’s about it.

Table 2.Mean RWA/SDO Scores
Among Republicans
 RWASDO
All Republicans + leaners (n=465)11286
GOP identifiers only (n=300)11586
GOP leaners only (n=165)10686
Men (n=292)11388
Women (n=169)11183
Age 18-44 (n=53)9582
Age 45-64 (n=188)11090
Age 65+ (n=215)11884
High school or less (n=71)12089
Some college (n=137)12086
4-year college degree (n=160)10886
Post-graduate (n=96)10284
White (n=415)11287
Latino, Black, Asian, other (n=39)10983

There’s a similar lack of differentiation among Democrats (Table 3). All demographic groups score significantly lower than their Republican counterparts, but there are only minor variations within the party itself. Democrats who never attended a college class tend to be most authoritarian and white Democrats tend to score lower on the RWA scale than Democrats of other racial or ethnic backgrounds. Otherwise, the scores are fairly similar across the board.

Table 3.Mean RWA/SDO Scores
Among Democrats
 RWASDO
All Democrats + leaners (n=408)5345
DEM identifiers only (n=284)5344
DEM leaners only (n=124)5246
Men (n=192)5349
Women (n=216)5241
Age 18-44 (n=93)4945
Age 45-64 (n=147)4843
Age 65+ (n=168)5947
High school or less (n=36)7547
Some college (n=69)5545
4-year college degree (n=137)4945
Post-graduate (n=166)4944
White (n=323)4945
Latino, Black, Asian, other (n=81)6645

I looked at potential variations among Democrats in another way by using a question on presidential preference as a proxy for different orientations within the party. Remember, the Democratic nomination was still wide-open when this survey was conducted in the fall of 2019. Mean RWA scores ranged from a relative high of x̅=59 among Joe Biden supporters (n=101), to x̅=48 among Bernie Sanders supporters (n=41), x̅=48 among Pete Buttigieg supporters (n=88), and x̅=44 among Elizabeth Warren supporters (n=75). The mean RWA score among Democrats with no candidate preference (n=63) was x̅=59.  [Interesting side note: Tulsi Gabbard was named by 32 respondents as a preferred Democratic nominee, but not by anyone who actually identified as a Democrat (the nominee preference question was asked of all survey respondents). Her supporters averaged x̅=106 on the RWA scale, which is much closer to the Republican average than the Democratic one.]

Both parties include their share of “authoritarians,” but there is a clear lack of critical mass on the left. Among Democrats, 11% of Biden supporters and 10% of Sanders supporters had a high score on the abbreviated 5-item RWA scale in the poll, while only 3% of Buttigieg and Warren supporters did. There simply aren’t enough authoritarian Democrats in the electorate to rally around a single leader and take control of the party. Of course, this may also be a “chicken and egg” problem. Voters who prefer authoritarian style leadership may be less comfortable with the Democratic Party as much for its lack of centralized leadership as for any specific set of policy issues. But considering the strong link between authoritarianism and religious fundamentalism in general, social issues must certainly play a key role in these wide partisan gaps on the RWA scale.

I attempted to strip the original RWA scale of its obviously religious items (discussed in my prior post). I found that the scales held up without the items that referenced values on gender and sexuality. However, this finding needs to be taken with an important caveat. The full battery of questions in the survey interspersed the value-laden items among the leadership items, so it is possible that respondents answered the full question set in ways that attained a level of internal attitude consistency (what we call context effects in the polling biz). This means that we could possibly get a different set of answers to the 5 questions in the shorter RWA measurement if they had been the only part of the scale included in the survey. We would need to test this in isolation to be certain.

As I mentioned earlier, my research interest lies in exploring the psychosocial underpinnings of political behavior. But I also have a normative interest in this particular study. To what extent does authoritarianism pose a danger to the U.S. Constitution?

The authors added a question to the survey that attempted to directly measure this question. However, the results may seem counterintuitive, at first glance. The question wording was: “The U.S. Constitution gets in the way of things too much nowadays and should just be ignored when it interferes with taking action on some issue.”  Overall, only 6% of respondents agreed with this statement or even took a neutral stance on it. In fact, 69% picked the most extreme “disagree” option.

Now, the first thing that comes to my mind in examining these results is social desirability bias. The idea that we should respect the U.S. Constitution is so baked in to the American psyche that you can hardly expect anyone to admit they are willing to disregard it when asked so blatantly: “I’m not undermining the Constitution. You’re undermining the Constitution!”

In fact, the responses to this question tended not to correlate with any of the RWA items in the survey. In the few cases where there was a significant – albeit weak – relationship, the correlation tended to be negative. In other words, willingness to ignore the Constitution to attain a policy end tended to be held – to the extent it was held at all – by people who had lower authoritarian tendencies.

I have not spoken with Dean and Altemeyer about this, but I expect they may have been surprised by the results.  Fully 83% of respondents who strongly approve of Trump said they very strongly disagree with this statement – the most negative option in the survey’s 9-point response set. This level of disagreement was less widely shared among voters who were less stalwart Trump supporters, including those who somewhat approve (68%), somewhat disapprove (51%), and strongly disapprove (62%) of the president’s job performance.

If you look at this question by partisanship, just 4% of Republicans either agreed or took a neutral stance on the question of the Constitution getting in the way of things, compared with 8% of Democrats and 10% of independents. This is a small, but still statistically significant difference. There is even further differentiation among Democrats. Sanders supporters (20%) are more likely to agree with this sentiment than backers of Biden (10%), Warren (5%), or Buttigieg (3%).

This may be a bit of a shocker because the charge against Trump has been that he is willing to trample on the Constitution to further his aims and his followers have been willing to go along with it. Does this mean strong Sanders supporters are more likely to ignore the Constitution to achieve their ends than strong Trump supporters are? I don’t think so. In fact, I hypothesize that these Sanders backers are just more likely to admit it – and indeed even recognize that they are doing it.

If you look at the Constitution as an operational framework for government rather than an ideological document, its purpose can basically be boiled down to concerns rooted in its historical origins. One key purpose was to create a functioning central government that would engender enough public trust to prevent anarchy and chaos (think Shay’s Rebellion). The other purpose was to prevent authoritarianism; that is, allowing any individual leader to put themselves above the law (think King George III). Basically, the Constitution works, not when everyone gets what they want from a policy standpoint, but when the public trusts that those who hold the reins of power are observing its rules of engagement.

Now, there are people on the left who want anarchy and those on the right who want an omnipotent ruler. For both, the U.S. Constitution’s purposeful ambiguity is an impediment to their ends. It just seems those on the left are more likely to actually recognize this and admit it. What neither side seems to realize, though, is this: in a nation as vast and diverse as ours, those tedious checks and balances embedded in the Constitution are in fact what keep this country on an even keel during uncertain times.

The bigger danger to our Republic does not, in fact, come from those who admit to seeking a wholesale change in our form of government. You can see that coming.  No, the more perilous hazard comes from those who are willing to erode Constitutional norms from within by manipulating the natural fears and psychological dispositions of a segment of the American public. And the biggest danger of all comes from those who do this while duplicitously giving lip service to the core principles of our Republic’s founding document.

Authoritarian Tendencies in the American Electorate (Part 1)

by Patrick Murray

The release of Authoritarian Nightmare by John Dean and Bob Altemeyer raises important questions about the underlying values and motivations of the American electorate. A core part of their analysis is based on a survey of voters conducted with the assistance of Monmouth University’s Polling Institute. I discuss our participation in the project in another post, but here I take a deeper dive into the explanatory power of authoritarianism for American voter attitudes and behavior.

[Note: this is a lengthy “extra for experts” post aimed at those wanting to understand psychosocial dimensions of political attitudes and behavior – and whether it is even possible to measure these constructs. If you are only interested in polling to forecast the next election, look elsewhere. The analysis here is based on a survey of 990 registered voters conducted online from late October to November, 2019.]

The framework for the book’s analysis is found in a number of psychological scales developed by Altemeyer and others to measure perceptions of prejudice, social equality, morality and preferences for strong leadership. Altemeyer’s premise is that traditional religious values and authoritarian tendencies are interrelated. [However, I must emphasize that the following thoughts are mine alone and do not represent Altemeyer’s research or writing.]

The existence of a correlation between traditional social values and authoritarianism makes sense from a lay person’s point of view. People who value tradition are more likely to be threatened by changes to the social order they know – whether those changes are real or perceived. And if you are concerned about the world changing too rapidly, the more likely you are to cede control to a strong authority figure who will do whatever is necessary to stem or reverse that cultural shift.

In fact, many evangelical voters offer a similar rationale in their continued support of Donald Trump. His behavior may be antithetical to their stated belief system in many ways, but they can rely on him to fight for their priority concerns. And while none of them have actually articulated it in this way, it basically boils down to: “If some Constitutional norms need to be undermined to overturn legal abortion, then so be it.” The ends justify the means.

This description is admittedly a simplistic depiction based on one type of single-issue voter. For many other Trump supporters, though, the cultural shifts they hope to reverse are more amorphous than any particular policy. This type of person’s calculation is more about having to confront unknowns in their daily life – a sense of discomfort and discontent that they are not getting ahead while “others” are. In this context, the passive authoritarian is willing to cede control to a strong leader who can identify and vilify the “other.”

Altemeyer has been utilizing his scales in a variety of settings for nearly four decades, but this new book marks the first time they have been put to the test with a representative sample of the American electorate. As the discussion in the book illustrates, this new data supports many of his prior claims about authoritarianism.

How Does One Measure Authoritarianism?

As someone new to the scene, though, I examined his Right Wing Authoritarian (RWA) scale and asked whether it is a measurement of traditional values more so than it taps into a willingness to cede authority to a strong leader. A look at the 20 questions in the scale finds a mix of items, such as “Our country will be great if we… do what the authorities tell us,” as well as “God’s laws about abortion, pornography and marriage must be strictly followed before it is too late.”  The scale includes constructs around strong leadership as well as items tapping into traditional views of sex and sexuality.

My question is whether we can tease out these two constructs. Altemeyer’s 20-item scale seems pretty solid as a measurement tool, with high reliability score of α=.96.  I wondered if the scale hold up without the sexual morality component, so I excluded the most blatant religious value questions and replaced them with a couple of additional items in the survey that asked about adherence to strong leadership (see question list here). This new 11 item scale had a similarly strong reliability score (α=.91). And because academic social scientists have a tendency to overegg the sauce, I also created an even more efficient 5-item scale. This produced a similar level of reliability (α=.90).

Lo and behold, these new scales exhibit nearly the same distributions as the findings discussed in the Dean and Altemeyer book that showed increasing authoritarian tendencies among Trump supporters (Table 1).  The mean authoritarian score of strong Trump approvers is more than twice that of strong Trump disapprovers across the board.  The results here suggest that removing the most blatantly religious items from the RWA scale demonstrates a squarely different mindset among Trump supporters on the proper exercise of leadership to maintain “order.” In other words, there are greater authoritarian tendencies among Trump supporters regardless of whether the scale includes references to specific social “norms” of gender and sexual orientation.

Table 1.RWA Scale Means by Trump Job Rating
RWA scale…Strongly
disapprove
(n=466)
Somewhat
disapprove
(n=43)
Somewhat
approve
(n=119)
Strongly
approve
(n=348)
Original 20-item scale
(range=20-176;  x̅=.84)
5482101119
11-item non-valence
(range=20-178;  x̅=.82)
567996113
5-item non-valence
(range=20-176;  x̅=.87)
5681101125

Despite the different composition of these RWA scales, all three of them correlate significantly with a separate scale measuring religious fundamentalism, although they do so at notably different levels (Table 2). These correlations range from r=.83 for the original 20-item scale to r=.68 for the 5-item scale with the sexual norms items removed. It is also interesting that all three RWA scales correlate highly with a separate scale in the survey designed to measure racial and religious prejudice (between r=.78 and r=.84). These concepts – racial equality and religious piety – were not referenced directly in any of the RWA scale items. While the religious aspect may be implicitly tied to authoritarianism because of how the original scale was constructed, racial prejudice is not. Yet, the two scales are indeed highly correlated. [By the way, the direct correlation between the prejudice scale and the fundamentalism scale is r=.65. Prejudice is not as strongly related to evangelism qua evangelism as it is to authoritarianism. These findings hold even when the analysis is run among white voters only.]

Table 2.Scale Correlations
(all are significant at p<.01)
RWA scale…Religious
Fundamentalism
PrejudiceSocial
Dominance
Original 20-item scale.83.84.68
11-item non-valence.72.78.62
5-item non-valence.68.79.63

The survey included another scale – one that may be even more telling about the underlying psychosocial dimensions of political behavior. The 16-item Social Dominance Orientation (SDO) scale measures a belief in and preference for maintaining hierarchy in society in general. [Note: the developers of this scale note that it includes two distinct dimensions – group dominance and anti-egalitarianism – that are meant to be scored separately. However, the correlations of the separate components to the other scales in our survey are nearly identical, so for ease of discussion I am using a summary scale value for SDO.]

The SDO scale does not specify any group identities, but it still has a strong correlation with the Prejudice scale (r=.80). This is much stronger than its correlation with the Religious Fundamentalism scale (r=.50). SDO also correlates with RWA – between r=.62 and r=.68 depending on which RWA scale is used. Other research has shown a weaker correlation between SDO and RWA, but those studies were conducted primarily with university students, whereas this study was conducted with a representative sample of American voters. The gap between Trump supporters and other voters is apparent on both scales, although the social dominance dimension varies over a smaller ranger (Table 3).

Table 3.Scale Means by Trump Job Rating
 Strongly
disapprove
(n=463)
Somewhat
disapprove
(n=43)
Somewhat
approve
(n=119)
Strongly
approve
(n=346)
RWA scale5482101119
SDO scale (rebased to
match RWA scale range)
45708190

The Dean and Altemeyer book suggests that while authoritarian followers provide Trump’s base of support, social dominators in his camp may pose the bigger threat. Basically, they see authoritarianism as passive and social dominance as aggressive (e.g. think of QAnon). About 1-in-8 survey respondents scored in the top quartile on both the RWA and the SDO scales and among this group, 89% are solidly in the pro-Trump camp. Among those who score highly only on the RWA scale, solid Trump support stands at 74% and among those who score highly only on the SDO scale solid Trump support is 58% of this group.

We know these “Double-highs” – as Dean and Altemeyer call them – on the RWA and SDO scales are firmly in Trump’s corner. The question is how far are they willing to go to back him?  One question in the survey is illustrative. It asked what should happen if Trump loses in November but he “declares the election was fixed and crooked” (Table 4). Nearly 1 in 4 (23%) double-highs agree that Trump should continue as president in this situation, while 53% disagree. Other high RWAs are not far behind (19% agree and 66% disagree) while other high SDOs are not as willing to back an unconstitutional extension of Trump’s term (5% agree and 79% disagree). That puts them in basically the same category on this question as voters who score moderately high on either the RWA or SDO scales (5% agree and 85% disagree).

Table 4.Trump should continue in office despite a loss  
if he declares election was fixed and crooked
 Double High
(RWA&SDO)
High RWA
only
High SDO
only
Moderate
RWA/SDO
Low on
both
Agree (7%)23%19%5%5%1%
Neutral (10%)24%15%16%10%1%
Disagree (83%)53%66%79%85%98%

Now, I am not a big fan of hypothetical questions. There are too many instances where what people thought they would do in a given situation do not match up with their actions once they are faced with the reality of the situation. But these results do suggest that Trump does maintain a core base of high RWA/SDO voters who might take to the ramparts for him and another group of high RWA voters who will tacitly offer their support. The question is whether these two groups form a critical mass in American politics.

Altemeyer’s position is that a certain number of people in any given population will always have authoritarian tendencies. The question is whether those inclinations are validated and authoritarian behaviors are deemed acceptable by a critical mass in society. Such situations seem to require a perfect storm of social uncertainty and economic volatility, but also seem to depend on the willingness of political leaders to cravenly play on those fears – or stand idly by while others do this.

It is possible that much has changed since Altemeyer first developed these scales. Religious fundamentalism as a political force was in its nascent stage four decades ago. There may have been other belief systems that correlated with authoritarianism just as well.  Regardless, these two constructs seem to be tightly linked today, at least in American politics.

This does not mean that all evangelicals or all Trump supporters are predisposed to authoritarianism. Nor does it mean that only those on the right of the political spectrum exhibit these tendencies. But the correlations are rather strong. I discuss these exceptions to the rule, as well as consequences for a constitutional republic, in the second part of this post.

Monmouth Poll Research on Authoritarianism

by Patrick Murray

The Monmouth University Polling Institute recently provided research support for a new book by John Dean and Bob Altemeyer entitled Authoritarian Nightmare. Our objective was to further public understanding of what drives voter attitudes and behavior in the 21st century. This is how our association with the authors came about. Thoughts on the data in the survey can be found in separate posts (here and here).

I have a long-standing interest in the role of identity in politics – how people see themselves and how that translates into political attitudes and behaviors, be it through the lens of race, class, geography, etc. The rational choice models typical of political science tend to miss the more visceral need to be part of a group and how that group identity can subsume other political calculations. This is the main reason why I studied political psychology in graduate school and eventually found my way into the polling business.

These factors became especially evident with the ascendance of Donald Trump’s 2016 candidacy. It is worth remembering that Trump had only a 20% favorable to 55% unfavorable rating among Republican voters before he got into the race. Within weeks of his June 2015 announcement, that opinion had flipped to 52% favorable and 35% unfavorable. He hit 70% favorability by Election Day and now routinely tops 80% among his fellow Republicans. I remarked at the time that a universally known candidate suddenly upending what voters think of him on the basis of a campaign speech was unprecedented.

Trump, however, was no typical politician and his appeal could not be readily explained by the usual political paradigms – no matter how much academics and pundits tried to do so. It wasn’t that a critical mass of Republican voters were looking for certain policy boxes to be ticked. These voters were not saying to themselves, “The lack of a border wall is keeping me up at night. Which candidate can I trust to build one?” They were looking for someone who could articulate why they felt unsettled in a changing world; someone who could identify the culprits responsible and vilify them. They were looking for someone who would ostensibly allay their fears – ironically by stoking those fears.

Clearly, something more than a rational choice model of political behavior was at work. But standard public opinion polling has a difficult time tapping into these psychosocial dimensions. These factors are always present but have become much more prominent – perhaps even overwhelming – in the Trump era. There were a few attempts to measure these extra-political factors during the 2016 campaign. None was particularly robust, though, largely because the lengthy question sets needed to tap into these dimensions cannot be administered reliably in a telephone poll. Surveys with psychological batteries tend to be long, wordy, and intentionally provocative, while standard polls are short, succinctly worded, and intentionally bland.

I began looking into ways to obtain better measurements of these phenomena as Trump supporters’ loyalty solidified once he took office. Enter John Dean.

Dean has spoken on the Monmouth University campus twice in the past five years and I had the opportunity to join him for lunch on those occasions. During a conversation in October 2018, I mentioned my concern that standard political polling and the media coverage of voter attitudes was not up to the task of fully explaining the current political climate, and specifically Trump’s ability to maintain his core support. John contacted me shortly after that meeting to discuss a book idea he was working on. In 2006, he published Conservatives Without Conscience, a play on the title of Barry Goldwater’s seminal work, The Conscience of a Conservative. Dean’s book profiles the evolution of the Republican Party as he saw it, drawing on the authoritarianism research of Altemeyer, a retired professor of psychology from the University of Manitoba. While the 2006 volume profiled Republican leaders, Dean and Altemeyer were collaborating on a new book examining Trump’s rank-and-file supporters in the context of the typologies Bob started developing four decades ago.

One thing they were missing, Dean explained, was a broad-based survey of American voters that could directly test their hypotheses. Upon hearing my own interest in developing a fuller understanding of voter attitudes and behavior in the Trump era, Dean asked if Monmouth might be interested in conducting a survey using Altemeyer’s psychological batteries. I saw this as an opportunity to further our body of knowledge by combining academic research with public polling methodology.

We came to an agreement where the authors underwrote a representative online survey of American voters (although we slightly oversampled Republicans to ensure a sufficient group size in the study). Monmouth maintained full control over the selection and management of the sample in line with our standard protocols. And while we agreed to work cooperatively with Altemeyer on the questionnaire content, Monmouth effectively had veto power over any questions, aside from the validated psychological batteries, that did not meet our standards for objective measurement. Dean made it clear on multiple occasions that he was fully prepared for the possibility they might need to rethink their premise if the data led to different conclusions. The numbers would fall where they may. In the end, the findings did support their hypothesis.

Monmouth delivered the final data set to the authors late last year, but played no role in the findings and conclusions of their book. However, part of the agreement was that Monmouth would have the ability to publish its own analysis of the data once the book was released. As such, we provide some initial thoughts on the results here and here.