You are on page 1of 34

719507

research-article2017
APRXXX10.1177/1532673X17719507American Politics ResearchMotta

Article
American Politics Research
2018, Vol. 46(3) 465­–498
The Dynamics and © The Author(s) 2017
Reprints and permissions:
Political Implications sagepub.com/journalsPermissions.nav
DOI: 10.1177/1532673X17719507
https://doi.org/10.1177/1532673X17719507
of Anti-Intellectualism journals.sagepub.com/home/apr

in the United States

Matthew Motta1

Abstract
Recently, Americans have become increasingly likely to hold anti-intellectual
attitudes (i.e., negative affect toward scientists and other experts). However,
few have investigated the political implications of anti-intellectualism, and
much empirical uncertainty surrounds whether or not these attitudes can be
mitigated. Drawing on cross-sectional General Social Survey (GSS) data and
a national election panel in 2016, I find that anti-intellectualism is associated
with not only the rejection of policy-relevant matters of scientific consensus
but support for political movements (e.g., “Brexit”) and politicians (e.g.,
George Wallace, Donald Trump) who are skeptical of experts. Critically,
though, I show that these effects can be mitigated. Verbal intelligence
plays a strong role in mitigating anti-intellectual sympathies, compared
with previously studied potential mitigators. I conclude by discussing how
scholars might build on this research to study the political consequences of
anti-intellectualism in the future.

Keywords
anti-intellectualism, antiscience attitudes, verbal intelligence, public opinion,
political psychology

1University of Minnesota–Twin Cities, Minneapolis, USA

Corresponding Author:
Matthew Motta, University of Minnesota–Twin Cities, 267 19th Avenue South S1414
SOCSCI, Minneapolis, MN 55405, USA.
Email: motta018@umn.edu
466 American Politics Research 46(3)

The distrust and dislike of scientists, academics, and experts more broadly
was a widely noted theme in the 2016 presidential election (e.g., Editorial
Board, 2016). Often, this affective aversion to experts is conceptualized as a
form of anti-intellectualism (Hofstadter, 1963; Rigney, 1991). Throughout
the 2016 presidential campaign, and in the months that followed, Republican
nominee Donald Trump, for example, made several notable appeals to anti-
intellectual sentiments.
For example, Trump questioned on several occasions whether or not cli-
mate scientists were secretly working with Chinese business interests to
falsely promote evidence of climate change, hinted at researchers’ ulterior
motives in producing research about the safety of vaccines, and called
researchers “idiots” for creating (and advocating the use of) environmentally
friendly but potentially carcinogenic lightbulbs (Swain & Wong, 2016). He
also repeatedly questioned whether or not polling methodologists were con-
spiring to sink his presidential bid by artificially deflating his standing in the
national polls (Golshan, 2016). Similar rhetoric was present too in Trump’s
transition to the White House, when a senior campaign advisor characterized
federal climate research programs as “politicized science” (Milman, 2016).
Overall, experts received a considerable amount of criticism on and off the
campaign trail in 2016.
Yet, while anti-intellectualism may seem like it has been a unique feature
of Donald Trump’s rhetoric in the 2016 campaign, recent research (e.g.,
Gauchat, 2012) suggests that anti-intellectual attitude endorsement has been
growing in the mass public for decades, especially on the ideological right.
Less well understood, however, are the political implications of growing anti-
intellectualism. Does voters’ distrust and dislike of experts fuel skepticism
about scientific consensus relevant to public policy matters? And might it
increase support for politicians and political movements who share their anti-
intellectual views? Moreover, can factors like individuals’ levels of education
and cognitive skill can mitigate holding these attitudes?
In this article, I advance two theories about the political relevance of mass
anti-intellectualism in American political life. First, while the causes of anti-
intellectualism have received some recently scholarly attention, few have
asked how negative affect toward scientists and experts might shape public
opinion and political behavior. The spillover effects thesis contends that neg-
ative affect toward experts makes citizens more likely to deny matters of
scientific consensus, and to lend support for campaigns and political move-
ments that are skeptical of experts and the research they produce.
Second, I suggest that it may be possible to discourage citizens from hold-
ing anti-intellectual attitudes. This possibility has received some scholarly
attention in recent years, with mixed empirical results. The mitigation thesis
Motta 467

introduces verbal intelligence as a potential mitigating factor and proposes


that holding high levels of verbal intelligence should decrease the endorse-
ment of anti-intellectual attitudes.
Making use of data from the General Social Survey (GSS; 1972-2014),
and a four-wave panel from the Center for the Study of Political Pyschology
(CSPP) at The University of Minnesota I find empirical support for both the-
ories. In both 2016 and in years prior, anti-intellectualism is associated with
decreased acceptance of anthropogenic (i.e., human-caused) climate change,
as well as the safety of nuclear power (both of which are matters of broad
scientific agreement). I also find that anti-intellectualism is strongly associ-
ated with support for politicians and political movements who made the dis-
trust of experts prominent components of their campaigns; including voting
for George Wallace in 1968, holding positive views and voting for Donald
Trump in 2016, and supporting the United Kingdom’s decision to leave the
EU (also known as “Brexit”). These results hold even when controlling for
factors that may also predict support for the same outcomes (e.g., negative
racial and gender attitudes, authoritarianism), and panel analyses suggest that
they are robust to reverse-causal accounts.
Last, and perhaps most importantly, I put the mitigation thesis to the test in
several iterations of the GSS by assessing the effect of verbal intelligence on
anti-intellectual attitudes. The results show that individuals with high levels of
verbal intelligence are considerably less likely to hold anti-intellectual attitudes
than individuals with low levels of verbal intelligence. The mitigating power of
verbal intelligence is uniquely strong, as the results also show that educational
attainment—often hypothesized to decrease anti-intellectualism, with mixed
empirical results—plays a significantly weaker role in doing so. I conclude by
discussing the importance of these results in contemporary American political
life and noting several directions for future research.

Anti-Intellectualism and Its Historical Origins


In his well-known book, Richard Hofstadter (1963) identified three forms of
anti-intellectual thought present in American public life since the nation’s
founding (see Rigney’s, 1991, typology). Two of these have only limited
applicability to politics and elections. The first, anti-rationalism, is the rejec-
tion of critical thought as a desirable quality for individuals to hold. The
second, unreflective instrumentalism, describes the preference for short-term
payoffs irrespective of carefully reasoned long-term consequences. While
these types of anti-intellectual thought may plausibly be related to support
uncritical political rhetoric and for policies with immediate payoffs, the link
to electoral politics and public opinion is somewhat unclear.
468 American Politics Research 46(3)

However, a third type of anti-intellectualism he identified has a more


obvious connection to politics and public opinion—what Rigney terms the
anti-elitist form of anti-intellectualism. Broadly, this can be thought about in
affective terms. This form of anti-intellectualism refers to the distrust, and
perhaps even dislike, for individuals who claim to have superior knowledge
or wisdom about a subject matter. When I refer to anti-intellectualism in this
article, I am referring to the anti-elitist conceptualization of the term—nega-
tive affect toward scientists and experts.
This form of anti-intellectualism is thought to have a rich historical con-
nection to American elections and public opinion. At several moments in
American history, presidential candidates have attempted to capitalize on the
public’s distrust of politicians and other elites by employing this type of anti-
intellectual rhetoric. For example, in reaction to gains in mass educational
attainment during the Progressive era, Theodore Roosevelt was successful in
distancing his naturalistic and experiential “education” from that of allegedly
effeminate male intellectuals (Hofstadter, 1963; Rigney, 1991). Rigney notes
that similar tensions existed between John Quincy Adams and Andrew
Jackson, as well as in George Wallace’s unsuccessful presidential campaign
as an American Independent. Although he is perhaps best known for his
stances on segregation, Wallace famously dismissed college professors, jus-
tices, and policy experts as “pointy-headed intellectuals” on several occa-
sions. More recently, George H. W. Bush attempted to distance himself from
Michael Dukakis by linking him to intellectuals with “fastidious disdain” for
popular forms of religious tradition (Bonikowski & Gidron, 2016).
In 2016, anti-intellectualism appeared frequently in American political dis-
course, although whether or not it shaped voters’ attitudes and behaviors is not
completely understood. As noted earlier, Donald Trump’s winning presiden-
tial bid often denigrated and cast doubt on experts. This rhetoric may have had
some amount of popular appeal. Oliver and Rahn’s (2016) early and important
research on the subject shows that citizens’ distrust of experts was correlated
(r = .35) with support for Donald Trump. Similar rhetoric appears to have been
present in United Kingdom’s decision to leave the EU. Member of Parliament
and outspoken Brexit advocate Michael Gove, for example, drew headlines in
the summer of 2016 for claiming that people in the United Kingdom “have
had enough of experts,” referencing warnings from European economists
about the potential financial dangers of leaving the union.1 Gove also noted
that he was “glad” to not have experts on his campaign’s side,2 and para-
phrased Donald Trump in asking voters to tell “economic authorities” in the
EU “you’re fired.” Even if Brexit leaders had not directly invoked the lan-
guage of Trump, the United Kingdom’s decision to leave had crossed over the
Atlantic to become a salient political issue in the United States. Notably,
Motta 469

Donald Trump called himself “Mr. Brexit,” prompting many comparisons


between his candidacy and the movement in the United Kingdom (e.g., North,
2016).
As these above examples illustrate, political elites—from time to time—
appeal to citizens’ mistrust of experts. Hofstadter suspects that politicians do
this because they expect it to be an effective rhetorical strategy, leading those
holding anti-intellectual attitudes to rally behind those who are averse to
experts and their research. Whether or not this anti-intellectual rhetoric actu-
ally influences citizens’ political attitudes and behavior, however, remains an
understudied empirical question.
To proceed with putting this possibility to the test, though, it is first impor-
tant to consider several properties of the distribution of anti-intellectual atti-
tudes in the mass public. If anti-intellectualism is a strictly ideological
phenomenon, its applicability to politics is somewhat limited, affecting citi-
zens’ attitudes and behaviors only inasmuch as it characterizes whether or not
individuals consider themselves to be liberals or conservatives. If, on the
contrary, both liberals and conservatives hold anti-intellectual attitudes, even
if not to the same degree, studying these attitudes themselves (and whether or
not they can be mitigated) is a far more interesting empirical question.

The Polarization of Anti-Intellectualism


Recently, the ideological polarization of attitudes toward experts has received
important attention from scholars across the social sciences. In the aggregate,
ideological conservatives’ levels of trust in the scientific community have
decreased gradually since the early 1990s, while liberals have maintained
moderately positive views since the early 1970s (Gauchat, 2012).
Unsurprisingly, at the individual level, conservatism has been shown to be a
strong predictor of distrust toward scientists and experts (Gauchat, 2012;
Nisbet, Cooper, & Garrett, 2015). Relatedly, conservatism has also been
shown to be associated with skepticism about the research experts produce
(for an extensive review, see Lewandowsky & Oberauer, 2016), even on
issues where consensus does not contradict or even supports conventional
conservative values and political opinions (e.g., the mass production of
genetically engineered food; Blank & Shaw, 2015).
While much empirical research seems to demonstrate higher levels of
anti-intellectualism among ideological conservatives, the broader cognitive
underpinnings of these effects are less straightforward. Some believe that
distrust in science is a predominantly conservative phenomenon (e.g.,
Mooney, 2012), as conservatives tend to hold cognitive styles associated with
less scrupulous information processing than liberals (e.g., Jost, Glaser,
470 American Politics Research 46(3)

Kruglanski, & Sulloway, 2003). However, Kahan (2012) finds little differ-
ence between liberals and conservatives on this score and shows that conser-
vatives who do engage in effortful cognitive processing are more likely to
engage in motivated reasoning.
The link between cognitive processing and motivated reasoning is important,
as many scholars have conceptualized scientific skepticism as a form of ideo-
logically motivated reasoning. Individuals are thought to be at least somewhat
aware of the parties’ positions on scientific matters. Because belief change is
cognitively taxing, individuals prefer to expend less cognitive effort in service
of the defense of the prior ideological commitments (Kraft, Lodge, & Taber,
2015; Lodge & Taber, 2013; see also Kunda, 1990). Consistent with this view,
Fowler and Gollust (2015) find that, as political elites polarize on newly intro-
duced issues pertaining to science (e.g., human papillomavirus vaccination
requirements in 2006-2007), support for related programs and trust of medical
experts more generally tends to decline. Liberals and conservatives alike have
been shown to take part in partisan-motivated reasoning of this variety, becom-
ing more trusting of scientists (Nisbet et al., 2015) and policies concerning the
research they produce (Bolsen, Druckman, & Cook, 2015) when the political
context in which information is presented matches their prior partisan commit-
ments. Nisbet and colleagues (2015) suggest that the contemporary prevalence
of anti-intellectual thinking among conservatives may simply be due to political
context, such that more matters of expert consensus run more contrary to con-
servative values and policy principles than they do for liberals.
What all of this research critically implies is that anti-intellectual attitudes
may not solely be the by-product of citizens’ partisan and ideological com-
mitments. While anti-intellectualism is undeniably affected by partisanship
and ideology, it also has the potential to exert its own influence on public
opinion. Even though conservatives have recently come to hold lower levels
of trust in scientific experts, Gauchat’s (2012) analysis reveals that levels of
trust among liberals have been only moderately positive. While the polariza-
tion of anti-intellectual attitudes may be a partisan phenomenon, some con-
servatives may not hold these attitudes and some liberals might also hold
some distrust toward experts.
Demonstrative of this point, Figure 1 extends Gauchat’s (2012) aggregate
analysis of public trust in the scientific community, which concluded in 2010,
through the most recent wave of the GSS (2014). As Gauchat does, the left-
hand panel plots mean levels of trust in the scientific community (a 3-point
scale ranging from 0 to 2, where a score of 0 indicates holding “hardly any”
trust and a score of 2 indicates holding “a lot” of trust) over time. Because
mean values on an ordinal scale can be difficult to interpret substantively, the
right-hand panel does the same using an alternate measure of trust in the
Motta 471

Figure 1.  The ideological polarization of anti-intellectualism.


Note. Trend lines in the left-hand panel are a reanalysis of Gauchat (2012) and display average
levels of trust toward the scientific community (on a scale ranging from 0 to 2). Those in
the right-hand panel display the proportion of individuals holding high levels of trust toward
scientists. Trend data are unweighted (see Gauchat, 2012).

scientific community: the proportion (ranging from 0 to 1) of respondents


selecting the highest possible score on that measure.
The figure shows that conservatives have become more anti-intellectual
over time and that polarization has continued in the years following Gauchat’s
(2012) analysis. Yet, it is also important to note that liberals’ mean levels of
trust have rarely exceeded a mean of 1.5 on the 3-point scale. Likewise, the
proportion holding “a lot” of trust in scientists has hovered around just 50%
over the past 40 or so years.
Thus, while the polarization of anti-intellectual attitudes is undoubtably
more common among conservatives in recent years, liberals too hold some
distrust toward experts. This indicates that anti-intellectualism is not simply a
proxy for ideology and might, above and beyond partisan and ideological com-
mitments, play an important role in shaping political attitudes and behavior.3

The Political Implications of Anti-Intellectualism


Although anti-intellectualism’s political relevance likely extends beyond
prior partisan commitments, its relationship to public opinion and voter
472 American Politics Research 46(3)

behavior is not well understood. Drawing on recent public opinion research


and Hofstadter’s theoretical insights, I suspect that anti-intellectual attitudes
may “spill over” to shape citizens’ political attitudes and behaviors on at least
two fronts. The first pertains to matters of public policy, while the second
focuses on politicians and political movements.
First, anti-intellectualism, because it reflects public distrust of experts,
should be associated with opposition to policy positions that are related to
subjects of expert consensus. For example, while nearly all climate scientists
agree that changes in the climate can be ascribed to human activities (Cook
et al., 2013), this consensus has become politicized (Bolsen & Druckman,
2015), with more than a third of Americans do not attribute increases in the
Earth’s temperature to human activity (Saad & Jones, 2016). Individuals
holding anti-intellectual attitudes should be less willing to acknowledge this
matter of consensus, and policy positions related to it, given their negative
feelings toward experts.
Another example concerns nuclear energy. Energy experts agree that
nuclear power poses only minimal risks to the safety of the environment and
those living nearby power plants (Environmental Protection Agency [EPA],
2016), but public opinion on the safety of nuclear power is quite mixed
(Bolsen & Cook, 2008). Highly anti-intellectual members of the mass public
might be unwilling to accept that the energy source is safe and resistant to its
viability as an alternative energy source.
While there are certainly other examples worth studying, the empirical
tests I describe shortly focus on these issues for two key reasons. First, and
most importantly, liberals and conservatives differ in the extent to which they
express skepticism about global warming and nuclear power safety. This
variation is critical. If anti-intellectualism shapes opinions about scientific
consensus independently of partisanship, it should do so for issues that
engender different levels of skepticism between liberals and conservatives.
Demonstrative of this point, Democrats have been shown to hold more nega-
tive attitudes toward nuclear energy than do Republicans (Riffkin, 2016),
whereas Republicans are more likely to deny human-caused climate change
(Saad & Jones, 2016). Second, and more pragmatically, these issues were
readily available in multiple waves of high-quality national survey data.
If anti-intellectualism is merely a reflection of partisanship, opposition to
positions like these should primarily be explained by citizens’ partisan and
ideological preferences. However, as argued earlier, there is good reason to
suspect that anti-intellectual attitudes exist on both the right and left, even
though they have become increasingly common on the right. Consequently, I
hypothesize that individuals who hold anti-intellectual attitudes will be more
likely to reject opinions and policies that reflect matters of expert agreement.
Motta 473

Hypothesis 1a: Spillover Effects (Expert Consensus Denial)—Increased


levels of anti-intellectualism will be associated with increased opposition
to policy positions and other political beliefs that are informed by scien-
tific and expert consensus, including belief in the danger of global warm-
ing and the safety of nuclear energy.

Second, citizens’ aversion to scientists and experts may lead them to


support politicians and political movements who share their distrust of
expert opinion and/or expressly deride it. As noted earlier, these types of
sentiments were prominently featured in George Wallace’s unsuccessful
1968 presidential bid, as well as Donald Trump’s victory in 2016.
Politicians in the United Kingdom also made use of anti-intellectual rheto-
ric in justifying their decision to leave the EU, which may have encour-
aged Americans holding negative attitudes toward experts to sympathize
with the cause.
If, as Hofstadter predicts, anti-intellectual rhetoric is a tool by which poli-
ticians can try to sway public opinion in their favor (see also Bonikowski &
Gidron, 2016), citizens’ distrust of experts should be associated with support
for expert-averse politicians and political movements. Indeed, Oliver and
Rahn (2016) take an important first step in testing this possibility by offering
correlational evidence of a link between anti-intellectualism and support for
Donald Trump in the early stages of his candidacy. However, whether or not
this finding holds up across other contexts, and in multivariate analyses, is in
need of further investigation.
Of course, the appeal of individuals like Wallace and Trump, and move-
ments like Brexit, is not limited to their attitudes toward experts. For exam-
ple, recent research (e.g., Schaffner, MacWilliams, & Nteta 2016) has linked
support for Donald Trump to citizens’ hostile sexism (Glick & Fiske, 1996)
and racial resentment (Kinder & Sanders, 1996). Others have noted that citi-
zens’ authoritarian dispositions (MacWilliams, 2016) might have similar
effects.
Whether or not anti-intellectualism increases support for Wallace, Trump,
and Brexit independently of these other factors is an open empirical question,
and one I put to the test. With that in mind, I propose Hypothesis 1b.

Hypothesis 1b: Spillover Effects (Support for Expert-Averse Politicians


and Movements)—Increased levels of anti-intellectualism will be asso-
ciated with increased support for expert-skeptic politicians and political
movements, including candidates George Wallace and Donald Trump,
and the United Kingdom’s referendum on whether or not to leave the
EU.
474 American Politics Research 46(3)

Anti-Intellectualism and Potential Mitigating


Factors
Scholars have suggested a number of factors that might mitigate holding anti-
intellectual attitudes, and (relatedly) skepticism about expert consensus, typi-
cally focusing on the ideological right. While some factors have been shown
promise in mitigating scientific skepticism, none have consistently been
shown to dissuade citizens from holding anti-intellectual attitudes.
For example, the “deficit model” of trust in science (Gauchat, 2008, 2012;
Sturgis & Allum, 2004) argues that conservatives are simply less educated
and knowledgeable about basic scientific facts than liberals, which holds
them back from exhibiting higher levels of support for scientists and experts.
However, researchers have documented only modest associations between
education, knowledge, and attitudes about science and scientists (Allum,
Sturgis, Tabourazi, & Brunton-Smith, 2008; Gauchat, 2008; Lewandowsky
& Oberauer, 2016). Some even show that education has been shown to
increase the extent to which conservatives distrust the scientific community
(Gauchat, 2012).
In a similar vein, higher levels of quantitative reasoning skills (Kahan,
Peters, Dawson, & Slovic, 2013), knowledge of basic scientific facts (Gauchat,
2008; Kahan et al., 2012), and depth of cognitive processing (Kahan et al.,
2012) have been shown to increase (or fail to decrease) motivated reasoning.
Some scholars have been successful, however, in “turning off” skepticism
about science by encouraging subjects to hold accuracy motivations when pro-
cessing new information about the issue (e.g., being held accountable to relay
facts about the issue to one’s peers) in laboratory experiments (Bolsen et al.,
2015). Others have found similar de-biasing effects among individuals who
exhibit high levels of curiosity, but not high levels of knowledge, about science
(Kahan, Landrum, Carpenter, Helft, & Jamieson, 2017).
Collectively, these results underscore a growing scholarly interest in
understanding which factors make citizens more or less likely to hold anti-
intellectual attitudes and to hold related antiscience attitudes more broadly.
However, their implications for the mitigation of holding anti-intellectual
attitudes are unclear. One way to make sense of this research is to group
potential mitigating factors in terms of motivation and ability. Whereas moti-
vational constructs, like citizens’ interest in science and willingness to pursue
accuracy goals in information processing, can have mitigating effects, they
have not been studied extensively in the domain of anti-intellectualism. In
contrast, knowledge about science and various types of cognitive skills have
received more attention in this regard, but it is unclear whether they exacer-
bate negative feelings toward scientists or inhibit them.
Motta 475

I suggest that an important ability-related factor has been overlooked in


prior research as a potential mitigator of anti-intellectualism—verbal intelli-
gence. Verbal intelligence refers to a collection of cognitive skills that reflect
one’s ability to broadly demonstrate comprehension of one’s vernacular lan-
guage (Dawson, 2013). Often, it is measured using short vocabulary tests like
Wordsum (Thorndike, 1942; see Malhotra, Krosnick, & Haertel, 2007, for a
review of its usage in political science). While some worry that the Wordsum
test makes use of antiquated words (Dorius, Alwin, & Pacheco, 2016), it has
shown to be a valid indicator of more general measures of intelligence (Miner,
1957; see also Malhotra et al., 2007) and is suitable for implementation in
online and mixed mode surveys (Gooch, 2015).
Verbal intelligence is a sensible candidate to mitigate holding anti-intel-
lectual attitudes, because it is thought to provide citizens with what scholars
term “communicative skills” (Condon, 2015). Communicative skills include
individuals’ abilities to think abstractly about political concepts, eschew heu-
ristics that are grounded in negative affect and prejudice toward marginalized
groups, and discuss politics with others. One communicative skill stands out
as being particularly relevant to anti-intellectualism, the unwillingness to rely
on cognitive shortcuts that stereotype or deride certain social groups when
formulating opinions about them (e.g., Deary, Batty, & Gale, 2008; Dhont &
Hodson, 2014; Nie, Junn, & Stehlik-Barry, 1996).4
Stereotypes about scientists and experts can lessen the cognitive workload
individuals expend to formulate judgments about those groups, which may
make reliance on them a particularly attractive strategy for those with lower
verbal ability (Dhont & Hodson, 2014). In contrast, individuals with greater
verbal abilities may be willing and able to expend more cognitive effort to
arrive at more nuanced views about the scientific community. Indeed, highly
verbally intelligent individuals have been shown to be more likely to hold
“open-minded” cognitive styles (DeYoung, Peterson, & Grazioplene, 2012;
Stanovich & West, 1997). In politics, open-minded thinking implies a will-
ingness to entertain alternatives to prior preferences, widely held values, and
pervasive stereotypes (Price, Ottati, Wilson, & Kim, 2015). A cognitively
effortful and open-minded approach to formulating judgments about scien-
tists and experts may recognize that some experts, under some conditions,
make factual errors or give in to self-interested motives, despite being gener-
ally trustworthy. In contrast, a less effortful approach, relying on the heuristic
that experts are uniformly untrustworthy, need not consider these potential
sources of variability and ambiguity.
Overall, then, verbal intelligence is a plausible and understudied candidate
to mitigate anti-intellectual attitudes in the mass public. In contrast to studies
focusing mainly on the ideological right, there is good reason to suspect that
476 American Politics Research 46(3)

this may be true should be true for both liberals and conservatives. While
conservatives have become more likely to hold anti-intellectual attitudes in
recent years, and therefore exhibit the greatest potential for change, Figure 1
demonstrates that liberals too hold at least some skepticism toward experts.
Both groups therefore have at least some potential for change.
Consequently, in testing the Mitigation Hypothesis, I consider whether or
not verbal intelligence’s association with increased positivity toward scien-
tists and experts differs across citizens’ ideological preferences. With this in
mind, I propose Hypothesis 2a:

Hypothesis 2a: Mitigation—Individuals who are highly verbally intelli-


gent—irrespective of their ideological preferences—should be less likely
to endorse anti-intellectual attitudes than those with lower levels of verbal
intelligence.

One important objection to this hypothesis may be that holding high levels
of verbal intelligence may actually exacerbate anti-intellectual attitudes for
some individuals. As mentioned above, nonverbal measures of intelligence
have indeed been shown to increase partisan polarization on matters pertain-
ing to science and expert opinion (e.g., Kahan, 2012; Kahan et al., 2013).
Individuals with high levels of verbal intelligence tend to be highly capable
of debating and counterarguing (Delia, Kline, & Burleson, 1979), which may
make them particularly good at crafting arguments and internal justifications
in service of their prior beliefs. Consequently, it may be the case that verbal
intelligence exacerbates holding anti-intellectual attitudes, among those most
likely to distrust and dislike experts, such as ideological conservatives
(Gauchat, 2012). With this alternative perspective in mind, I propose the
Exacerbation Hypothesis:

Hypothesis 2b: Exacerbation—Individuals who are highly verbally intel-


ligent will be more likely to endorse anti-intellectual attitudes than those
with lower levels of verbal intelligence, but only among ideological
conservatives.

While the Exacerbation Hypothesis poses a plausible alternative to the


Mitigation Hypothesis, there are important theoretical reasons to suspect that
verbal intelligence is more likely to mitigate holding anti-intellectual atti-
tudes. If verbal intelligence tends to exacerbate holding stereotypes and other
negative attitudes toward various social groups, we might expect to see that
ideology moderates the effect of verbal intelligence on attitudes toward these
groups. According to this view, verbal intelligence would decrease holding
Motta 477

these attitudes among liberals but increase them among conservatives.


However, recent work in social psychology suggests that—in addition to the
large direct effects of verbal intelligence (and other forms of cognitive abil-
ity) on mitigating stereotype holding—the relationship between ideology
negative attitudes toward various social groups tends to be not moderating
but mediating (Deary et al., 2008; Dhont & Hodson, 2014), such that verbal
intelligence boosts holding socially liberal attitudes, which in turn decrease
holding racially prejudicial attitudes. Thus, while the nature of the relation-
ship between verbal intelligence, ideology, and anti-intellectual attitudes is
certainly an open empirical question, I suspect that verbal intelligence will
play a mitigating, and not exacerbating, role with respect to holding anti-
intellectual attitudes.

Data and Measures


Data
The data used in this study come from two sources. One is the GSS cumulative
time-series file, which surveyed nearly 60,000 individuals between 1972 and
2014. While a few of the years feature full probability samples of the voting age
population, sampling methods usually differed across years. Therefore, all data
are weighted using the WTSSALL variable, as recommended in the user guide
(Appendix A of the Supplementary Materials) . The second dataset is a 2016
Election Panel from the Center for the Study of Political Psychology (CSPP) at
the University of Minnesota. Respondents were recruited through Survey
Sampling International (SSI), who emailed thousands of individuals to partici-
pate in the study. In July of 2016, 3,550 individuals successfully completed the
first wave of the study, all of whom had the opportunity to take part in three
follow-up waves (conducted in September and October). The panel had a built-
in attrition rate of 43% in Wave 2 (n = 2,020) and 63% in Wave 3 (n = 1,234).
All data were weighted to known population benchmarks with respect to age,
gender, race, ethnicity, income, and education.

Measures
This study makes use of several different outcome and independent variables
across a number of datasets, which I describe below. Summary statistics and
more detailed information can be found in the Supplementary Materials.

Anti-intellectualism.  The key independent variable in the spillover tests, and


key dependent variable in the mitigation tests, is individuals’ affective
478 American Politics Research 46(3)

aversion to scientists and experts. In the GSS, respondents were asked to


denote how much confidence they have in the scientific community (Gau-
chat, 2012)—3-point scale ranging from “a great deal” of confidence to
“hardly any,” and recoded to range from 0 to 1 when employed as an indepen-
dent variable (otherwise, left in its natural metric). The second item, in the
CSPP study, is derived from the anti-expert domain of a broader measure of
populism developed by Oliver and Rahn (2016). Respondents were asked
whether they agree or disagree with placing trust in the wisdom of ordinary
people, as opposed to experts and intellectuals. The 5-point scale (ranging
from “strongly disagree” to “strongly agree”) was again recoded to range
from 0 to 1, such that 1 reflects strong agreement.
Here, some might object that these two measures of anti-intellectual-
ism might capture different (albeit conceptually related) constructs. This
is an important point but one that underscores a potential strength of the
strategy. If these measures collectively point to similar results, despite
their differences in focus and construction, they should provide further
confidence that the broader anti-intellectualism concept is an important
determinant of voter behavior and public opinion. To establish the relat-
edness of the measures, Table S2 in the Supplemental Materials considers
several demographic and political predictors of holding anti-intellectual
attitudes, and whether or not they vary across datasets. The table shows
that the correlates of anti-intellectualism are quite similar across mea-
surement strategies.
Another question some might raise is whether or not these measures of
anti-intellectualism might be standing in for another related construct, such
as antiestablishment attitudes or populism more generally. Whereas some
conceptualize anti-intellectualism as one of several components of populism
(e.g., Oliver & Rahn, 2016), others have noted that academics and other
experts have—at times—aligned themselves with populist movements and
anti-elite ideals (Rigney, 1991). Thus, there is plenty of room for conceptual
debate as to whether or not anti-intellectualism is primarily a facet of popu-
lism, a distinct construct, or some mix of the two.
While settling this conceptual debate falls outside the purview of this
research, I can at least empirically isolate the effects of anti-intellectualism,
above and beyond measures of these other related factors. The GSS contained
measures that could be used as indicators of anti-elitism, whereas the CSPP
study had a few (but not all) remaining items of Oliver and Rahn’s (2016)
Populism Scale. These analyses, and information about how these variables
were measured, can be found in Appendix C of the Supplementary Materials.
As I review shortly, the results suggest that the effects of anti-intellectualism
documented in this article are not explained away by these other factors.
Motta 479

Belief in policy-relevant scientific consensus.  The key outcomes for Hypoth-


esis 1a include variables that pertain to citizens’ belief in the existence
and severity of climate change, and their views on the safety of nuclear
power. In the GSS, respondents were asked whether they think that the
greenhouse effect is dangerous (1) in general or (2) to their family, on an
ordered 5-point scale ranging from “extremely dangerous” to “not dan-
gerous.” In the CSPP studies, respondents were asked to select one of two
possible causes of climate change (“human activities” or “natural
changes” in the environment), or indicate that they believe that there “has
been no change in the climate.” Responses were dichotomized, such that
a score of 1 indicates believing in human-caused climate change (0,
otherwise).

Support for expert-averse politicians and political movements.  The key out-
comes for Hypothesis 1b include variables that denote respondents’ affec-
tive or electoral support for expert-averse politicians and associated
political movements. In the 1972 GSS, I create a measure of 1968 vote
choice for American Independent George Wallace, dichotomized so that a
score of 1 indicates voting for him, and 0 indicates voting for the Repub-
lican or Democratic nominee. I do the same in the 2016 CSPP study, cre-
ating a variable where a score of 1 indicates voting for Donald Trump, and
0 indicates voting for Hillary Clinton. I also include measures of differen-
tial affect for Trump over Clinton by subtracting scores on 101 point
Trump feeling thermometers (ranging from 0 to 100) from those of Hill-
ary Clinton; in the CSPP study, the resulting scale was recoded to range
from 0 to 1, such that a score of 1 (0) indicates high differential affect for
Trump (Clinton), respectively, and a score of 0.5 indicates rating the two
similarly. Respondents were also asked whether they support, oppose, or
have not heard enough about the United Kingdom’s decision to leave the
EU (Brexit). Responses to this question were fashioned into a dichoto-
mous indicator of Brexit support (taking on a value of 1 if they support it,
and 0 otherwise).5

Verbal intelligence.  Verbal intelligence is the key independent variable in this


article’s mitigation tests and is scored by summing up individuals’ scores on
10 multiple-choice vocabulary questions (where individuals are given a word
and asked to select a word that most closely resembles the initial word).This
variable was available in most waves of the GSS and is recoded to range from
0 to 1 (such that a score of 1 reflects correctly identifying all 10 synonyms).
The specific words are not released to the public, but I include an example in
the Supplementary Materials (Appendix A).
480 American Politics Research 46(3)

Demographic and political controls.  All models control for respondents’ ideol-
ogy (0-1, where 1 indicates being extremely conservative), partisan identity
(dummy variables indicating self-identifications with the Democrats or
Republicans), strength of partisan identity (rescaled to range from 0 to 1,
such that 1 reflects strong partisanship), race (1 if White, 0 if non-White),
total family income (recoded to range from 0 to 1), education (dummy vari-
ables indicating whether or not individuals completed high school, some col-
lege, college, or postcollege degrees), age (recoded to range from 0 to 1, such
that an age of 1 reflects the oldest person in the dataset), gender (1 if female,
0 if male), and labor force status (1 if working, 0 if not working). As noted
earlier, the CSPP models also control for a number of factors that might alter-
natively explain support for Brexit and Donald Trump, including respon-
dents’ knowledge about civics (0-1, such that a score of 1 indicates holding
high knowledge), racial resentment (0-1, such that a score of 1 indicates high
resentment), hostile sexism (0-1, where 1 indicates holding high levels of
ambivalent sexism), authoritarianism (0-1, where 1 indicates being highly
authoritarian), and feeling thermometer toward Muslims (recoded to range
from 0 to 1, such that 1 indicates highly positive affect).

Results
The Political Implications of Anti-Intellectualism
To test the first component of spillover effects thesis (Hypothesis 1a), I
regress each indicator of belief in policy-relevant scientific consensus on
anti-intellectualism (and a host of demographic and political controls). The
full results of each model can be found in the appendix (Table A1) and are
summarized in Table 1.
In all seven models, anti-intellectualism was significantly (p < .05 in all
cases) associated with decreased belief in expert consensus with respect to
climate change and nuclear power safety. Because the models vary in their
estimation strategies, it is simplest to talk about these results in terms of lin-
ear predictions (ordinary least squares) and predicted probabilities (binary
and ordered maximun likelihood models). Movement from the minimum to
maximum observed values of anti-intellectualism in the 1993, 1994, and
2000 GSS (i.e., a “first difference”) was associated with an 8% decrease in
the likelihood of believing that the greenhouse effect poses a general risk to
the population, as well as to one’s own family (9%). Similar movement on the
anti-intellectualism variable was associated with a 7% increase in the belief
that nuclear power is extremely dangerous in general, and a 6% increase in
the belief that nuclear power poses threats to one’s family. In the CSPP study,
movement from the minimum to maximum value on the anti-intellectualism
Motta 481

Table 1.  Summary of Effects From Tables A1 and A2.


ΔMinimum–
Dataset β SE p Minimum Maximum Maximum

Table A1
  Pr(Greenhouse = Ext. GSS 93, 94, 00 0.69 0.17 * 0.20 0.12 −8%
Dangerous, in general)
  Pr(Greenhouse = Ext. GSS 93, 94, 00 0.66 0.19 * 0.19 0.10 −9%
Dangerous, to family)
  Pr(Nuke Power = Ext. GSS 93, 94 −0.47 0.17 * 0.19 0.26 +7%
Dangerous, in general)
  Pr(Nuke Power = Ext. GSS 93, 94 −0.44 0.19 * 0.16 0.22 +6%
Dangerous, to family)
  Pr(Climate Change CSPP W1 −0.81 0.22 * 0.66 0.48 −18%
Cause = Humans)
  Pr(Climate Change CSPP W2 −0.94 0.33 * 0.68 0.51 −17%
Cause = Humans)
  Pr(Climate Change CSPP W3 −1.10 0.41 * 0.61 0.39 −22%
Cause = Humans)
Table A2
  Pr(Vote Wallace) GSS 1972 0.76 0.42 † 0.12 0.22 +9%
  Net Trump Support CSPP W1 0.06 0.02 * 0.44 0.49 +5%
  Net Trump Support CSPP W2 0.07 0.03 * 0.42 0.50 +9%
  Net Trump Support CSPP W3 0.08 0.04 † 0.43 0.51 +8%
  Pr(Vote Trump) CSPP W1 0.27 0.41 ns 0.41 0.44 +3%
  Pr(Vote Trump) CSPP W2 1.10 0.53 * 0.36 0.46 +10%
  Pr(Vote Trump) CSPP W3 0.34 0.60 ns 0.47 0.50 +3%
  Pr(Support Brexit) CSPP W1 1.02 0.24 * 0.25 0.44 +18%

Note. Predicted probabilities (rows prefixed with “Pr”) and linear predictions presented, based on
models in Tables A1 and A2. All other covariates are held at their sample means. The first three columns
reprint the parameter estimates (β), standard errors, and p values associated with the anti-intellectualism
variable in the appendix. The remaining columns denote the probabilities and scores predicted when
anti-intellectualism is held at its lowest and highest (respectively) observed values, followed by the
resulting change in effect size moving from the minimum to maximum values (ΔMinimum–Maximum). For the GSS
parameters and probabilities, the effects of anti-intellectualism in rows 1 and 2 and 2 to 4 are expected
to differ in sign (i.e., because scientific consensus views greenhouse emissions as dangerous, and nuclear
power as not dangerous). CSPP = Center for the Study of political psychology.GSS = General Social Survey.
†p < .10. *p < .05.

scale leads to an 18% reduction in the probability of believing that climate


change is caused by humans in the July wave of the CSPP study, a 17%
decrease in the September wave, and a 22% decrease in the November wave.
Anti-intellectualism also plays a key role in explaining why individuals sup-
port expert-averse movements and political candidates, supportive of Hypothesis
1b. In the eight models presented in Table A2 (summarized in Table 1), anti-
intellectualism was approached or surpassed conventional levels of significance
in six out of eight models. A first difference in anti-intellectualism was associated
482 American Politics Research 46(3)

with a 9% increase in the likelihood of voting for George Wallace in the 1968
presidential election, as well as a 10% increase in the likelihood of voting for
Donald Trump in 2016 (in the September, but not July or October waves of the
CSPP study). Similarly, and building on findings noting strong bivariate correla-
tions between expert distrust and Trump support, anti-intellectualism boosted
positive affect toward Trump in the multivariate models, relative to Clinton, by
5% in the July wave of the CSPP Study, 9% in the September wave, and 8% in
the October wave. Anti-intellectualism was also associated with a substantial
18% increase in support for the Brexit movement.6

Addressing Potential Confounds and Reverse-Causal Accounts


Notably, the effects summarized in Table 1 are robust to the inclusion of several
potential confounds. First, as can be found in Tables A1 and A2, they are not
explained away by individuals’ partisanship or ideological leanings, which we
might suspect if anti-intellectualism was simply a proxy for partisanship. The
results are also hold when including factors that might be thought to be relevant
in shaping vote choice in 2016 (MacWilliams, 2016; Schaffner, MacWilliams,
& Nteta, 2016), including citizens’ levels of racial resentment, hostile sexism,
affect toward immigrants and Muslims, and authoritarianism.
I also test whether or not antiestablishment or populist attitudes more gener-
ally confound the results displayed in Table 1. First, I add a measure of anties-
tablishment attitudes to all GSS models summarized in Table 1; relying on
items (constructed identically to the anti-intellectualism item) that assess atti-
tudes toward officials in all three branches of government, as well as the popu-
lar press. The results can be found in Table S4 in the Supplementary Materials
and suggest that anti-intellectualism is not empirically conflated with antiestab-
lishment attitudes.7 Second, in the CSPP study, I test whether or not the effect
of anti-intellectualism is explained away by the inclusion of three additional
items from the anti-elite domain of Oliver and Rahn’s (2016) Populism Scale
(see Table S5 for more information). The results (Table S5) show that the
effects of anti-intellectualism documented in Table 1 hold in every model.
Still, some might wonder if the results presented in the Trump support
models are confounded by a reverse-causal account. Perhaps support for
Trump, and his rhetoric about scientists and experts, encourages individuals
to hold anti-intellectual attitudes. While this account seems plausible, I find
little evidence in support of it.
Because the CSPP data contained measures of both Trump support and
anti-intellectualism in Waves 1 and 2, it was possible to create a cross-lagged
model of the relationship between changes in support for Trump and anti-
intellectual attitude holding. Figure S1 in the Supplementary Materials
reveals that anti-intellectual attitudes are generally stable over time
Motta 483

(β* = 0.39, p < .01) and are uninfluenced by support for Trump (β* = 0.03, p
> .10). Thus, a reverse-causal story seems unlikely.8

The Prevalence of Anti-Intellectual Attitudes


Another matter to consider, before moving on, is the prevalence of anti-intel-
lectual attitude holding in the United States. If few Americans distrust and/or
dislike experts, the consistency and size of these spillover effects would be of
less substantive interests. However, this does not appear to be the case.
In the 2014 wave of the GSS, for example, about 58% of respondents
(weighted) placed “only some” trust in the scientific community or less. In
the 2016 CSPP studies too, approximately 44% of the mass public (weighted)
agreed the claim that experts ought to be less trusted than ordinary people.
Because many Americans hold the types of attitudes that Table 1 suggests
spill over to shape public opinion, it is worthwhile to consider how these
attitudes might be mitigated.

The Mitigating Power of Verbal Intelligence


The mitigation thesis outlined in this article suggests that verbal intelligence
should discourage citizens from holding anti-intellectual attitudes, irrespective
of their ideological commitments. I construct two maximum likelihood models
to test this possibility (Hypothesis 2a), as well as the alternative exacerbation
thesis (Hypothesis 2b). Because I use maximum likelihood estimation in these
models (ordered logistic regression), I present odds ratios (eβs) and formal
Wald coefficient tests to facilitate effect size comparison across variables. To
demonstrate the strong mitigating power of verbal intelligence, I also compare
these effects with those of another hypothesized mitigator of anti-intellectual
attitudes—educational attainment (which, as noted earlier, may either increase
anti-intellectual attitudes or yield a smaller/null effect on them).
The first column of Table 2 tests whether or not verbal intelligence and/or
educational attainment mitigates holding anti-intellectual attitudes. This is
the most general test of the Mitigation Hypothesis. Model 1 regresses anti-
intellectualism on individuals’ levels of verbal intelligence, educational
attainment, and a series of demographic controls.
The second model compares whether or not the mitigating effect of ver-
bal intelligence varies across citizens’ ideological preferences, by interact-
ing citizens’ ideological preferences in the full sample with verbal
intelligence and educational attainment. To make this model as favorable as
possible to detect differences between groups—as the Mitigation Hypothesis
suspects that the effect of verbal intelligence on holding anti-intellectual
attitudes should not vary between liberals and conservatives, and the
484 American Politics Research 46(3)

Table 2.  The Mitigating Power of Verbal Intelligence.

DV = Anti-intellectualism Model 1 Model 2


Wordsum 1.01* 1.62*
(0.11) (0.48)
High School 0.17* −0.19
(0.05) (0.30)
Some College 0.26* −0.19
(0.09) (0.39)
College 0.70* 0.76*
(0.07) (0.34)
Post College 0.72* 0.58
(0.09) (0.38)
Conservatism −0.30* −0.01
(0.09) (0.63)
Income 0.33* 0.57*
(0.09) (0.19)
White 0.45* 0.28*
(0.05) (0.09)
Female −0.35* −0.45*
(0.04) (0.07)
Age −0.51* −0.51*
(0.09) (0.15)
Labor Force −0.16* −0.26*
(0.04) (0.08)
Democrat 0.27* 0.27*
(0.08) (0.13)
Republican 0.33* 0.29*
(0.08) (0.13)
PID Strength −0.03 −0.01
(0.08) (0.14)
Rel. Serv. Attend −0.29* −0.42*
(0.06) (0.10)
Wordsum × — −1.24
Conservatism   (0.87)
High School × — 0.43
Conservatism   (0.55)
Some College × — 0.53
Conservatism   (0.71)
College × Conservatism — −0.37
  (0.61)

(continued)
Motta 485

Table 2. (continued)

DV = Anti-intellectualism Model 1 Model 2


Post College × — 0.32
Conservatism   (0.70)
τ1 −2.14* −1.82*
(0.14) (0.38)
τ2 0.88* 1.29*
(0.13) (0.38)
n 14,880 5,097

Note. Unstandardized ordered logistic regression coefficients presented. Positive coefficients


indicate increased trust in the scientific community. Standard errors in parentheses. GSS
models include year fixed effects. DV = Dependent Variable; GSS = General Social Survey;
PID = Partisan Identity.
†p < .10. *p < .05.

exacerbation hypothesis suggests that it should—I limit analysis to all GSS


waves conducted after 1994. I do this because, as Figure 1 suggests, the
mid-1990s appear to be when liberals and conservatives diverged in their
attitudes about experts. Detecting no differences between groups when
including earlier waves of the GSS might provide misleading evidence in
favor of Hypothesis 2a, and against Hypothesis 2b, because liberals and
conservatives held similar attitudes toward scientists and experts for many
years in the series.
Turning to the first model in Table 2, the results indicate that both verbal
intelligence and, to a lesser extent, educational attainment played a role in
mitigating holding anti-intellectual attitudes.
The results show that higher scores on the Wordsum test were associated
with holding more positive feelings toward scientists (β = 1.01, p < .05).
Increases in educational attainment were also associated with increased trust
toward the scientific community, most notably, among individuals who com-
pleted college (β = 0.70, p < .05) or who obtained postsecondary degrees (β =
0.72, p < .05). However, these mitigating effects were significantly smaller for
individuals holding a graduate degree (eβ = 2.05; Wald χ2 = 3.31, p < .10) or
college degree (eβ = 2.01; Wald χ2 = 4.32, p < .05), compared with individuals’
levels of verbal intelligence (eβ = 2.74).
To more easily visualize the differences in these effects, Figure 2 plots
changes in the predicted probability of holding “a lot” of trust toward the sci-
entific community for the lowest and highest values of verbal intelligence
(i.e., answering all questions correctly vs. answering none correctly) and edu-
cational attainment covariates (i.e., each particular degree obtained, relative to
earning less than a high school degree). The results show that movement from
486 American Politics Research 46(3)

Wordsum

Post College

College

Some College

High School

0 .05 .1 .15 .2 .25

Change in Pr(Trust Sci Community “A Lot”)

Figure 2.  The mitigating power of verbal intelligence.


Note. Changes in predicted probabilities derived from ordered logistic regression models in
Table 2. All other factors are held constant.

the minimum to maximum value of verbal intelligence produced a substan-


tively larger (24%) change in trust toward the scientific community than did
holding a graduate degree (17% change).9 Overall, while both verbal intelli-
gence and educational attainment both mitigate holding anti-intellectual atti-
tudes, verbal intelligence appears to be the stronger mitigator, both
substantively and statistically.
At this point, some might wonder whether Model 1 has been misspecified.
Including verbal intelligence and educational attainment together in one model
might introduce posttreatment bias into the estimates, because individuals who
are more verbally intelligent may go on to obtain higher levels of schooling. Of
course, it could be the case individuals who are more highly educated also
acquire more verbal intelligence over the course of their schooling. Absent lon-
gitudinal data spanning large swaths of the life course, my data cannot deter-
mine the direction of a potential posttreatment effect. However, to reduce
concerns about potential bias in these models, I can at least rerun Model 1 to
estimate the effects of verbal intelligence and educational attainment indepen-
dently from one another. The results are displayed in Table S1 in the
Supplementary Materials and mirror the results displayed in Model 1. When the
model is run without the verbal intelligence covariate, the effects of completing
college (eβ = 2.51) or a graduate degree (eβ = 2.94) increase. But so too does the
effect of verbal intelligence (eβ = 4.35) when run without the educational attain-
ment variables. Again, the effect of verbal intelligence was significantly larger
than those of the educational attainment indicators in all cases (p < .05). Because
Motta 487

these results mirror those presented in Model 1, it seems unlikely that the inclu-
sion of both terms in the model is posing a misspecification issue.
Next, Model 2 tests whether or not these mitigating effects vary across citi-
zens’ ideological identities. As noted previously, the models restrict analysis to
only waves of the GSS taken following 1994, to more easily detect any potential
moderation effects of ideology. The results show, however, that neither the effect
of verbal intelligence (β = −1.24, p > .10) nor even the highest educational
attainment indicator (β = 0.32, p > .10) was moderated by ideology. The mitigat-
ing power of both sets of variables, therefore, appears to be the same for liberals
and conservatives, consistent with the Mitigation Hypothesis (Hypothesis 2a)
and inconsistent with the idea that verbal intelligence exacerbates holding anti-
intellectual attitudes among conservatives (Hypothesis 2b).
However, because collapsing educational attainment into indicator vari-
ables introduces several interaction terms into the model that are likely at
least somewhat collinear, standard error inflation might complicate the abil-
ity to detect potential interaction effects. To assess this possibility, I collapsed
the educational attainment indicators into a single quasi-interval variable,
ranging from 0 (less than high school) to 1 (a graduate degree). The results
produced substantively and statistically similar interaction effects for both
verbal intelligence (β = −1.06, p > .10) and education (β = −0.45, p > .10).

Discussion
The results presented in this study suggest that anti-intellectualism is associated
with several political attitudes and behaviors that are relevant to contemporary
political life in the United States. Anti-intellectualism discourages citizens from
siding with experts on matters of scientific consensus and can more broadly help
explain why voters support candidates who criticize and denigrate scientists and
experts. In most cases, these effects were substantively large and statistically
significant; even when assessing the effect of anti-intellectualism alongside a
number of other powerful determinants of voter behavior in 2016.
More importantly, the results suggest that while anti-intellectualism is far
from rare in the mass public, it does not appear to be immutable. This study
demonstrates that individuals holding high levels of verbal intelligence are
less likely to hold anti-intellectual views than those with lower levels of ver-
bal intelligence. These effects significantly exceeded those of educational
attainment and were not moderated by ideology, which are notable departures
from previous research. Generally, this article suggests that verbal intelli-
gence ought to earn more scholarly consideration as a potential mitigator of
holding anti-intellectual attitudes.
It is important to note that the distrust of experts is not a wholly deleterious
influence in American political life. Some amount of skepticism about experts’
488 American Politics Research 46(3)

motives and the quality of the research they produce can actually be constructive
in creating public policies informed by an accurate understanding of relevant
information. Anti-intellectualism differs from healthy skepticism, however, in
that it captures not only a broad distrust toward experts but dislike for those lay-
ing claim to expertise. The result, as demonstrated here, is that citizens eschew
heeding experts’ insight about the facts surrounding politically relevant matters
and lend support to candidates and political movements that are hostile to allow-
ing expert-based evidence to inform policy decisions.
Of course, this research is not without limitations. For example, while the
variables used here all attempt to measure negative affect toward scientists and
experts, their psychometric properties are not fully understood. In the future,
scholars should consider validating and implementing scales specifically
designed to measure anti-intellectualism. Furthermore, scholars should also
devote more attention to empirically unpacking the mechanism by which verbal
intelligence mitigates anti-intellectualism, and consider alternate theoretical
accounts. For example, some might consider pitting the theoretical account pre-
sented here versus a group-centric account, whereby highly verbally intelligent
individuals see themselves as experts, and therefore hold more positive attitudes
about expertise. More generally, this research points to an important question
scholars ought to consider when studying Americans’ anti-intellectual attitudes in
the future. How can mass positivity toward scientists and experts be improved?
On one hand, though verbal intelligence is highly heritable (Bouchard, 1998), it
is at least somewhat malleable over the life course, especially at young ages
(Campbell, Ramey, Pungello, Sparling, & Miller-Johnson, 2002). Scholars could
plan and track the effectiveness of interventions designed to boost English com-
prehension at young ages, assessing whether or not they boost levels of verbal
intelligence and consequently decrease holding anti-intellectual attitudes.
Scholars might also consider how other factors might mitigate holding
anti-intellectual attitudes. Science communicators, of course, cannot alter the
public’s level of verbal intelligence with policy reports or televised advertise-
ments. However, consistent with recent work on combating partisan asym-
metries in unorthodox belief-holding (Berinsky, 2017), they could seek the aid
of individuals close to anti-intellectual politicians and causes. Should high-
profile supporters of Donald Trump, for example, come out in defense of
experts (e.g., by noting their general trustworthiness), they may be effective in
encouraging individuals who hold anti-intellectual attitudes to reconsider their
prior beliefs. Science communicators might also find success in issuing warn-
ings about the negative consequences of dismissing scientists and the research
they produce (Bolsen & Druckman, 2015). Most generally, future research
should explore how both individual differences and various communication
strategies might mitigate holding anti-intellectual attitudes.
Appendix
Table A1.  Anti-Intellectualism and Skepticism about Policy-Relevant Science.
GSS CSPP

Climate Climate Nuke energy Nuke energy Human Human Human


  (General) (Family) (General) (Family) activities activities activities

Anti-Intellectualism 0.69* 0.66* −0.47* −0.44* −0.81* −0.94* −1.10*


(0.17) (0.19) (0.17) (0.19) (0.22) (0.33) (0.41)
Conservatism 1.02* 1.20* 0.45* 0.46† −1.88* −2.79* −3.05*
(0.23) (0.27) (0.22) (0.26) (0.28) (0.40) (0.49)
White 0.18 0.36* 0.63* 0.72* 0.03 0.47* 0.04
(0.15) (0.17) (0.14) (0.15) (0.14) (0.20) (0.27)
Income 0.24 0.13 0.69* 0.59* −0.29 0.49 −0.43
(0.22) (0.27) (0.23) (0.30) (0.29) (0.40) (0.50)
Age 1.12* 1.01* 1.39* 1.51* −0.68* −0.52 1.00
(0.24) (0.27) (0.23) (0.28) (0.32) (0.46) (0.61)
Female −0.56* −0.67* −0.74* −0.63* 0.08 −0.04 0.27
(0.10) (0.11) (0.09) (0.11) (0.12) (0.17) (0.22)
Labor Force 0.17 0.13 0.02 −0.06 0.16 0.07 0.32
(0.12) (0.14) (0.12) (0.14) (0.13) (0.19) (0.24)
Civic Knowledge — — — — 0.72* 0.70* 1.17*
  (0.21) (0.31) (0.41)
Democrat 0.03 0.12 −0.30 −0.36 0.58* 0.74† −0.06
(0.20) (0.24) (0.19) (0.23) (0.28) (0.40) (0.58)
Republican 0.33 0.42† 0.13 −0.12 0.09 −0.16 −0.44
(0.20) (0.24) (0.20) (0.23) (0.26) (0.40) (0.56)
PID Strength −0.02 −0.05 0.17 0.51* −0.09 −0.17 0.16
(0.20) (0.24) (0.20) (0.24) (0.26) (0.38) (0.50)

489
(continued)
Table A1. (continued)

490
GSS CSPP

Climate Climate Nuke energy Nuke energy Human Human Human


  (General) (Family) (General) (Family) activities activities activities

High School 0.25 0.18 0.39* 0.44* 0.67 0.45 −0.64


(0.16) (0.18) (0.16) (0.19) (0.49) (0.57) (0.72)
Some College 0.19 0.12 0.24 0.51† 0.68 0.45 −0.78
(0.22) (0.27) (0.24) (0.30) (0.48) (0.57) (0.72)
College 0.23 0.42† 0.82* 0.88* 1.09* 0.81 −0.18
(0.19) (0.22) (0.19) (0.22) (0.49) (0.59) (0.75)
Post College 0.32 0.41† 0.96* 1.15* 0.91† 0.37 −0.45
(0.23) (0.25) (0.22) (0.26) (0.50) (0.61) (0.76)
Religious Service 0.25† 0.08 −0.13 −0.19 — — —
Attendance (0.15) (0.17) (0.15) (0.18)  
β0 — — — — 0.51 0.90 1.59†
  (0.54) (0.71) (0.95)
τ1 0.32 0.19 0.23 0.16 — — —
(0.32) (0.36) (0.31) (0.39)  
τ2 1.67* 1.54* 1.50* 1.35* — — —
(0.32) (0.36) (0.31) (0.39)  
τ3 3.74* 3.41* 3.46* 3.15* — — —
(0.33) (0.37) (0.32) (0.39)  
τ4 6.05* 5.74* 5.46* 5.36* — — —
(0.37) (0.40) (0.34) (0.41)  
n 1,947 1,391 2,027 1,431 3,344 1,861 1,161

Note. Unstandardized ordered logistic regression coefficients presented in columns 1 to 4, and OLS coefficients presented in columns 5 to 7. Standard errors in
parentheses. All survey data are weighted (see text). GSS models calculated with year fixed effects. CSPP = Center for the Study of Political Psychology; GSS = General
Social Survey; OLS = Ordinary Least Squares; PID = Partisan Identity.
†p < .10. *p < .05.
Table A2.  Anti-Intellectualism and Support for Expert-Averse Politics.
GSS CSPP CSPP CSPP

Wallace Trump Feeling Trump Feeling Trump Feeling Trump Trump Trump
  vote Thermometer Thermometer Thermometer vote vote vote Brexit

Anti-Intellectualism 0.76† 0.06* 0.07* 0.08† 0.27 1.10* 0.34 1.02*


(0.42) (0.02) (0.03) (0.04) (0.41) (0.53) (0.60) (0.24)
Conservatism — 0.16* 0.14* 0.15* 1.40* 2.85* 1.53† 0.85*
  (0.03) (0.04) (0.06) (0.44) (0.60) (0.81) (0.30)
White — 0.05* 0.00 0.03 0.99* 0.63† 0.72 −0.32*
  (0.01) (0.02) (0.03) (0.23) (0.35) (0.48) (0.16)
Income — −0.01 −0.07 0.06 −0.48 −0.45 0.06 0.24
  (0.03) (0.04) (0.05) (0.47) (0.57) (0.77) (0.30)
Age −2.16* 0.00 0.03 0.07 0.50 0.74 −0.02 0.62†
(0.66) (0.03) (0.05) (0.06) (0.53) (0.72) (1.35) (0.32)
Female −0.62* −0.00 −0.04* 0.00 0.04 −0.17 −0.28 −0.60*
(0.26) (0.01) (0.02) (0.02) (0.23) (0.29) (0.39) (0.13)
Labor Force 0.16 −0.02 −0.03 −0.03 −0.28 −0.21 −0.72 0.18
(0.28) (0.01) (0.02) (0.02) (0.24) (0.34) (0.46) (0.14)
Civic Knowledge — 0.02 0.00 0.01 −0.25 −0.02 0.33 0.29
  (0.02) (0.03) (0.04) (0.33) (0.52) (0.77) (0.22)
Democrat −0.67 −0.15* −0.21* −0.18* −1.75* −1.56* −1.88* −0.66*
(0.48) (0.03) (0.03) (0.06) (0.42) (0.64) (0.83) (0.30)
Republican −1.29* 0.18* 0.11* 0.20* 1.63* 1.71* 1.13 0.06
(0.52) (0.03) (0.04) (0.07) (0.42) (0.65) (0.91) (0.28)
PID Strength −0.49 −0.03 0.03 −0.05 0.04 −0.40 0.06 0.64*
(0.54) (0.02) (0.03) (0.04) (0.45) (0.66) (0.85) (0.27)
Hostile Sexism — 0.02 0.04 0.14* 0.56 0.96† 1.85* 0.94*
  (0.03) (0.03) (0.05) (0.43) (0.55) (0.75) (0.28)

491
(continued)
Table A2. (continued)

492
GSS CSPP CSPP CSPP

Wallace Trump Feeling Trump Feeling Trump Feeling Trump Trump Trump
  vote Thermometer Thermometer Thermometer vote vote vote Brexit

Muslim FT — −0.12* −0.08* −0.12* −1.57* −1.09* −1.64* 0.03


  (0.03) (0.03) (0.05) (0.37) (0.47) (0.69) (0.25)
Racial Resentment — 0.17* 0.19* 0.14* 2.88* 3.84* 2.18* 1.22*
  (0.03) (0.04) (0.05) (0.40) (0.54) (0.77) (0.31)
Authoritarianism — −0.04* −0.04 −0.02 −0.23 −1.58* −0.16 −0.57*
  (0.02) (0.03) (0.04) (0.29) (0.49) (0.68) (0.21)
Religious Service 0.08 −0.00 0.00 0.02 0.28 0.32 0.51 0.41*
Attendance (0.38) (0.02) (0.02) (0.03) (0.27) (0.38) (0.47) (0.17)
High School — −0.14* −0.11† −0.10 −0.94 −1.46 0.04 0.65
  (0.05) (0.06) (0.07) (0.58) (0.96) (0.77) (0.49)
Some College — −0.11* −0.09 −0.09 −0.34 −0.80 0.30 0.81†
  (0.04) (0.06) (0.07) (0.55) (0.96) (0.66) (0.48)
College −1.36* −0.14* −0.09 −0.14† −1.01† −1.59 −0.22 0.72
(0.39) (0.05) (0.06) (0.08) (0.57) (1.02) (0.71) (0.49)
Post College — −0.13* −0.14* −0.13† −1.11† −1.97† −0.37 0.72
  (0.05) (0.06) (0.08) (0.59) (1.04) (0.78) (0.51)
Constant 0.32 0.43* 0.47* 0.33* −1.94* −2.64* −2.65 −3.60*
(0.60) (0.06) (0.08) (0.11) (0.77) (1.19) (1.70) (0.67)
n 699 3,213 1,737 1,126 2,832 1,565 1,039 3,231

Note. Unstandardized logistic regression coefficients presented in column 1 and columns 5 to 8. OLS coefficients presented in columns 1 to 4. Standard errors in
parentheses. All survey data are weighted. Column 1 was limited to White respondents only, as no non-White respondents in the GSS (with nonmissing data on all
other covariates) reported voting for Wallace in 1968. Second, due to insufficient observations of individuals with some college and postcollege degrees in the 1972
wave, the column 1 education variable was collapsed into a binary indicator of whether or not individuals completed college. CSPP = Center for the Study of Political
Psychology; GSS = General Social Survey; OLS = Ordinary Least Squares; PID = Partisan Identity.
†p < .10. *p < .05.
Motta 493

Acknowledgments
I thank Joanne Miller, Wendy Rahn, Christina Farhart, Kristina Jessen Hansen, Julian
Christensen, Paul Goren, Chris Federico, and Howie Lavine for their invaluable feed-
back and suggestions.

Declaration of Conflicting Interests


The author(s) declared no potential conflicts of interest with respect to the research,
authorship, and/or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research,
authorship, and/or publication of this article: The author was funded by the National
Science Foundation [grant no. 000039202].

Supplemental Material
Supplementary material is available for this article online.

Notes
1. For example, see https://www.timeshighereducation.com/blog/brexit-would-be-
victory-those-who-distrust-academics
2. See http://europe.newsweek.com/michael-gove-sky-news-brexit-economics-
imf-466365
3. This measure, on its face, pertains to scientists in general and has been inter-
preted as such in prior research (e.g., Gauchat, 2012). Consequently, it does not
address which specific considerations about scientists, experts, and/or scientific
research citizens bring to mind when responding to this question. It could be the
case, for example, that these responses are shaped by attitudes about particular
types of scientific research or scientists. Another alternative might be that atti-
tudes toward one or more scientific/expert groups “spill over” to shape attitudes
about scientists more broadly. Answering this question is outside the scope of
this project (and the present data), but it is a fascinating mechanism to study in
more detail in the future.
4. This account is primarily focused on the role that verbal intelligence plays in mit-
igating negative stereotypes about experts. Some might also wonder about the
relationship between verbal intelligence and positive stereotypes (Czopp, Kay, &
Cheryan, 2015) about experts, for example, that they are universally trustworthy
and/or likable. As the research summarized above suggests, verbal intelligence
is associated with an increased propensity to eschew heuristics in general, opting
for more careful, open-minded, and ultimately more cognitively taxing forms
of judgment formulation. Thus, to the extent that verbal intelligence influences
anti-intellectual attitudes, I suspect it to be based not on the replacement of one
494 American Politics Research 46(3)

heuristic for another but instead by formulating careful opinions that avoid defer-
ence to cognitive shortcuts.
5. My goal, in studying support for Brexit, is to better understand the correlates
of Americans’ sympathies with the movement. I do not intend to claim the
movement itself was caused by aversion to scientists and experts in the United
Kingdom. While this is an interesting question, and warrants future research, the
present data cannot speak to it.
6. Another plausible outcome to test in the General Social Survey (GSS) would
be support for George Bush over Michael Dukakis in 1988, although there are
a few complicating factors. As noted earlier, Bush made use of anti-intellectual
rhetoric from time to time in campaigning against Dukakis. Rigney argues, how-
ever, that Bush can actually be thought about as an intellectual himself (he was,
after all, both an ivy league grad and director of a large federal bureau). Thus,
my theoretical expectations about the link between Bush support are somewhat
unclear. Nevertheless, I consider the effect of anti-intellectualism on Bush sup-
port in the Supplementary Materials (Table S3). The results show no effect of
anti-intellectualism on Bush support, changing the substantive likelihood of vot-
ing for him by 0%.
7. The effect of anti-intellectualism was robust in all models, except for predicting
support for George Wallace. In that model, trust in the judicial system explained
away the effect of anti-intellectualism. However, as I noted earlier, the courts
were one of Wallace’s favorite targets of his anti-intellectual rhetoric. Thus, it
could be the case that court attitudes mediate the effect of anti-intellectualism
on Wallace support. As I detail in the note accompanying Table S4, formal
mediation tests show precisely this. Overall, while the argument for conflation is
important to bear in mind, this evidence is more consistent with the interpretation
of anti-intellectualism as a distinct construct.
8. Interestingly, the cross-lagged models displayed in Figure S1 do not yield a sig-
nificant effect of anti-intellectualism on Trump support in September, control-
ling for June levels of support. Conceptually, it is important to point out that
this null finding does not undermine the results presented in the main text so
far. While the alternative “reverse-causal” account is necessarily dynamic (i.e.,
supporting Trump leads to change in anti-intellectualism), the same is not neces-
sarily true about the effect of anti-intellectualism on Trump support. As I note in
Supplementary Materials, the effect of anti-intellectualism might best be thought
about as initially shaping support for Trump, which is then highly stable over
time. On an empirical level, this null finding is also somewhat unsurprising—as
June Trump support accounts for more than 72% of the variance in September
support, leaving little room for other factors to explain much variation. Given
high levels of temporal stability for both anti-intellectualism and Trump sup-
port, I respecified the cross-lagged models to control only for demographic and
political essentials (Figure S2). The reverse-causal account remains null under
this specification (β* = 0.03, p > .10), while the dynamic effect of anti-intellec-
tualism on Trump support does attain statistical significance (β* = 0.05, p < .05).
Motta 495

Thus, while anti-intellectualism is probably best thought about as influencing


candidate support at the outset of the campaign, it may exert a dynamic influence
under some specifications and is unlikely to be undermined by reverse-causal
accounts.
9. The probabilities in Figure 2 are calculated by moving to the minimum to maxi-
mum observed value of each listed variable. As Figure S2 in the Supplementary
Materials demonstrates, however, the number of respondents who earn the low-
est possible score on the Wordsum test is smaller than the number of individuals
earning the lowest possible degree (as the GSS collapses all those who have less
than a high school degree into a single category). I therefore reestimate these
predictions moving across smaller ranges of each distribution, employing single
standard deviation shifts for the interval Wordsum test (from a score of 40% to
a score of 80%), and comparing values across the interquartile range for the cat-
egorical degree attainment variable (from having a high school to degree to hav-
ing less than a college degree). The results show that verbal intelligence remains
a stronger mitigator, boosting trust toward scientists by slightly more than 9%,
whereas educational attainment only does the same by about 2%.

References
Allum, N., Sturgis, P., Tabourazi, D., & Brunton-Smith, I. (2008). Science knowledge
and attitudes across cultures: A meta-analysis. Public Understanding of Science,
17, 35-54.
Berinsky, A. J. (2017). Rumors and health care reform: Experiments in political mis-
information. British Journal of Political Science, 47, 241-262.
Blank, J. M., & Shaw, D. (2015). Does partisanship shape attitudes toward science and
public policy? The case for ideology and religion. The ANNALS of the American
Academy of Political and Social Science, 658, 18-35.
Bolsen, T., & Cook, F. L. (2008). The polls—Trends: Public opinion on energy pol-
icy: 1974-2006. Public Opinion Quarterly, 72, 364-388.
Bolsen, T., & Druckman, J. N. (2015). Counteracting the politicization of science.
Journal of Communication, 65, 745-769.
Bolsen, T., Druckman, J. N., & Cook, F. L. (2015). Citizens, scientists, and policy
advisors beliefs about global warming. The ANNALS of the American Academy
of Political and Social Science, 658, 271-295.
Bonikowski, B., & Gidron, N. (2016). The populist style in American politics:
Presidential campaign discourse, 1952-1996. Social Forces, 94, 1593-1621.
Bouchard, T. J., Jr. (1998). Genetic and environmental influences on adult intelli-
gence and special mental abilities. Human Biology, 70, 257-279.
Campbell, F. A., Ramey, C. T., Pungello, E., Sparling, J., & Miller-Johnson, S.
(2002). Early childhood education: Young adult outcomes from the abecedarian
project. Applied Developmental Science, 6, 42-57.
Condon, M. (2015). Voice lessons: Rethinking the relationship between education
and political participation. Political Behavior, 37, 819-843.
496 American Politics Research 46(3)

Cook, J., Nuccitelli, D., Green, S. A., Richardson, M., Winkler, B., Painting, R., . . .
Skuce, A. (2013). Quantifying the consensus on anthropogenic global warming
in the scientific literature. Environmental Research Letters, 8(2), 1-7.
Czopp, A. M., Kay, A. C., & Cheryan, S. (2015). Positive stereotypes are pervasive
and powerful. Perspectives on Psychological Science, 10, 451-463.
Dawson, M. (2013). Verbal Intelligence. In F. R. Volkmar (Ed.), Encyclopedia of
autism spectrum disorders (pp. 3243-3250). New York, NY: Springer.
Deary, I. J., Batty, G. D., & Gale, C. R. (2008). Bright children become enlightened
adults. Psychological Science, 19, 1-6.
Delia, J. G., Kline, S. L., & Burleson, B. R. (1979). The development of persua-
sive communication strategies in kindergarteners through twelfth-graders.
Communication Monographs, 46, 241-256.
DeYoung, C. G., Grazioplene, R. G., & Peterson, J. B. (2012). From madness to
genius: The openness/intellect trait domain as a paradoxical simplex. Journal of
Research in Personality, 46(1), 63-78.
Dhont, K., & Hodson, G. (2014). Does lower cognitive ability predict greater preju-
dice? Current Directions in Psychological Science, 23, 454-459.
Dorius, S. F., Alwin, D. F., & Pacheco, J. (2016). Twentieth century intercohort trends
in verbal ability in the United States. Sociological Science, 3, 383-412.
Editorial Board. (2016, September). Donald Trump’s lack of respect for science is
alarming. Scientific American. Retrieved from https://www.scientificamerican.
com/article/donald-trump-s-lack-of-respect-for-science-is-alarming/
Environmental Protection Agency. (2016). Radiation risks and realities. Washington,
DC: National Service Center for Environmental Publications, Environmental
Protection Agency.
Fowler, E. F., & Gollust, S. E. (2015). The content and effect of politicized health
controversies. The ANNALS of the American Academy of Political and Social
Science, 658, 155-171.
Gauchat, G. W. (2008). A test of three theories of anti-science attitudes. Sociological
Focus, 41, 337-357.
Gauchat, G. W. (2012). Politicization of science in the public sphere: A study of
public trust in the United States, 1974 to 2010. American Sociological Review,
77, 167-187.
Glick, P., & Fiske, S. T. (1996). The Ambivalent Sexism Inventory: Differentiating
hostile and benevolent sexism. Journal of Personality and Social Psychology,
70, 491-512.
Golshan, T. (2016, September). Breitbart thought the polls were biased against
Trump. So it did its own poll. Clinton won. Vox. Retrieved from https://www.
vox.com/2016/8/15/12480950/breitbart-poll-bias-clinton-winning-trump
Gooch, A. (2015). Measurements of cognitive skill by survey mode: Marginal differ-
ences and scaling similarities. Research & Politics, 2(3), 1-11.
Hofstadter, R. (1963). Anti-intellectualism in American life. (3. print.) (Vol. 713).
New York, NY: Vintage.
Jost, J. T., Glaser, J., Kruglanski, A. W., & Sulloway, F. J. (2003). Political conserva-
tism as motivated social cognition. Psychological Bulletin, 129, 339-375.
Motta 497

Kahan, D. M. (2012). Ideology, motivated reasoning, and cognitive reflection: An


experimental study. Judgment and Decision Making, 8, 407-424.
Kahan, D. M., Landrum, A., Carpenter, K., Helft, L., & Jamieson, K. H. (2017).
Science curiosity and political information processing. Advances in Political
Psychology, 38, 179-199.
Kahan, D. M., Peters, E., Dawson, E. C., & Slovic, P. (2013, September). Motivated
numeracy and enlightened self-government (Public Law Working Paper No.
307). New Haven, CT: Yale Law School.
Kahan, D. M., Peters, E., Wittlin, M., Slovic, P., Ouellette, L. L., Braman, D., &
Mandel, G. (2012). The polarizing impact of science literacy and numeracy on
perceived climate change risks. Nature Climate Change, 2, 732-735.
Kinder, D. R., & Sanders, L. M. (1996). Divided by color: Racial politics and demo-
cratic ideals. Chicago, IL: University of Chicago Press.
Kraft, P. W., Lodge, M., & Taber, C. S. (2015). Why people “don’t trust the evi-
dence”: Motivated reasoning and scientific beliefs. The ANNALS of the American
Academy of Political and Social Science, 658, 121-133.
Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulletin, 108,
480-498.
Lewandowsky, S., & Oberauer, K. (2016). Motivated rejection of science. Current
Directions in Psychological Science, 25, 217-222.
Lodge, M., & Taber, C. S. (2013). The rationalizing voter. Cambridge, UK: Cambridge
University Press.
MacWilliams, M. C. (2016). Who decides when the party doesn’t? Authoritarian vot-
ers and the rise of Donald Trump. PS: Political Science & Politics, 49, 716-721.
Malhotra, N., Krosnick, J. A., & Haertel, E. (2007). The psychometric properties
of the GSS Wordsum Vocabulary Test (GSS Methodological Report No. 11).
Chicago, IL: National Opinion Research Center.
Milman, O. (2016, November). NASA’s climate research will likely be scrapped by
Trump. Newsweek. Retrieved from http://www.newsweek.com/nasas-climate-
research-will-likely-be-scrapped-trump-525559
Miner, J. B. (1957). Intelligence in the United States. New York, NY: Springer.
Mooney, C. (2012). The republican brain: The science of why they deny science—and
reality. New York, NY: John Wiley.
Nie, N. H., Junn, J., & Stehlik-Barry, K. (1996). Education and democratic citizen-
ship in America. Chicago, IL: University of Chicago Press.
Nisbet, E. C., Cooper, K. E., & Garrett, R. K. (2015). The partisan brain how disso-
nant science messages lead conservatives and liberals to (dis)trust science. The
ANNALS of the American Academy of Political and Social Science, 658, 36-66.
North, A. (2016, September). Donald Trump’s Brexit moment. The New York Times.
Retrieved from https://takingnote.blogs.nytimes.com/2016/09/01/donald-trumps-
brexit-moment/?_r=0
Oliver, J. E., & Rahn, W. M. (2016). Rise of the Trumpenvolk: Populism in the 2016
election. The ANNALS of the American Academy of Political and Social Science,
667, 189-206.
498 American Politics Research 46(3)

Price, E., Ottati, V., Wilson, C., & Kim, S. (2015). Open-minded cognition.
Personality and Social Psychology Bulletin, 41, 1488-1504.
Riffkin, R. (2016). For first time, majority in U.S. oppose nuclear energy. Gallup
Politics. Washington, DC: Gallup.
Rigney, D. (1991). Three kinds of anti-intellectualism: Rethinking Hofstadter.
Sociological Inquiry, 61, 434-451.
Saad, L., & Jones, J. M. (2016). U.S. concern about global warming at eight-year
high. Gallup Politics. Washington, DC: Gallup
Schaffner, B. F., MacWilliams, M., & Nteta, T. (2016). Explaining white polariza-
tion in the 2016 vote for president: The sobering role of racism and sexism.
Unpublished manuscript. Paper prepared for presentation at the Conference on
the U.S. Elections of 2016: Domestic and International Aspects. January, 2017.
IDC Herzliya Campus.
Stanovich, K. E., & West, R. F. (1997). Reasoning independently of prior belief and
individual differences in actively open-minded thinking. Journal of Educational
Psychology, 89, 342-357.
Sturgis, P., & Allum, N. (2004). Science in society: Re-evaluating the deficit model of
public attitudes. Public Understanding of Science, 13, 55-74.
Swain, F., & Wong, S. (2016, August). What Donald Trump has said about science—
and why he’s wrong. New Scientist. Retrieved from https://www.newscientist.
com/article/2099977-what-donald-trump-has-said-about-science-and-why-hes-
wrong/
Thorndike, R. L. (1942). Two screening tests of verbal intelligence. Journal of
Applied Psychology, 26, 128-135.

Author Biography
Matthew Motta is a PhD candidate at the University of Minnesota in American
Politics and Quantitative Methodology. His research broadly focuses on public opin-
ion, political psychology, and survey methodology. He is also a research associate for
the Wesleyan Media Project.

You might also like