ArticlePDF Available

Abstract and Figures

Prior research suggests that liberals are more complex than conservatives. However, it may be that liberals are not more complex in general, but rather only more complex on certain topic domains (while conservatives are more complex in other domains). Four studies (comprised of over 2,500 participants) evaluated this idea. Study 1 involves the domain specificity of a self-report questionnaire related to complexity (dogmatism). By making only small adjustments to a popularly used dogmatism scale, results show that liberals can be significantly more dogmatic if a liberal domain is made salient. Studies 2–4 involve the domain specificity of integrative complexity. A large number of open-ended responses from college students (Studies 2 and 3) and candidates in the 2004 Presidential election (Study 4) across an array of topic domains reveals little or no main effect of political ideology on integrative complexity, but rather topic domain by ideology interactions. Liberals are higher in complexity on some topics, but conservatives are higher on others. Overall, this large dataset calls into question the typical interpretation that conservatives are less complex than liberals in a domain-general way.
Content may be subject to copyright.
Are Conservatives Really More Simple-Minded than Liberals?
The Domain Specificity of Complex Thinking
Lucian Gideon Conway III
The University of Montana
Laura Janelle Gornick
Roanoke College
Shannon C. Houck
The University of Montana
Christopher Anderson
University of Oklahoma
Jennifer Stockert
Minnesota State University, Mankato
Diana Sessoms
The University of Montana
Kevin McCue
The University of Montana
Prior research suggests that liberals are more complex than conservatives. However, it may be that liberals
are not more complex in general, but rather only more complex on certain topic domains (while conservatives
are more complex in other domains). Four studies (comprised of over 2,500 participants) evaluated this idea.
Study 1 involves the domain specificity of a self-report questionnaire related to complexity (dogmatism). By
making only small adjustments to a popularly used dogmatism scale, results show that liberals can be
significantly more dogmatic if a liberal domain is made salient. Studies 2–4 involve the domain specificity of
integrative complexity. A large number of open-ended responses from college students (Studies 2 and 3) and
candidates in the 2004 Presidential election (Study 4) across an array of topic domains reveals little or no
main effect of political ideology on integrative complexity, but rather topic domain by ideology interactions.
Liberals are higher in complexity on some topics, but conservatives are higher on others. Overall, this large
dataset calls into question the typical interpretation that conservatives are less complex than liberals in a
domain-general way.
KEY WORDS: complexity, ideology, attitudes
0162-895X V
C2015 International Society of Political Psychology
Published by Wiley Periodicals, Inc., 350 Main Street, Malden, MA 02148, USA, 9600 Garsington Road, Oxford, OX4 2DQ,
and PO Box 378 Carlton South, 3053 Victoria, Australia
Political Psychology, Vol. xx, No. xx, 2015
doi: 10.1111/pops.12304
It has practically achieved the state of an axiom in our field that liberals are more complex
thinkers than conservatives. This is not without reason. Meta-analyses—covering a vast array of evi-
dence related to dogmatism, uncertainty avoidance, openness to experience, need for closure, and inte-
grative complexity—suggest that liberals are indeed more complex than conservatives (see Jost,
Glaser, Kruglanski, & Sulloway, 2003; Van Hiel, Onraet, & De Pauw, 2010; see also Joseph, Graham,
& Haidt, 2009).
Nonetheless, we believe that the judgment that conservatives are broadly simple-minded may be
premature. In the present article, we provide an alternative framework for understanding existing dif-
ferences in complexity between conservatives and liberals and some initial evidence for that frame-
work. Our approach focuses on considering more fully the topic domain that the complexity
measurement is relevant to.
The Domain Specificity of Complex Thinking
Complex thinking is domain-specific.
Someone can think highly complexly about the Iraq War,
and yet still think very simply about broccoli. A lot of evidence using many different operations of
complex thinking underscores this point (Conway, Schaller, Tweed, & Hallett, 2001; Houck, Conway,
& Gornick, 2014; Judd & Lusk, 1984; Liht, Conway, Savage, White, O’Neill, 2011; Pancer et al.,
1995; Sidanius, 1984; Suedfeld, 2000; Tetlock, Peterson, & Lerner, 1996). For example, the complex-
ity of thinking can be affected by the importance of the content domain (Conway et al., 2008;
Suedfeld, 2000) by the experience people have with the domain (Conway et al., 2008; Dasen, 1975;
Suedfeld, 2000), by the heritability of the domain (Conway, Dodds, Hands Towgood, McClure, &
Olson, 2011), or by the value conflict implied by the domain (e.g., Suedfeld, Bluck, Loewen, &
Elkins, 1994; Tetlock, 1986).
This fact of domain specificity has implications for our understanding of group differences in
complexity. For example, while there is a tendency to think about cultural differences in complexity
in monolithic terms, evidence (see Conway et al., 2001, for a summary) suggests that the culture-
complexity link is in fact domain-specific. In methodological terms, the relationship between cultural
groups and complexity is probably better described as a culture by topic-domain interaction than it is
by a main effect of culture (Conway et al., 2001).
Ideology X Domain Interactions on Complex Thinking
A primary assumption of the present article is that this same culture 3domain interaction applies
to political culture (defined operationally by political ideology). This idea is not new. Over 25 years
ago, Tetlock (1986) pointed out that, because conservatives and liberals differ in which values are in
conflict, his Value Pluralism Model predicts ideology 3domain interactions. Of course, his model
also predicts that integrative complexity most typically breaks left of center; thus, it expects both a
main effect (liberals more complex) and interactions with issue domain. Yet it is worth noting that,
In the present article, we use the term “domain specific” to indicate the particular content that comprises the thought-
about subject in the same way that past researchers discuss “domains” or “issues” (e.g., Suedfeld, 2000; Tetlock,
1986). Deciding whether or not two thought-about subjects involve different “domains” in this sense is not itself
always a simple matter. We suspect that most scientists would agree that “broccoli” is a different domain than “the
Iraq War.” But consider, for example, that while it is possible that “smoking cigarettes” and “drinking alcohol” could
be viewed as different domains, it is also possible they could be viewed as being under the larger single domain of
“substance use.” In the present work, we took the view that it was best not to make too many inferences about the
larger categories participants might or might not impose on each topic, and thus we treat domains as different to the
degree that they clearly indicate potentially different linguistic topics. Thus, unless topic stems use identical language
or are direct synonyms, we treat them as separate domains.
2 Conway et al.
while the main effect prediction has been widely discussed, the expected interaction has not.
This is
not from an initial lack of encouragement of its importance. In 1986, Tetlock (p. 825) encouraged
researchers to study this topic: “Systematic study of such Ideology X Issue interactions should be a
major goal of future laboratory and archival studies on this topic.” Tetlock’s admonition has been
largely ignored. Indeed, in the intervening 25-plus years, very little research has been done to further
this “major goal.” The research that has been done relevant to issue domain and complexity (e.g.,
Lavallee & Suedfeld, 1997; Suedfeld, 2000; Suedfeld et al., 1994; Suedfeld, Steel, & Schmidt, 1994)
has generally not directly tested ideology X issue interactions, and, to date, no research program has
systematically explored these interactions. Further, almost none of the research cited in meta-analyses
on conservative simplicity (Jost et al., 2003; Van Hiel et al., 2010) has directly accounted for such
Why This Matters: The Present Research
If we assume that (1) ideology 3domain interactions are prevalent in reality, and (2) very few
formal and systematic tests have been made of such interactions, then it is possible that existing
research to date might misrepresent the actual main effect between liberals and conservatives on com-
plexity—it may be that the small number of tested domains are on average ones for which liberals
score higher on complexity and that the counterbalancing domains have yet to be tested. As such,
before fully deciding on the question of whether or not liberals and conservatives differ in complexity
in a domain-general way, it is worth first more fully exploring ideology 3topic domain interactions
on a wide range of topics (see also Duarte et al., in press).
To that end, the present research attempts to show how two of the major complexity-relevant
areas most typically used as evidence of conservative simplicity (Jost et al., 2003; Van Hiel et al.,
2010) might be accounted for by ideology 3domain interactions. Study 1 focuses on one self-report
measure relevant to complexity; studies 2–4 focus on complexity scoring of open-ended statements.
Study 1: Dogmatism is Domain-Specific
As Jost et al. (2003) point out, dogmatism has historically been conceptualized as closed-
mindedness that is indicative of rigid, black-and-white thinking; for example, Rokeach described dog-
matism in terms of “closed belief systems” (1960, p. 67). Such closed belief systems are a hallmark of
cognitive simplicity (see, e.g., Conway et al., 2008; Tetlock, 1986), and thus it is not surprising that
other researchers have noted the kinship between dogmatism and cognitive simplicity (e.g., Suedfeld,
Tetlock, & Streufert, 1992). For our purposes, this clear conceptual overlap suggests that it is reasona-
ble for us to view dogmatism as a proxy for closed-minded, simplistic thinking.
Indeed, this overlap is the impetus for dogmatism being used as one of the major arguments in
the case for conservative simplicity (Jost et al., 2003). Specifically, a large body of research, mostly
using versions of the Rokeach (1960) scale we use here, reveals a positive relationship between dog-
matism and political conservatism (see Jost et al., 2003).
The use of dogmatism as an argument in the case for conservative simplicity is partially depend-
ent upon its conceptualization as being free of specific topic content (henceforth, domain general;see
Because so many factors relevant to topic content affect complexity, there are many reasons—including value plural-
ism and attitude strength—to expect interactions between ideology and content domain on complexity. As such, it is
reasonable to expect interactions at a larger level without necessarily being able to identify specifically why such an
interaction exists in a particular context. Although we provide some preliminary evidence concerning one possible
explanatory mechanism in this context, the primary purpose of this article is to demonstrate the expected interaction
at a larger level and not to provide a coherent theoretical explanation for the interaction.
Ideology and Complexity 3
Jost et al., 2003, for a discussion). And many of its items do, on the surface, appear to be domain gen-
eral. On the other hand, some research suggests that students who rate the ideological content of the
Rokeach items rate them as leaning towards conservative content (see Simons, 1968), and other aca-
demics have noted (e.g., Van Hiel et al., 2010) that not all the items on the scale are free of specific
content. One of the items explicitly mentions religion, another item states an explicit idea about
humans being helpless and frail, and a third item states that focusing on one’s happiness is contemptu-
ous. These items explicitly identify a domain of interest and in some cases state an opinion on that
domain. Thus, it is possible that the standard dogmatism scale is not really a pure measurement of
domain-general dogmatism, but rather a measurement of dogmatism that captures domains on which
conservatives are more dogmatic. In the present study, we directly test the degree that the specific con-
tent of the dogmatism questionnaire matters by altering the items to reflect one of two content
domains (environmental issues versus religion).
Study 1 Method
Four-hundred and seventy-five undergraduates at the University of Montana participated for
course credit in large-group sessions.
Questionnaire Packets
Participants completed a questionnaire packet which contained a dogmatism scale,
a political
ideology scale, and some demographic information.
Dogmatism. Eighteen items from Rokeach’s (1960) standard dogmatism scale were used in the
present study.
Participants were randomly assigned one of three versions of the dogmatism scale. Some partici-
pants received the standard version of the scale as typically used in previous research on ideology
(obtained from Ray, 1970). Other participants received one of two domain-specific versions of the
scale. In one condition, participants received a scale that was designed to measure their dogmatism
about religion, and in another condition, participants received a scale designed to measure their dog-
matism about environmental issues.
These domain-specific dogmatism scales were nearly identical to the standard scale and to each
other, but they differed only in intentionally injecting content domains into the items (please see the
online supplemental information for the entire scales). An example will help illustrate. A standard
item on the dogmatism scale is “A group which tolerates too much difference of opinion among its
own members cannot exist for long.” For the religious dogmatism scale, this item was adapted (italics
and bold added for emphasis here) to say “A religious group which tolerates too much difference of
opinion among its own members cannot exist for long.” The parallel environmental dogmatism ques-
tionnaire item read “An environmental group which tolerates too much difference of opinion among
its own members cannot exist for long.”
In this way, the two alternate domain-specific dogmatism questionnaires kept almost all of the
language from the original items but interjected a content domain (either religion or environmental
issues) into the majority of those items. Inter-item reliability for the scale was satisfactory in all three
conditions (standard-scale alpha 5.74; environmental-scale alpha 5.74; religious-scale alpha 5.88).
Participants also completed measurements of authoritarianism and modern racism. These measurements are not
directly relevant to the present study and are not discussed further.
4 Conway et al.
Four items on each domain-specific scale were kept in their original (domain-general) wording.
These originally worded items appeared last in the list of 18 items.
Political Ideology. Participants also completed several items relevant to their political ideology.
We focus on two of those here: standard bipolar items anchored by liberal/conservative and Demo-
cratic/Republican that have been used in prior research (e.g., Conway et al., 2012) and are similar to
the vast majority of standard ideology measures (see e.g., Jost et al., 2003; Federico, Deason, &
Fisher, 2012). These two items were highly correlated (r5.77) and thus averaged into a single mea-
sure of political conservatism (standardized alpha 5.86).
Study 1 Results
Correlations within Condition
We first looked at correlations within each condition. Replicating prior research, the standard
dogmatism scale was positively correlated with political conservatism (r[111] 5.27, p<.01).
Does this finding represent a domain-general or domain-specific phenomenon? Our next two
findings suggest it is domain-specific. First, the religion dogmatism scale—a scale specifically
designed on a domain on which conservatives are more likely to be dogmatic—showed virtually the
same effect as the supposedly domain-general scale (r[184] 5.33, p<.001). More importantly, are
conservatives unilaterally more dogmatic across all domains? The answer is no: They were signifi-
cantly less dogmatic on environmental domains, as illustrated by the negative correlation between
conservatism and environmental dogmatism (r[180] 5–.26, p<.001).
Absolute Values for Conservatives and Liberals
Evaluating only correlations between scales, it is conceptually possible that the negative cor-
relation between conservatism and the liberal-focused scale could be driven more by a rejection
of those items by conservatives than by an acceptance by liberals. To look at the plausibility of
this alternative, we divided participants up categorically into conservatives (those who scored
above 5 on the conservatism scale) and liberals (those who scored below 5 on the scale). This
analysis thus drops those directly at the midpoint of the scale. We then ran parallel 2 (Type of
Scale) 32 (Ideology) ANOVAs for dogmatism (for this analysis, we dropped the standard ques-
tionnaire). This analysis showed a significant Type of Scale 3Ideology interaction (F>29.9,
More pertinent to our purpose are the absolute values for each scale, broken down by ideol-
ogy. These are presented in Table 1. As can be seen there, the highest score for simplicity was
for liberals (the highest cell was liberals on environmental dogmatism). These additional
We also compared correlations within each condition by the type of item (domain-general versus domain-specific).
Removing the four domain-general items from these domain-specific scales increased the size of the difference
between the religion and environmental dogmatism scales: The main conservatism measure was significantly nega-
tively correlated for the environmental scale (r5–.36, p<.001) and significantly positively correlated for the religion
scale (r5.34, p<.001; Fisher’s Z-test for comparing correlations 56.86, p<.001). When looking at the four
domain-general items on the otherwise domain-specific scales, conservatives were nonsignificantly positively corre-
lated for the environmental scale (r5.10) and significantly positively correlated for the religion scale (r5.23,
p<.01; Fisher’s Z-test for comparing correlations 51.32, p>.05). These results indicate that there is some “leakage”
from the domain-specific environmental items that reduce the typical size of the conservatism-political ideology corre-
lation but still reveal that the nature of the item itself does matter, even within a domain-specific context.
Ideology and Complexity 5
analyses with absolute values make any interpretation of Study 1 results based on a lack of lib-
eral simplicity implausible.
Study 1 Discussion
These results suggest that the relationship between ideology and dogmatism is domain-specific.
Conservatives are indeed more dogmatic on the religious domain; but liberals are more dogmatic on
the environmental domain.
It might be easy to dismiss these effects as reflecting the content preferences of liberals and con-
servatives (and thus as not reflecting anything about dogmatism per se). There are two reasons why
we think such a dismissal would be premature.
First, the dismissal is a double-edged sword. If the question is “do prior results suggest that con-
servatives are more dogmatic?” then simply dismissing our results as only having to do with content
raises the possibility that all dogmatism scales are picking up on content primarily (and not dogma-
tism per se).
Second, more importantly, a quick dismissal of these findings does not capture the subjective
nature of the items themselves. For example, consider that for dogmatism, liberals scored higher on
the following questions:
There are two kinds of people in this world: those who are for the truth that the planet
is warming and those who are against that obvious truth.
When it comes to stopping global warming, it is better to be a dead hero than a live
A person who thinks primarily of his/her own happiness, and in so doing disregards the
health of the environment (for example, trees and other animals), is beneath contempt.
Table 1. Studies 1 through 4: Liberals and Conservatives by Topic Domain and Complexity Measurement
Liberal Complex Domains Conservative Complex Domains
Liberals Conservatives Liberals Conservatives
Study 1 (Student) Dogmatism 2.55 3.21 3.61 3.07
Study 2 (Student) Complexity 1.90 1.75 1.54 1.76
Study 3 (Student) Complexity 2.02 1.67 1.54 1.79
Study 4 (Bush/Kerry) Complexity 1.81 1.34 1.39 1.93
Note: For Studies 1–3, Conservatives 5participants above midpoint on political conservatism scale, and Liber-
als 5participants below midpoint. For Study 4, conservative 5Bush and Liberal 5Kerry. Studies 2–4 5dialectical com-
plexity for ad hoc categories (see text).
Some prior research suggests that the relationship between ideology and outcome variables may be curvilinear and as
such represents more about ideological extremism than about ideological content (e.g., Tetlock et al., 1994). As a
result, we tested for the possibility that our results represent a curvilinear, rather than a linear, relationship. In particu-
lar, we ran linear regression on all key results while entering both a linear and two separate nonlinear terms for politi-
cal conservatism as simultaneous predictors: (1) A mean-centered quadratic term for conservatism and (2) an
extremism score for conservatism (computed as the absolute difference from the midpoint of the conservatism scale).
All analyses were performed within-condition in a way parallel to that described above. Results overwhelmingly sup-
port a linear, rather than a nonlinear, interpretation of our results. For dogmatism, when linear and quadratic/extre-
mism scores are entered simultaneously, all linear conservatism terms remained significant (p’s <.01), while no
significant nonlinear effects emerged in any condition on either nonlinear measurement (p’s>.380). Thus, (1) all lin-
ear effects remained significant—and were of similar size and direction as in zero-order analyses—when accounting
for nonlinear effects, and (2) nonlinear effects overall accounted for very little of the variance. Thus, our results are
much better construed as linear effects than as nonlinear.
6 Conway et al.
The subjective tone of those statements is not merely “I am an environmentalist” but rather “all
people who disagree with me are fools.” In these and other items from the scale, liberals are consent-
ing to (1) categorizing the world into only two kinds of people, those that are right and those that are
wrong, (2) a scorn of those unwilling to die for a cause, (3) a belief that persons who disagree with
them are “beneath contempt,” (4) a belief that the only method for understanding the truth is to rely
on experts, (5) an expression that true living involves believing in their cause, and (6) an appeal to the
temporal urgency of the cause. Those are not just statements about having an environmental position:
They are explicitly and overwhelmingly dogmatic statements. And liberals are more likely to agree
with such sentiments—for an environmental domain.
Studies 2–4: Integrative Complexity is Domain-Specific
The dogmatism measurement used in Study 1 has been widely used to make a case for conserva-
tive simplicity, but it is not without its problems (see Van Hiel et al., 2010). It is dependent on partici-
pants’ own self-perceptions and willingness to express them; it also contains explicit ideological
content. Although, as we have discussed, that “ideological content” argument cuts both ways and
does not undermine our present purpose, it would nonetheless be advisable to also use a more open-
ended measure that ameliorated some of these problems.
With that in mind, we turn next to one such open-ended measurement of the complexity of think-
ing: Integrative complexity (e.g., Suedfeld & Streufert, 1966; Suedfeld & Tetlock, 1976; see also Har-
vey, Hunt, & Schroder, 1961; Schroder, Driver, & Streufert, 1965).
Integrative Complexity
Integrative complexity, which formed an important part of Jost et al.’s (2003) case for conserva-
tive simplicity, is used to assess the complexity of spoken or written communications according to
their basic structure (see, e.g., Suedfeld & Bluck, 1988; Suedfeld, Bluck, & Ballard, 1994; Suedfeld
& Leighton, 2002; Suedfeld & Rank, 1976). Passages are coded and assigned a score between 1 and 7
based on the level of differentiation (i.e., the extent to which differing dimensions are used to describe
a given topic) and, if more than one dimension is present, integration (i.e., the joining of these multi-
ple dimensions; see Baker-Brown et al., 1992, for integrative complexity scoring details).
In assigning complexity scores, the particular position argued for by the speaker/writer is irrele-
vant; the score is based on the structure of the passage. As such, the construct is able to capture the
underlying mechanisms of the complexity of thought on a broad level that is conceptually independent
of the content domain of the passage. It is in part for this domain-general breadth that integrative com-
plexity is the most widely used scoring system for measuring the complexity of open-ended state-
ments (e.g., Conway, Conway, Gornick, & Houck, 2014; Conway & Gornick, 2011; Conway et al.,
2012; Conway, Suedfeld, & Clements, 2003; Conway, Suedfeld, & Tetlock, 2001; Houck et al.,
2014; Suedfeld & Bluck, 1988; Suedfeld et al., 1994; Suedfeld & Leighton, 2002; Suedfeld & Tet-
lock, 1976; Tetlock, 1984, 1986).
Domain Specificity
Although it is intended to be a “content free” measurement, this does not mean that participants’
complexity is uninfluenced by the domain of interest. Indeed, although content domain is not often a
subject of inquiry in integrative complexity research, topic domain has been shown to influence inte-
grative complexity in some lines of research (Conway et al., 2008; Conway et al., 2011; Conway
et al., 2012; Pancer et al., 1995; Suedfeld, 2000; Suedfeld, Bluck et al., 1994; Suedfeld & Wallbaum,
1992; Tetlock, 1986; Tetlock, Peterson, & Lerner, 1996). In all this work, the specific content domain
Ideology and Complexity 7
that people wrote or talked about mattered for the ultimate complexity they produced. Thus, it is
worth considering more fully the possibility of ideology 3topic domain interactions on integrative
Our Focus on Dialectical Forms of Integrative Complexity
Recently, a new scoring system for parsing integrative complexity scores into different types of
complexity has been scientifically validated (Conway et al., 2008; Conway et al., 2011). In particular,
some integrative complexity scores are driven by dialectical complexity, which is complexity
achieved by giving legitimacy to two opposing viewpoints. On the other hand, some integrative com-
plexity scores are driven by elaborative complexity, which is complexity achieved by defending or
expanding upon one particular viewpoint (see Conway et al., 2008; Conway et al., 2011).
All materials in studies 2–4 in the present article were scored for integrative complexity and the
two subtypes outlined in Conway et al. (2008). However, we here opt to present results in those stud-
ies on dialectical forms of complexity. Our reasons for doing so are three-fold. First, almost all of the
prior work on integrative complexity cited in meta-analyses by Jost et al. (2003) and Van Hiel et al.
(2010) comes from Tetlock. When scoring integrative complexity, Tetlock has only coded dialectical
forms of complexity in his work (see Conway et al., 2008, for a review). As a result, the best direct
comparison with prior work showing conservative simplicity is dialectical complexity. Second, this
focus makes conceptual sense, because dialectical forms of complexity most clearly map on to the
“rigidity of the right” idea. It is in their inability or unwillingness to think about things from different
points of view that conservatives are supposed to be lacking: And dialectical complexity best captures
that aspect (see Conway et al., 2008). Third, our results are inferentially stronger and more consistent
using dialectical forms of complexity, and thus we acknowledge that part of our decision to present
this set of results is ad hoc.
However, we also performed all analyses using the larger integrative complexity construct, and
the overall pattern of results in most cases is similar (see footnote 12 for details). It is indeed notewor-
thy that the key moderating effects reported here are strongest for the form of complexity (dialectical
complexity) on which conservatives are supposed to be weakest.
Using dialectical complexity below in Studies 2–4, we show that, contrary to the conservative
simplicity hypothesis, no main effect emerges of ideology on complexity. Instead, these results are
better characterized by an ideology 3topic domain interaction.
Study 2
Participants. Over a three-year span, 1,529 undergraduate participants at the University of Mon-
tana completed questionnaire packets, usually in large sessions exceeding 100 persons.
Complexity Question Stems. Participants completed one of 13 possible question stems that
mostly dealt with political and social issues (example topic stems include “death penalty,” “abortion,”
and “organized religion”). These questions were later coded by trained scorers for integrative com-
plexity. Most of these items were chosen because they had been previously assessed for their heritabil-
ity in one of two prior, well-known heritability research programs (Eaves, Eysenck, & Martin, 1989;
Portions of the data from Studies 2 and 3 were used also to test hypotheses about psychological extremism (Conway
et al., 2008) and heritability of attitudes (Conway et al., 2011). However, those studies did not deal with political ide-
ology at all, and thus all analyses presented here are entirely novel.
8 Conway et al.
Martin, Eaves, Heath, Jardine, Feingold, & Eysenck, 1986; see Conway et al., 2008; Conway et al.,
2011 for descriptions of topic selection).
Complexity Coding. Study 2 was scored by coders who had taken an intensive training course
and achieved a .85 reliability score with an expert integrative complexity coder and who had subse-
quently received training in scoring the subconstructs. Responses were coded by 4–5 coders in
“blocks” of around 500 responses each. For each block, every coder of that block scored all partici-
pants (and thus all topics) for that block. Thus, summary scores provided below are the average of 4–
5 coders for each participant. To check for reliability, we computed standardized alphas for each
block separately (because all coders scored all responses in each block, an alpha is an appropriate
metric of reliability). Reliability on each of the different blocks was satisfactory, with standardized
alphas for dialectical complexity ranging from .86 to .89 (M5.88).
Ideology Measurement. All participants completed the same two continuous measurements of
political ideology used in Study 1. These were averaged to produce a single continuous “political con-
servatism” score. For both Studies 2 and 3, we converted this score into a categorical measurement in
a manner identical to our method in Study 1 for testing absolute values. We did this because the cur-
rent conceptual case being made is that conservatives are simple-minded. Because the slope of the
line in a correlation might not fully capture differences between persons on each side of the liberal/
conservative divide, there is value in considering what people who classify themselves as being on the
“conservative” side of the ledger are like, as compared to those on the “liberal” side, in more categori-
cal terms. In addition, using a regression-based approach for Studies 2 and 3 would be fairly cumber-
some, due to the large number of topic domains (with topic domain serving as one of the primary
IVs). Thus, for ease of testing interactions, an ANOVA-based approach that uses categories for politi-
cal ideology is more practically useful.
Persons who scored above the midpoint were categorized as conservative, while those below the
midpoint were categorized as liberal. This removed participants right at the midpoint, leaving 1,323
for our main analyses in Study 2 (liberal 5852; conservative 5471).
Results and Discussion
Analyses were first performed within a 2 (Ideology: Leans Right versus Leans Left) 313 (Topic
Domain) Factorial ANOVA. By far the strongest effect was for Topic Domain F(12, 1297) 59.30,
p<.001. No main effect of ideology emerged, with Conservatives (M51.72) and Liberals
(M51.72) having virtually identical overall complexity means. However, an Ideology 3Topic
Domain interaction emerged, with Conservatives higher on some topics and Liberals higher on others,
interaction F(12, 1297) 52.03, p<.02.
To understand the degree that this interaction was driven by conservative or liberal complexity
(or both equally), we created some ad hoc categories representing the upper tertile of topics on which
conservatives and liberals, respectively, were highest in complexity (defined by the difference
between conservatives and liberals on complexity for that topic).
Topics for which Conservatives
This ad hoc strategy is primarily an organizing device to simplify data analyses and presentation. First, one of the
most important considerations in the domain ideology 3interaction is whether or not conservatives show an equal
effect on their highest-complexity topics as liberals do on their highest-complexity topics. The ad hoc strategy we
employ is useful for quickly illustrating that the nature of the interaction across topics is equal on both sides in a
manner that allows for easy comparison across studies. Second, this strategy helps simplify additional analyses (e.g.,
it provides a straightforward way to test the effect of potential explanatory mechanisms in the “Additional Analyses
of Studies 1–3” section). Although we recognize that this method has the potential of exaggerating the strength of the
interaction effect, it is important to note that two of the three domain 3ideology interaction terms are significant
(Studies 2 and 4) without any ad hoc organizing—and the one that is not significant (Study 3) has such small cell
numbers and so many topic domains that it would be hard to find an interaction term. Thus, we think this ad hoc
method is constructive way of summarizing these studies that accurately captures the nature of the data.
Ideology and Complexity 9
were higher were: Death penalty is barbaric and should be abolished, Socialism, Refugees, and
George W. Bush. Topics for which Liberals were higher on complexity were: People should find out
if they are sexually suited before marriage, Bible truth, Alcohol, and Censorship. (These analyses
remove the middle tertile of topics for which liberals and conservatives were roughly equal in
A 2 (Ideology) 32(Ad Hoc Domain Type: Conservatives Higher or Liberals Higher) ANOVA
revealed, predictably, an interaction between Ideology and Domain Type, F(1, 734) 512.31,
p<.001. As can be seen in Table 1, this interaction is clearly a true crossover interaction, with conser-
vatives scoring higher than liberals on some topics, while the reverse is true for others. Indeed,
within-domain type correlations showed roughly similar effect sizes for topics on which conservatives
were higher in complexity (conservatism-complexity r[346] 5.12, p<.03) and topics on which con-
servatives were lower in complexity (conservatism-complexity r[511] 5-.08, p<.06; Fisher’s Z-test
comparing correlations 52.87, p<.01).
In other words, the interaction between domain type and ideology is roughly equally attributable
to the fact that conservatives were sometimes higher than liberals on complexity as it is to the fact
that liberals were sometimes higher than conservatives.
Study 3
Participants. Over a two-year span, 728 undergraduate participants at the University of Montana
completed an open-ended question pertaining to an array of attitudes in exchange for course credit,
usually in large sessions exceeding 100 persons.
Complexity Question Stems. The 30 question stems were taken directly from a research program
that was not motivated by ideology, but rather because they had been previously assessed for their her-
itability in a project (Olson, Vernon, Harris, & Jang, 2001) that was independent of the heritability
program used in Study 2.
Complexity Coding. As in Study 2, Study 3 was scored by 4–5 trained coders in “blocks” of
around 400 responses each. For each block, every coder of that block scored all participants (and thus
all topics) for that block. Thus, summary scores provided below are the average of 4–5 coders for
each participant. To check for reliability, we computed standardized alphas for each block separately.
Reliability on each of the different blocks was satisfactory, with standardized alphas for dialectical
complexity ranging from .81 to .82 (M5.81).
Political Ideology. We measured ideology in a manner identical to Study 2 and further com-
puted a single categorical score identical to that study. This method (which removes participants scor-
ing directly at the midpoint of the scale) left 633 for our main analyses (liberal 5395;
conservative 5238).
Studies 2 and 3 have 423 overlapping participants (those participants completed an item used both in Study 2 and in
Study 3). As in prior research using this dataset (Conway et al., 2008; Conway et al., 2011), we treat these overlap-
ping responses from each participant as independent. While it is conceptually possible that (for that subsample of 423
only) participants’ responses on the first set influenced their responses on the second set, we statistically controlled
for that possibility by computing all key analyses on that sample of 423 for each set, while statistically controlling for
the participants’ complexity from the other set. These analyses revealed no change when the other topic set was
accounted for: Participants showed a pattern of results identical, both descriptively and inferentially, to the same anal-
yses without controlling for variables for the other topic set. This demonstrates the validity of treating participants’
responses to those attitude items independently (see Conway et al., 2008; Conway et al., 2011 for similar treatments).
Indeed, it is worth noting that the complexity from the first set was virtually uncorrelated with complexity from the
second set, r5.02, p>.70.
10 Conway et al.
Results and Discussion
Initial analyses were performed within a 2 (Ideology: Leans Right versus Leans Left) 330
(Topic Domain) Factorial ANOVA. Once again, by far the strongest effect was for Topic Domain
F(29, 573) 52.91, p<.001. No main effect of ideology emerged, with Conservatives (M51.74) and
Liberals (M51.76) having virtually identical complexity means. Unlike in Study 2, no clear Ideology
3Topic Domain interaction emerged in Study 3.
However, the descriptive data suggested that sometimes conservatives were higher than liberals
on some topics and vice versa, but that small cell sizes (in some cases n57) and the large number of
cells made it difficult to find an effect in the 2 330 ANOVA. We subsequently followed an ad hoc
procedure identical to that in Study 2: Specifically, we created some ad hoc categories representing
the upper tertile of topics on which conservatives and liberals, respectively, were highest in complex-
ity. Topics for which Conservatives were higher were: Wearing clothes that draw attention, Exercis-
ing, Death penalty, Open-door immigration, Smoking, Reading books, Castration, Loud music, Roller
coaster rides, and Easy access to birth control.
Topics for which Liberals were higher on complexity were: Being assertive, Organized religion,
Crosswords, Public speaking, Abortion on demand, Big parties, Playing organized sports, Making
racial discrimination illegal, Education, Being the center of attention. (This analyses removes the
middle tertile of topics for which liberals and conservatives were roughly equal in complexity.)
A 2 (Ideology) 32 (Ad Hoc Domain Type) ANOVA revealed an interaction between Ideology
andDomainType,F(1, 374) 521.39, p<.001. As can be seen in Table 1, this interaction is clearly a
true crossover interaction, with conservatives scoring higher than liberals on some topics, while the
reverse is true for others. Indeed, within-domain correlations showed roughly similar effect sizes for
topics on which conservatives were higher in complexity (conservatism-complexity r[230] 5.15,
p<.03) and topics on which conservatives were lower in complexity (conservatism-complexity
r[207] 52.20, p<.01; Fisher’s Z53.67, p<.01).
In other words, the interaction between domain type and ideology is roughly equally attributable
to the fact that conservatives were sometimes higher than liberals on complexity as it is to the fact
that liberals were sometimes higher than conservatives.
Additional Results from Studies 1–3
To this point, we have talked loosely about ideology 3domain interactions without clearly speci-
fying the psychological explanatory variables that might account for such interactions. This is partially
purposeful—our primary goal in this article is to discuss the potential for such interactions at a large
Although there are solid conceptual and practical reasons for treating liberalism/conservatism as a dichotomous vari-
able, we ran several sets of additional analyses using conservatism as a continuous variable. First, we used regression/
correlation analyses to test the key interactions from Studies 2 and 3 while keeping political conservatism as a contin-
uous measurement. In particular, we (1) correlated political conservatism with complexity within each topic domain,
(2) created an ad hoc dummy variable (215liberals higher, 115conservatives higher) representing the top and bot-
tom tertile for the conservatism-complexity relationship, then (3) ran a regression entering standardized political con-
servatism, topic domain, and their interaction term on complexity. Results were consistent with those presented in the
text for the categorical measurement of political conservatism: For both Study 2 and Study 3, there was no main
effect of political conservatism (betas 5.01 and 2.03), but a significant interaction between conservatism and topic
domain (betas >.12, p’s <5.001).
For Studies 2 and 3, we further tested for curvilinear effects of conservatism in a manner identical to Study 1 by cre-
ating two nonlinear political conservatism terms. A summary of these results is that (1) the difference in the
conservatism-complexity relationship between conservative-higher and liberal-higher topics remains when accounting
for nonlinear terms, (2) nonlinear effects overall accounted for a proportionally smaller amount of the variance than
linear effects. In short, the effects reported here are better described as linear effects than curvilinear effects. A
detailed report of these additional results is available upon request.
Ideology and Complexity 11
level and not to engage in a debate about the specific psychological mechanisms underlying those
interactions. However, on a subset of our data for Studies 1–3, we collected a few variables relevant
to mechanisms that we believed might help explain those interactions. (We did not collect any of
these additional variables for Study 4.) Thus, while not our primary purpose, we briefly summarize
the outcome of those analyses here. Although we measured other variables (for a summary, see foot-
note 11), we specifically focus on a set of variables related to a given domain’s attitude strength.
Conceptually and empirically, attitude strength/involvement is negatively related to dialectical
forms of complexity (see Conway et al., 2008; Conway et al., 2011). Thus, it is possible that liberals
and conservatives differ on what topics they hold strong attitudes on, and this fact might help us better
understand the domain-specific effects of conservatism on complexity. We tested the implications of
this in Studies 1–3.
Study 1. For Study 1, all participants who completed the domain-specific dogmatism scales also
completed several measurements relevant to attitude strength/involvement with the topic domain: (1)
topic importance, (2) involvement with the topic, (3) experience with the topic, (4) confidence in their
opinion on the topic, and (5) attitude extremity (represented by how far they were from the midpoint
in absolute terms on an agreement item). We converted each measurement to a z-score and averaged
them into an overall attitude strength score.
Conceptually, if conservatives and liberals differed in what topics they were dogmatic about due
to differences in attitude strength, then liberals should hold stronger attitudes on the topics they were
more dogmatic on (and vice versa for conservatives). This conceptual logic directly predicts an ideol-
ogy 3topic type interaction on attitude strength, such that liberals would hold weaker attitudes on
topics for which they scored lower in dogmatism (in this case, religion), while conservatives would
hold weaker attitudes on topics for which they scored lower in dogmatism (in this case, environmental
issues). We tested this by computing an ideology (liberal versus conservative) 3topic type (environ-
ment versus religion) interaction on attitude strength.
Results were consistent with our conceptual logic: Conservatives held stronger attitudes for the
domain on which they were more dogmatic (conservatives 5.27, liberals 5.10), while liberals held
stronger attitudes for the domain on which they were more dogmatic (conservatives 52.34, liberals 5
2.04), interaction F(1,304) 58.69, p<.01.
Studies 2 and 3: Attitude Strength/Involvement. We performed a similar set of analyses on
Studies 2 and 3. In particular: On a subset of our data that comprised part of the sample for Studies 2
and 3 (n5423), immediately after completing the open-ended statements that were scored for com-
plexity, participants also completed measurements of attitude strength identical to those used in Study
1. As in Study 1, we converted each measurement to a z-score and averaged them into an overall atti-
tude strength score.
Conceptually, if conservatives and liberals differed in what topics they were more complex about
due to differences in attitude strength, then liberals should hold stronger attitudes on the topics they
were less complex on (and vice versa for conservatives). Thus, we computed ideology (liberal versus
conservative) 3ad hoc topic type (liberal higher in complexity versus conservative higher in com-
plexity) interactions.
Results were largely descriptively consistent with attitude strength being an explanatory variable,
but the inferential statistics were not overwhelming. In particular, for Study 2, conservatives held
stronger attitudes for topics on which liberals were higher in complexity (conservatives 5.16,
liberals 52.04), while liberals held stronger attitudes for topics on which conservatives were higher
in complexity (conservatives 52.07, liberals 5.00). However, this interaction was not statistically
significant (p5.184). A similar pattern emerged for Study 3, where conservatives held stronger atti-
tudes for topics on which liberals were higher in complexity (conservatives 5.21, liberals 52
while this difference was essentially not in evidence for topics on which conservatives were higher in
12 Conway et al.
complexity (conservatives 5.05, liberals 5.04). However, this interaction was also not statistically
significant (p5.262).
Study 4: 2004 Bush-Kerry Debates
George W. Bush has often been discussed as a prototypical representative of simple-minded con-
servatives (e.g., Simonton, 2006). But if, as we have argued, complexity is largely driven by topic
domain, then it is important to consider what domain politicians are talking about. In Study 4, we
compare Bush versus his opponent in the 2004 election campaign, John Kerry, across 15 different
topics that were discussed during the presidential debates.
Paragraph selection. Across three presidential debates, the two candidates were specifically
directed to discuss 15 different topics, ranging from domestic issues (e.g., the economy) to moral dis-
cussions (e.g., abortion) to foreign policy (e.g., the Iraq war). From each debate, we randomly selected
five paragraphs per topic per candidate. If the candidate did not provide five paragraphs in a given
debate for a given topic, we used all the available paragraphs for that topic.
Paragraph preparation and scoring. As is standard in archival integrative complexity research
(e.g., Conway & Conway, 2011; Suedfeld & Rank, 1976; Tetlock, 1984; Thoemmes & Conway,
2007), we removed all information from the selected paragraphs that might directly identify who the
speaker is and replaced that with generic information, and then we presented the paragraphs in ran-
dom order to four trained scorers. All scorers coded all paragraphs. Interrater reliability for the current
project was satisfactory (dialectical complexity alpha 5.75). The four coders’ scores were averaged
into a single complexity score.
Results and Discussion
Analyses were first performed within a 2 (Candidate: Bush Versus Kerry) 315 (Topic Domain)
Factorial ANOVA. By far the strongest effect was for Topic Domain F(14, 65) 52.40, p<.01. No
main effect of ideology emerged, with Conservative Bush (M51.45) and Liberal Kerry (M51.48)
having virtually identical overall complexity means. However, an Ideology 3Topic Domain interac-
tion emerged, with Bush higher on some topics and Kerry higher on others, interaction F(14,
65) 51.88, p<.05.
To understand the relative strength of Bush’s and Kerry’s complexity in contributing to this inter-
action, we created some ad hoc categories by grouping topic types together on which Bush’s and
Kerry’s complexity differed. Specifically, we created difference scores for each topic representing the
degree that either Bush or Kerry was higher on complexity for that topic, and then used the upper ter-
tile of these difference scores for each candidate to create two categories: The five topics for which
Bush was higher in complexity and the five topics for which Kerry was higher. (This strategy is analo-
gous to the strategy for creating ad hoc categories in Studies 2 and 3). Using these criteria, the topics
For Studies 2 and 3, we also performed similar exploratory analyses using several measurements relevant to attitude
discrepancy from consensus opinion (both real and perceived), the degree to which participants perceived consensus
to exist on the issue in question, and the amount of effort they put into writing the topic. No significant interaction
effects—and no clear pattern—emerged across the two studies on any of these variables. Finally, for Study 2 only, we
had a measurement of the value pluralism participants felt relevant to the topic they wrote about (constructed in a
manner drawn from Tetlock, 1986). While—consistent with theorizing from Moral Foundations Theory (see Graham,
Haidt, & Nosek, 2009; Joseph et al., 2009)—conservatives had more overall value pluralism than did liberals
(p5.01), conservatism did not interact with topic type to predict value pluralism (interaction p>.89), and thus value
pluralism cannot offer a clear explanation as to why topic type moderated the effect of conservatism on complexity.
Ideology and Complexity 13
on which Bush was higher than Kerry were: Religion, Terrorism/Homeland Security, Stem Cells,
Healthcare, and Affirmative action. The topics on which Kerry was higher than Bush were: Iraq, non-
Iraq foreign policy issues, economic issues, Abortion, and Education.
A 2 (Candidate) 32 (Ad Hoc Domain Type: Bush Higher or Kerry Higher) ANOVA revealed an
interaction between Ideology and Domain Type, F(1, 58) 515.44, p<.001. As can be seen in Table 1,
this interaction is clearly a true crossover interaction, with Bush scoring higher than Kerry on some
topics, while the reverse is true for others. Indeed, comparisons within-domain type showed roughly sim-
ilar effects for topics on which Bush was higher in complexity (effect of candidate t[28] 52.93, p<.01)
and topics on which Kerry was higher in complexity (effect of candidate t[30] 522.63, p<.02).
In other words, the interaction between domain type and ideology is roughly equally attributable
to the fact that Bush was sometimes higher than Kerry on complexity as it is to the fact that Kerry
was sometimes higher than Bush.
General Discussion
Are conservatives simple-minded? The present results suggest the answer to this question is
yes...but only on some topics.” On other topics, conservatives are more complex than liberals. Using
a large array of topic domains and methods, we found that the ideology-complexity relationship is per-
haps best described as an interaction between ideology and topic domain.
Is this interaction hiding a larger main effect of conservative simplicity? Although we found no
evidence here of the much-assumed main effect difference between liberals and conservatives, our arti-
cle is not an attempt to definitively answer that question with a “no.” Rather, the results presented here
suggest that more caution should be given to definitively answering that question with a “yes.” It may
be that liberals are, as many have claimed (Jost et al., 2003; Tetlock et al., 1996; Van Hiel et al., 2010)
and as some of even our own prior work suggests (Conway et al., 2012; Thoemmes & Conway, 2007),
pulled towards complexity more than conservatives: But we think such a judgment is premature.
What Mechanism Might Explain Domain Differences in Complexity?
Our primary purpose in this article is not to explain domain differences but to demonstrate
domain 3ideology interactions. It is nonetheless important moving forward to address what
For Studies 2–4, we tested whether the absolute value of the effect sizes for these ad hoc topic-type comparisons dif-
fered for topics on which conservatives versus liberals were higher. In Studies 2 and 4, the effect was slightly stronger
for topics on which conservatives were higher; for Study 3, the effect was slightly stronger for topics on which liber-
als were higher. However, across all three studies, Z-tests comparing the absolute value of these effect sizes revealed
little evidence that they differed (Study 2 Z50.58; Study 3 Z50.54; Study 4 Z50.24; all p’s >.56). In other words,
the absolute values of the ideology-complexity effects are essentially equivalent between conservative-higher and
liberal-higher topics in Studies 2, 3, and 4. This is consistent with the interpretation offered in the text.
We also performed analyses for Studies 2–4 for integrative complexity. As occurred for the dialectical complexity
results reported in the text, no significant main effects for ideology on complexity occurred on integrative complexity
in Studies 2–4. Also the same as for dialectical complexity, for Studies 2 and 3 for integrative complexity there was a
significant main effect of topic domain. The main difference between integrative complexity and dialectical complex-
ity results occurred in Studies 2 and 4 (recall that Study 3 did not have an initial ideology3domain interaction):
Namely, the initial domain 3ideology interactions (using all topic domains in those studies) were not significant for
integrative complexity. However, following a similar ad hoc strategy for Studies 2, 3, and 4 using integrative com-
plexity (as opposed to dialectical complexity) yielded significant interactions in each case (interaction p’s <.01), and
the overall pattern is very similar. Also, Study 4 showed no main effect difference for topic domain for integrative
complexity. Given our focus on dialectical forms of complexity throughout this article, these inferential differences
are largely irrelevant. We report them here for completeness. Our larger point remains the same: For the dialectical
forms of complexity most directly related to the rigidity of the right idea, the pattern presented here is better captured
by domain ideology interactions than by a main effect of ideology.
14 Conway et al.
psychological aspects of topic domains might help us better understand when conservatives are more
or less complex than liberals.
We tested several possibilities on a subsample of our data, and the most promising explanatory
variable to emerge was attitude strength. Conservatives and liberals differ on the topics for which they
hold strong attitudes; and it may be that this variability in attitude strength helps us understand vari-
ability on complexity. Our data reveal evidence that is modestly consistent with this hypothesis:
Across all three studies for which data was available, both conservatives and liberals held stronger
attitudes for the topics on which they were more simple-minded (operationalized in Study 1 as topics
for which they were more dogmatic and in Studies 2 and 3 as topics for which they were less com-
plex). Although this pattern was weaker for Studies 2 and 3,
these results provide some preliminary
evidence that the ideology 3domain interactions on complexity-relevant variables may be partially a
function of domain differences in attitude strength.
This suggests two possible ways forward for future researchers. First and most obviously, it
would be useful to run studies with greater power for testing the explanatory ability of attitude
strength measurements. This could be done by collecting a larger group of participants, but it also
might be useful to select topics a priori that were especially prone to show attitude strength differences
between liberals and conservatives and run more focused tests with those domains or otherwise
directly manipulate attitude strength (rather than merely measuring it as we did in the present studies).
Second, the measurements reported here are only one set of possible (and potentially imprecise) meth-
ods for measuring the attitude strength construct. Given the potential explanatory value of attitude
strength in our understanding of complexity-related domain variability, future work would do well to
include more sophisticated methods of measuring attitude strength that do not rely solely on direct
self-report (e.g., Fazio, Jackson, Dunton, & Williams, 1995).
Sample Limitations
This research is not without its limitations. First, our work is entirely constrained to U.S. samples
and thus should be interpreted with appropriate caution. We do not know if the results presented in
Studies 1–3 would generalize beyond U.S. borders, and the results from Study 4 are from a single
U.S. election.
What might this mean for our interpretation of the results? First, it is important to note that much
of the case for conservative simplicity has been compiled on U.S. samples. For example, in Jost
et al.’s (2003) meta-analysis, 81% of the reported Nfor integrative complexity and 44% of the
reported Nfor dogmatism occurred on U.S. participants. Since part of our aim is to offer a potential
alternative explanation for existing evidence in these areas, our data at least suggest that for the part
of the current case which has been built on U.S. samples (a fairly large percentage), we should per-
haps reconsider our collective interpretation of the evidence.
However, it is of course still worth considering the potential effect of the cultural context on
our results. Consider, for example, that one of the most salient differences distinguishing the
United States from other potential Western samples is that the United States as a whole is more
conservative than many other Western nations. What effect might this have? According to one of
the most influential models of the origins of complex thinking, Tetlock’s Value Pluralism Model
(e.g., Tetlock, 1986; Tetlock et al., 1996), if this had any effect, it would be likely to bias the
sample by making U.S. liberals more complex than liberals in other places (in particular on
For Studies 2 and 3, it is worth noting that we only had relevant data on a subset of our larger sample. Although atti-
tude strength measurements did not show significant effects in those studies, they were in the correct direction and
showed a similar pattern across both studies. We do not want to overinterpret these data; it is possible that this pattern
does not represent a real finding. However, we view it as likely that a larger sample would find the expected signifi-
cant effect for attitude strength.
Ideology and Complexity 15
dialectical forms of complexity; see Conway et al., 2008).
If the United States leans right as a
nation, that means that those we call “liberals” are actually more “centrist”—and it is centrists
that the Value Pluralism Model (and a related model of belief defense; for discussions, see Con-
way et al., 2008; Conway et al., 2011) predicts would be most complex. As a population, this
would make the United States a skewed test that would more likely make liberals complex,
because liberals in the United States would actually be less likely to be true liberals—but would
in fact be moderates who are more likely on average to be complex (see, e.g., Tetlock et al.,
1994). Thus, in a sense, our finding no main effect (and instead finding domain 3ideology inter-
actions) in a U.S. sample is more—rather than less—compelling evidence against the conserva-
tive simplicity hypothesis.
Consistent with this notion, it is worth noting that additional research on integrative com-
plexity and related measurements—research not cited in Jost et al. (2003)—suggests that if
anything, the conservative simplicity effect may be less likely to be found in non-U.S. contexts.
For example, Van Hiel and Mervielde (2003) correlated political conservatism amongst Bel-
gian college students with three measurements of complexity, and none of the measurements
was statistically significant. Further, in Soviet Russia, leftist communist leaders were less inte-
gratively complex than capitalistic reformers who had a more conservative ideology (Tetlock
& Boettger, 1989). Quite a bit of research on Canadian political parties is also inconsistent
with the conservative simplicity hypothesis. For example, comparing Canadian college stu-
dents affiliated with two conservative and two liberal political parties, Suedfeld et al. (1994)
found no significant differences among them, and the highest overall group on complexity was
the conservative Progressive Conservative party. Similar results for Canadian politicians were
found by Suedfeld’s (2000) scoring of the major political parties’ candidates for Prime Minister
during the 1997 Federal election. Finally, Lavallee and Suedfeld (1997) found that more liberal
environmental groups scored lower in integrative complexity (though not significantly so) than
more conservative forest advocacy groups in a conflict in British Columbia over Clayquot
In summary, we do not view it is likely that our results will end up being relevant to only the
United States—they may indeed be more powerful in other regions.
However, it is further worth
noting that to the degree that our sample does in fact differ from other samples and would yield differ-
ent results, this itself poses a problem for the rigidity of the right explanation. Even should our results
prove to be specific only to the United States, this suggests at a minimum that cultural context is a
potential moderator of the conservatism-complexity relationship. Given how alarmingly few cultural
contexts have actually been tested, this leaves open the possibility that, averaged across all cultures,
we may find a far weaker (or even nonexistent) main effect for the conservatism-complexity
The relationship of extremism/value pluralism to complexity itself is more complex than this implies. Some work
shows that more extreme views lead to more complexity (e.g., Conway et al., 2008; Sidanius, 1984). Conway et al.
(2008) attempted to resolve this apparent conflict by demonstrating that extremism leads to less dialectical complexity
and more elaborative complexity. Because the focus of the “rigidity of the right” hypothesis is on dialectical forms of
complexity, and most of the evidence in favor of it has focused on those forms, we opt to discuss the relationship
here and elsewhere in terms of that well-established “extremism reduces complexity” pattern.
There is a potential tension here. These results highlight that other studied nations may show less of a tendency for
conservative simplicity, which is best construed as a main effect of context on the effect in question. It is possible
that ideology topic domain interactions—the topic of this article—could conceivably be less in those nations, even as
they are simultaneously showing less support for conservative simplicity. While we acknowledge this potential, the
tension between the main effect and interactions in other nations is beyond the scope of this article. Our point here is
that it is likely unreasonable to imagine that different contexts outside of the United States would produce a wildly
different landscape in terms of the ideology-complexity relationship, and based on what we know, if they produce
anything different, that difference would likely not be favorable to the conservative simplicity hypothesis.
16 Conway et al.
Thus, while we cannot of course definitively say that our work would generalize, our work at
least makes it clear that we should pause in our larger conclusions about the relationship between con-
servatism and domain-general simplicity.
Research Content Limitations
Our discussion has covered two of the major constructs in the argument for conservative
simplicity: dogmatism and integrative complexity. But the case for conservative simplicity
includes much more than just research on those constructs, and it includes quite a bit of evidence
our domain-specificity theory does not account for even within those two lines of work. It does
not seem very likely that ideology 3domain interactions account for all of the additional evi-
dence discussed by Jost et al. (2003). As such, caution is warranted in interpreting these results
in the larger picture.
Indeed, consider three lines of evidence discussed in Jost et al. (2003) and Van Hiel et al.
(2010): Need for closure/structure, openness, and preference for complexity (e.g., preferences
for complex visual images and complex poetry). In each case, quite a bit of evidence exists tying
political conservatism to a motive for simplicity; conservatives are higher in self-reported need
for closure/structure (and, in a more nuanced recent account, liberals who are higher in need for
closure show more conservative policy positions; see Federico et al., 2012), lower in self-
reported openness, and lower in self-reported preferences for complex images or poems. Each of
those areas could potentially capture a more domain-general motive—for example, need for clo-
sure is conceptualized as a need for nonspecific closure that cuts across domains—and yet our
research cannot directly speak to that work. How, then, do we reconcile our work to this prior
work? We discuss three different points of intersection below (see the online supplementary
information for additional discussion in this regard).
The multidimensionality of complexity and why it matters. Even if we assume that this prior
work on need for closure/structure, openness, and preferences for complex materials represent phe-
nomena on which conservatives are indeed simpler, that would not invalidate the importance of our
present findings. As many researchers have pointed out, complexity itself is multifaceted (see, e.g.,
Conway et al., 2014; Houck et al., 2014; Tetlock, Emlen Metz, Scott, & Suedfeld, 2014). Thus, even
if the present results turn out to be limited only to dogmatism and integrative complexity—two
aspects of the case being made for conservative simplicity—that would nonetheless suggest for com-
plexity relevant to those types, the general case being made against conservatives in those areas needs
revision. This would present to us a more nuanced and accurate picture of the relationship between
ideology and complexity and suggest at the least that the current picture must be qualified by the type
of complexity measurement under the microscope.
The potential pitfalls of self-report measurement. Further, for the purposes of determining the
average complexity of a group of persons, the kinds of self-report measurements comprising the bulk
of prior meta-analyses have some additional potential pitfalls in interpretation. A self-report measure-
ment is at best an indirect marker of a potential complexity-relevant motive. To take one example,
Person A may be higher than Person B on the need for closure scale because they are actually more
motivated for closure; but they also may be higher because they are more willing to report higher
need for closure (even though both persons may actually have the same level of motive), or have a dif-
ferent set of social desirability templates, or a host of other factors that do not have to do with the
actual motive.
Importantly, many of these self-report scales that are purportedly domain-general may also contain very specific
domain-related content. (For discussion of this possibility relevant to openness, see Sibley & Duckitt, 2008; Sibley,
Osborne, & Duckitt, 2012. For discussion relevant to need for closure, see the supplementary information.)
Ideology and Complexity 17
Further, self-report measurements vary in their degree of clear overlap with complexity. Consider
that the need for structure scale was intentionally designed to avoid explicitly mentioning cognitive
structure (see, e.g., Neuberg & Newsom, 1993). Both the need for structure scale and the highly corre-
lated and conceptually similar need for closure scale contain items that appear like conscientiousness
as much as cognitive structure, and the need for structure scale is in fact correlated with both self-
reported conscientiousness and conscientious behaviors (Neuberg & Newsom, 1993). While need for
structure/closure has been sometimes correlated with more face-valid measurements of complexity
(e.g., Neuberg & Newsom, 1993; Webster & Kruglanski, 1994) and is related to attitudinal “seizing”
that seems clearly conceptually related to complex thinking (e.g., Webster & Kruglanski, 1994), it is
nonetheless worth considering that the scales do not directly measure a desire to think in a complex
In contrast, the scoring of open-ended statements is a much more direct marker of the aver-
age complexity of particular groups (see Houck et al., 2014). While it, too, has its drawbacks,
this at the very least means that when Person A scores higher than Person B, we can feel more
confident that the output of Person A is indeed more complex than Person B on that domain.
Thus, this provides a more direct marker of the outcome of interest—complexity—than do more
indirect measurements of a self-reported motive that is expected to be correlated to the outcome.
The importance of the potential divergence between self-report and open-ended measures is
underscored by the fact that the two types of measurements are often not very highly correlated
(see Van Hiel et al., 2010, for a summary).
Empirical scope of the present research. Our work shows a different pattern of results than
quite a bit of prior work. Why is that the case? There are three conceptual reasons why the literature
writ large might differ with the results presented here. The first is the primary argument made in this
article: That prior work has not fully accounted for domain 3ideology interactions. But it may also
be that our results are simply anomalous. After all, for any real effect in psychology, sometimes you
will not find that effect, or find a reversal of the effect, just by chance.
Relatedly, it is also possible that our results represent an overselection of topics on which
conservatives score more complexly. Even if one grants the power of ideology 3domain interac-
tions, it of course does not follow that there is no meaningful main effect (see, e.g., Tetlock
et al., 1996). Conservatives may be more complex than liberals on a certain handful of topics,
but this may run counter to the general tendencies outlined by Jost et al. (2003) to pull them, at
some larger psychological level, towards simplicity. Indeed, it is possible that we ourselves are
exhibiting a bias in topic selection: Although the topics selected here (with a few exceptions)
were not largely selected for reasons directly related to ideology at all (see Conway et al., 2008;
Conway et al., 2011), it is nonetheless certainly possible that we selected a group of topics on
which conservatives are particularly likely to score high in complexity, thus offsetting the larger
tendency for liberals to score higher.
We acknowledge the possibility, as well as the possibility that our results are simply anom-
alous. But, in considering the scope of our research set against the existing body of work, it is
also worth noting that most prior research on ideology and complexity is very narrow in its
scope of possible topics, whereas our research covered 43 separate (though sometimes overlap-
ping) topic stems for the student population and 15 separate topic categories for the Bush/Kerry
comparison. For comparison, in the integrative complexity research cited in Jost et al. (2003),
only one topic distinction is made within-study at all, and it onlyoccurredforonestudyand
included only two broad topic types. Thus, although a lot of research indicates that liberals are
higher than conservatives in complexity, it is unclear whether or not that research covers a
wide range of topics.
About integrative complexity specifically, it is further worth noting that prior research on ideology
and integrative complexity has been mostly limited to scoring politicians (Jost et al., 2003; Van Hiel
18 Conway et al.
et al., 2010), and thus may not apply to the vast majority of the population.
Further, our research cov-
ers over 2,000 participants, which is more participants than all the integrative complexity research com-
bined cited in Jost et al.’s (2003) meta-analysis (combined participant Nfor all integrative complexity
studies 5307). So while it would be premature to suggest overturning all prior evidence on the basis of
our work presented here, it would similarly be premature to dismiss our evidence as having no bearing
on the larger question. We do not know yet whether our topic selection method might show bias in
favor of conservatives, and our evidence contains the largest set of domains studied to date.
Concluding Thoughts
Even the best research is potentially subject to qualification through scientific scrutiny. Our pur-
pose here is not to claim definitively that conservatives are equally as complex as liberals. We are cer-
tainly open to the possibility that conservatives are simpler on average, and agree with Tetlock (1986)
and Jost et al. (2003) that there are reasons it may be so.
Our purpose is much more modest: It is to point out that, just as there are reasons why it may be
so, there are also reasons it may not be so. We have presented one alternative possibility here and
offered some empirical evidence in support of that model. As such, we believe that the present
research might fit in with a growing body of work suggesting that negative attributes once attributed
to conservatives might be domain specific (see, e.g., Brandt, Reyna, Chambers, Crawford, & Wether-
ell, 2014; Crawford, 2012; Suedfeld, Bochner, & Wnek, 1972). We hope this to be the beginning of
the discussion with respect to complexity and have no illusions of it being the last word.
Portions of this article were presented at the 2012 Annual Meeting of the International Society
of Political Psychology, Chicago, IL. Correspondence concerning this article should be addressed to
Lucian Gideon Conway III, University of Montana, Psychology Department, Skaggs Building 239,
Missoula, MT, 59812. E-mail:
A recent study (Brundidge, Reid, Choi, & Muddiman, 2014)—using a computer-based complexity measurement—analyzed
the complexity of over 500 blog posts from liberals and conservatives and found that liberal blog posts were more complex
than conservative blog posts (r5.20). They also scored for four different topic types and found no clear interaction pattern
for ideology and domain. These data are clearly relevant to the larger issue of differences between conservatives and liber-
als, but it is less clear how relevant they are to integrative complexity specifically. Several considerations include: (1)
Computer-based measurements are at best an indirect approximation of linguistic complexity (see Tetlock et al., 2014). It
is noteworthy that the LIWC complexity measurement used by Brundidge et al. (2014) was based on a measurement that
has shown fairly low correlations with human-scored integrative complexity (r5.14; Conway et al., 2014). This low corre-
lation means that differences between liberals and conservatives on the measurement may very well have to do with some
other linguistic property besides complexity itself. In contrast, human-scored integrative complexity on which prior meta-
analyses are based—and which are the subject of this paragraph—is a much more direct measurement of the actual struc-
tural complexity of a paragraph (see., e.g., Tetlock et al., 2014). Thus, while clearly a relevant piece of evidence that
should be considered, we should be careful in overinterpreting such indirect data. (2) It is further noteworthy that, while
not finding domain 3ideology interactions, the study had a low number of domains (four) when compared to the present
work. (3) The study does not deal directly with lay populations, instead focusing on political pundits who author blogs.
Thus, even if interpreted as representing real complexity differences among liberals and conservatives, it should be inter-
preted with caution in terms of the potential breadth of application. It is entirely possible that the reported effects between
liberals and conservatives are in actual fact very specific only to political elites and/or political pundits; if so, that is an
important and meaningful fact but is nonetheless a comparatively small percentage of the population—and this would
require a revisiting of the underlying cause of the differences between the two groups. In summary, while we acknowledge
that the work has value and relevancy, there is also good reason to interpret the work very cautiously as indicative of inte-
grative complexity differences for liberals and conservatives more broadly.
It is also important to note that we are not arguing that Jost et al.’s (2003) theory is wrong. In fact, the theory could
be completely right and yet conservatives and liberals could still be equally complex. Please see the supplementary
information for additional discussion on this issue.
Ideology and Complexity 19
Baker-Brown, G., Ballard, E. J., Bluck, S., de Vries, B., Suedfeld, P., & Tetlock, P. E. (1992). The conceptual/integrative
complexity scoring manual. In C. P. Smith (Ed.), Motivation and personality: Handbook of thematic content analy-
sis (pp. 605–611). Cambridge, UK: Cambridge University Press.
Brandt, M. J., Reyna, C., Chambers, J. R., Crawford, J. T., & Wetherell, G. (2014). The Ideological-Conflict Hypothesis:
Intolerance among both liberals and conservatives. Current Directions in Psychological Science,23, 27–34. doi:
Brundidge, J., Reid, S. A., Choi, S., & Muddiman, A. (2014). The “deliberative digital divide:” Opinion leadership and
integrative complexity in the U.S. political blogosphere. Political Psychology. doi:10.1111/pops.12201
Conway, L. G., III, & Conway, K. R. (2011). The terrorist rhetorical style and its consequences for understanding terro-
rist violence. Dynamics of Asymmetric Conflict,4, 175–192.
Conway, L. G., III, Conway, K. R., Gornick, L. J., & Houck, S. C. (2014). Automated integrative complexity. Political
Psychology,35, 603–624.
Conway, L. G. III, Dodds, D., Hands Towgood, K., McClure, S, & Olson, J. (2011). The biological roots of complex
thinking: Are heritable attitudes more complex? Journal of Personality,79, 101–134.
Conway, L.G., III, & Gornick, L. J. (2011). Cognitive complexity. In D. Christie (Ed.), The encyclopedia of peace psy-
chology (pp. 849–853). Hoboken, NJ: Wiley-Blackwell.
Conway, L. G., III, Gornick, L. J., Burfiend, C., Mandella, P., Kuenzli, A., Houck, S. C., & Fullerton, D. T. (2012).
Does simple rhetoric win elections? An integrative complexity analysis of U.S. presidential campaigns. Political
Psychology,33, 599–618.
Conway, L. G., III, Schaller, M., Tweed, R. G., & Hallett, D. (2001). The complexity of thinking across cultures: Inter-
actions between culture and situational context. Social Cognition,19, 230–253.
Conway, L. G., III, Suedfeld, P., & Clements, S. M. (2003). Beyond the American reaction: Integrative complexity of
Middle Eastern leaders during the 9/11 crisis. Psicologia Politica,27, 93–103.
Conway, L. G., III, Suedfeld, P., & Tetlock, P. E. (2001). Integrative complexity and political decisions that lead to war
or peace. In D. J. Christie (Ed.), Peace, conflict, and violence: Peace psychology for the 21st century (pp. 66–75).
Upper Saddle River, NJ: Prentice Hall/Pearson Education.
Conway, L. G., III, Thoemmes, F., Allison, A. M., Towgood, K. H., Wagner, M. J., Davey, K., et al. (2008). Two ways
to be complex and why they matter: Implications for attitude strength and lying. Journal of Personality and Social
Psychology,95(5), 1029–1044.
Crawford, J. T. (2012). The ideologically objectionable premise model: Predicting biased political judgments on the left
and right. Journal of Experimental Social Psychology,48, 138–151.
Dasen, P. R. (1975). Concrete operational development in three cultures. Journal of Cross-Cultural Psychology,6, 156–
Duarte, J. L., Crawford, J. T., Stern, C., Haidt, J., Jussim, L., & Tetlock, P. E. (in press). Political diversity will improve
social psychological science. Behavioral and Brain Sciences.
Eaves, L. J., Eysenck, H. J., & Martin, N. G. (1989). Genes, culture, and personality: An empirical approach. London,
UK: Academic Press.
Fazio, R. H., Jackson, J. R., Dunton, B. C., & Williams, C. J. (1995). Variability in automatic activation as an unobtru-
sive measure of racial attitudes: A bona fide pipeline? Journal of Personality and Social Psychology,69, 1013–
Federico, C. M., Deason, G., & Fisher, E. L. (2012). Ideological asymmetry in the relationship between epistemic moti-
vation and political attitudes. Journal of Personality and Social Psychology. doi:10.1037/a0029063
Graham, J., Haidt, J., & Nosek, B. (2009). Liberals and conservatives rely on different sets of moral foundations. Journal
of Personality and Social Psychology,96, 1029–1046.
Houck, S. C., Conway, L. G., III, & Gornick, L. J. (2014). Automated integrative complexity: Current challenges and
future directions. Political Psychology,35(5), 647–659.
Harvey, O. J., Hunt, D. E., & Schroder, H. M. (1961). Conceptual systems and personality organization. Oxford, UK:
Joseph, C. M., Graham, J., & Haidt, J. (2009). The end of equipotentiality: A moral foundations approach to ideology-
attitude links and cognitive complexity. Psychological Inquiry,20, 172–176.
Jost, T., Glaser, J., Kruglanski, A. W., & Sulloway, F. J. (2003). Political conservatism as motivated social cognition.
Psychological Bulletin 129(3), 339–375.
Judd, C. M., & Lusk, C. M. (1984). Knowledge structures and evaluative judgments: Effects of structural variables on
judgmental extremity. Journal of Personality and Social Psychology,46, 1193–1207.
20 Conway et al.
Lavallee, L., & Suedfeld, P. (1997). Conflict in Clayoquot Sound: Using thematic content analysis to understand psycho-
logical aspects of environmental controversy. Canadian Journal of Behavioural Science,29, 194–209.
Liht, J., Conway, L. G. III, Savage, S., White, W., O’Neill, K. A. (2011). Religious fundamentalism: An empirically
derived construct and measurement scale. Archive for the Psychology of Religion,33, 299–323.
Martin, N. G., Eaves, L.J., Heath. A. R., Jardine, R., Feingold, L. M., & Eysenck, H. J. (1986). Transmission of social
attitudes. Proceedings of the National Academy of Science,83, 4364–4368.
Neuberg, S. L., & Newsom, J. T. (1993). Personal need for structure: Individual differences in the desire for simpler
structure. Journal of Personality and Social Psychology,65, 113–131.
Olson, J. M., Vernon, P.A., Harris, J., & Jang, K. L. (2001). The heritability of attitudes: A study of twins. Journal of
Personality and Social Psychology. 80, 845–860.
Pancer, S. M., Jackson, L. M., Hunsberger, B., Pratt, M. W., et al. (1995). Religious orthodoxy and the complexity of
thought about religious and nonreligious issues. Journal of Personality,63, 213–232.
Ray, J. J. (1970). The development and validation of a balanced dogmatism scale. Australian Journal of Psychology,22,
Rokeach, M. (1960). The open and closed mind: Investigations into the nature of belief systems and personality systems.
New York, NY: Basic Books.
Schroder, H. M., Driver, M. J., & Streufert, S. (1965). Information processing systems in individuals and groups. New
York, NY: Holt, Rinehart, & Winston.
Sibley, C. G., & Duckitt, J. (2008). Personality and prejudice: A meta-analysis and theoretical review. Personality and
Social Psychology Review,12, 248–279.
Sibley, C. G., Osborne, D., & Duckitt, J. (2012). Personality and political orientation: meta-analysis and test of a threat-
constraint model. Journal of Research in Personality,46, 664–677.
Sidanius, J. (1984). Political interest, political information search, and ideological homogeneity as a function of sociopo-
litical ideology: A tale of three theories. Human Relations,37, 811–828.
Simons, H. W. (1968). Dogmatism scales and leftist bias. Speech Monographs,35, 149–153.
Simonton, D. K. (2006). Presidential IQ, openness, intellectual brilliance, and leadership: Estimates and correlations for
42 U.S. chief executives. Political Psychology,27, 511–526.
Suedfeld, P. (2000). Domain-related variation in integrative complexity: A measure of political importance and respon-
siveness? Clinton, Gingrich, Gorbachev, and various Canadian political leaders. In O. Feldman & C. De Landtsheer
(Eds.), Beyond public speech and symbols: Explorations in the rhetoric of politicians and the media (pp. 17–34).
Westport, CT: Praeger.
Suedfeld, P., & Bluck, S. (1988). Changes in integrative complexity prior to surprise attacks. Journal of Conflict Resolu-
tion,26, 626–635.
Suedfeld, P., Bluck, S., & Ballard, E. J. (1994). The effects of emotional involvement and psychological distance on
integrative complexity. Journal of Applied Social Psychology,24, 443–452.
Suedfeld, P., Bluck, S., Loewen, L., & Elkins, D. (1994). Sociopolitical values and integrative complexity of members
of student political groups. Canadian Journal of Behavioural Science,26, 121–141.
Suedfeld, P., Bochner, S., & Wnek, D. (1972). Helper-sufferer similarity and specific request for help: Bystander inter-
vention during a peace demonstration. Journal of Applied Social Psychology,2, 17–23.
Suedfeld, P., & Leighton, D. C. (2002). Early communications in the war against terrorism: An integrative complexity
analysis. Political Psychology,23, 585–599.
Suedfeld, P., & Rank, A. D. (1976). Revolutionary leaders: Long-term success as a function of changes in conceptual
complexity. Journal of Personality and Social Psychology,34, 169–178.
Suedfeld, P., Steel, G. D., & Schmidt, P. W. (1994). Political ideology and attitudes toward censorship. Journal of
Applied Social Psychology,24, 765–781.
Suedfeld, P., & Streufert, S. (1966). Information search as a function of conceptual and environmental complexity. Psy-
chonomic Science,4(10), 351–352.
Suedfeld, P., & Tetlock, P. (1976). Integrative complexity of communications in international crises. Journal of Conflict
Resolution,21(1), 169–184.
Suedfeld, P., Tetlock, P., & Streufert, S. (1992). Conceptual/integrative complexity. In C. P. Smith (Ed.), Motivation
and personality: Handbook of thematic content analysis (pp. 393–400). Cambridge, MA: Cambridge University
Suedfeld, P., & Wallbaum, A. B. C. (1992). Modifying integrative complexity in political thought: Value conflict and
audience disagreement. International Journal of Psychology,26, 19–36.
Tetlock, P. E. (1984). Cognitive style and political belief systems in the British House of Commons. Journal of Person-
ality and Social Psychology,46, 365–375.
Ideology and Complexity 21
Tetlock, P. E. (1986). A value pluralism model of ideological reasoning. Journal of Personality and Social Psychology,
50, 819–827.
Tetlock, P. E., & Boettger, R. (1989). Cognitive and rhetorical styles of traditionalist and reformist Soviet politicians: A
content analysis study. Political Psychology,10, 209–232.
Tetlock, P. E., Emlen Metz, S., Scott, S. E., & Suedfeld, P. (2014). Integrative complexity coding raises integratively
complex issues. Political Psychology,35, 625–634.
Tetlock, P. E., Peterson, R. & Lerner, J. (1996). Revising the value pluralism model: Incorporating social content and
context postulates. In C. Seligman, J. Olson, & M. Zanna (Eds.), Values: The Ontario symposium on personality
and social psychology (Vol. 8, pp. 25–51). Hillsdale, NJ: Lawrence Erlbaum.
Thoemmes, F. J., & Conway, L. G., III. (2007). Integrative complexity of 41 U.S. presidents. Political Psychology,28,
Van Hiel, A., & Mervielde, I. (2003). The measurement of cognitive complexity and its relationship with political extre-
mism. Political Psychology,24(4), 781–801.
Van Hiel, A., Onraet, E., & De Pauw, S. (2010). The relationship between social-cultural attitudes and behavioral meas-
ures of cognitive style: A meta-analytic integration of studies. Journal of Personality,78, 1765–1800.
Webster, D. M., & Kruglanski, A. W. (1994). Individual differences in need for cognitive closure. Journal of Personality
and Social Psychology,67, 1049–1062.
Supporting Information
Additional supporting information may be found in the online version of this article at the pub-
lisher’s website:
Study 1 Questionnaires
Additional Discussion: The Potential for Domain-Specificity for Prior Scales.
22 Conway et al.
... providing strong evidence for its empirical distinctiveness (see Ronkko & Cho, 2020). Moreover, the primary measure of dogmatism in the literature, the DOG Scale, may be confounded with religiosity and social conservatism (Conway et al., 2016;Duckitt, 2009;see Stanovich & Toplak, 2019) 3 . Given these considerations, as well as the fact that dogmatism is quite theoretically distinct from all other "rigidity" variables (Johnson, 2009), we treat dogmatism as a standalone rigidity domain in the present review. ...
... Conservatives tend to closely embrace absolutist and intuitive ideas and practices, which is quite congenial to dogmatism (Jost et al., 2003). Second, the primary measure of dogmatism in the literature, the DOG Scale, seems to be confounded with religiosity and social conservatism (Conway et al., 2016;Duckitt, 2009;see Stanovich & Toplak, 2019). Thus, our findings may be attributable to measurement error. ...
... For example, manipulations and measures relevant to perception of terrorism-related threats are often found to predict conservatism and are consequently taken as support for the RRH(Jost et al., 2007, Study 3; Thorisdottir & Jost, 2011, Study 2). Further, many studies rely on Rokeach's Dogmatism (D) scale as a rigidity indicator, despite the presence of right-wing political content in this scale (seeConway et al., 2016). Similarly, the longstanding finding that political conservatism is associated with prejudice (see Hodson & Dhont, 2015, for a review) appears to dissipate when groups that are perceived as ideologically dissimilar to political liberals, such as Christian fundamentalists and wealthy individuals, are included as targets in measures of prejudice(Brandt & Crawford, 2019;Crawford, 2017). ...
Full-text available
The rigidity-of-the-right hypothesis (RRH), which posits that cognitive, motivational, and ideological rigidity resonate with political conservatism, is an influential but controversial psychological account of political ideology. Here, we leverage several methodological and theoretical sources of this controversy to conduct an extensive quantitative review—with the dual aims of probing the RRH’s basic assumptions and parsing the RRH literature’s heterogeneity. Using multi-level meta-analyses of relations between varieties of rigidity and ideology measures alongside a bevy of potential moderators (s = 329, k = 708, N = 187,612), we find that associations between conservatism and rigidity are tremendously heterogeneous, suggesting a complex—yet conceptually fertile—network of relations between these constructs. Most notably, whereas social conservatism was robustly associated with rigidity, associations between economic conservatism and rigidity indicators were inconsistent, small, and not statistically significant outside of the United States. Moderator analyses revealed that non-representative sampling, criterion contamination, and disproportionate use of American samples have yielded over-estimates of associations between rigidity-related constructs and conservatism in past research. We resolve that drilling into this complexity, thereby moving beyond the question of if conservatives are essentially rigid to when and why they might or might not be, will help provide a more realistic account of the psychological underpinnings of political ideology.
... 1 DEMOCRACY AND IDEOLOGY motivated bias in their reasoning and have more rigid thinking styles (Carney et al., 2008;Jost, 2017). Conversely, proponents of symmetry argue that both conservative and liberals are similarly willing to discriminate against one another (e.g., Brandt et al., 2014;Chambers et al., 2013;Crawford & Pilanski, 2014;Moore-Berg et al., 2020;Toner, Leary, & Asher, 2013), show similar levels of emotional and cognitive bias against one another (Ditto et al., 2019;Steiger et al., 2019), and use similarly simple thinking in approaching partisan issues (Conway et al., 2016). ...
... 1 DEMOCRACY AND IDEOLOGY motivated bias in their reasoning and have more rigid thinking styles (Carney et al., 2008;Jost, 2017). Conversely, proponents of symmetry argue that both conservative and liberals are similarly willing to discriminate against one another (e.g., Brandt et al., 2014;Chambers et al., 2013;Crawford & Pilanski, 2014;Moore-Berg et al., 2020;Toner, Leary, & Asher, 2013), show similar levels of emotional and cognitive bias against one another (Ditto et al., 2019;Steiger et al., 2019), and use similarly simple thinking in approaching partisan issues (Conway et al., 2016). ...
... Our original theorizing had also connected with counterarguments against claims of ideological asymmetry. Some argue that dangerous radicalism and political intolerance are endemic at both ends of the political spectrum, not just the conservative one (Brandt et al., 2014;Chambers, Schlenker, & Collisson, 2013;Crawford & Pilanski, 2014). Like some conservatives, some liberals may be authoritarian (Costello et al., 2020). ...
Full-text available
Despite widespread support for the principles of democracy, democratic norms have been eroding globally for over a decade. We ask whether and how political ideology factors into people's reactions to democratic decline. We offer hypotheses derived from two theoretical lenses, one considering ideologically relevant dispositions and another considering ideologically relevant situations. Preregistered laboratory experiments combined with analyses of World Values Survey (WVS) data indicate that there is a dispositional trend: Overall, liberals are more distressed than conservatives by low democracy. At the same time, situational factors also matter: This pattern emerges most strongly when the ruling party is conservative and disappears (though it does not flip into its mirror image) when the ruling party is liberal. Our results contribute to ongoing debates over ideological symmetry and asymmetry; they also suggest that, if democracy is worth protecting, not everyone, everywhere will feel the urgency. (PsycInfo Database Record (c) 2021 APA, all rights reserved).
... We used latent variable modeling to explore the DOG Scale's dimensional structure and calculate factor scores that best reflect a substantive core of dogmatism (i.e., EFA in 40% of the data followed by a CFA comparing plausible factor structures in the other 60%; see Supplemental Files 1 and 2). Still, there is reason to suspect the DOG Scale demonstrates test bias for political ideology (i.e., scores may have differential validity across political groups; Conway et al., 2016). As such, we also used a modified measure designed to assess Rokeach's (1960) conceptualization of dogmatism (see Costello, Bowes, Stevens, et al 2021;. ...
Full-text available
The present investigation examined curvilinear relations between political ideology, on the one hand, and absolute certainty and dogmatism, on the other, across six online samples (N = 2,889). Ideological extremists were more likely than others to be absolutely certain: About one in three extremists reported being absolutely (i.e., 100%) certain of the correctness of their political beliefs, whereas about one in 15 non-extremists reported being absolutely certain. Although absolute political certainty was relatively symmetrical across the political left and right, conservatives tended to report greater domain-general dogmatism than liberals. Extremism effects for domain-general dogmatism were also present, however; and ideological asymmetries in dogmatism appeared to be driven by social, rather than economic, ideology. Taken together, these findings underscore the complexity of relations between absolute certainty, dogmatism, and ideology, ultimately challenging the sufficiency of contemporary psychological accounts of ideological (a)symmetries to describe our complex political reality.
... Meanwhile, the number of respondents who indicated having mixed liberal and conservative views declined from 49% in 2004to 39% in 2014(Pew Research Center, 2014b. Many authors use political parties such as Republicans interchangeably with an ideology such as conservativism (e.g., Longo & Baker, 2014) and they are often strongly correlated (see Conway et al., 2016). ...
Full-text available
Researchers have found differences between the ideological beliefs and preferred news sources of the major political parties in the United States (U.S.). However, few researchers explored potential differences in logical reasoning by party. We sought to replicate previously published ideological differences between Democrats and Republicans, examined their rationality using items from classic decision-making literature, and investigated the role of voters' political news sources in decision-making. Surveying 239 Democrats and 117 Republicans, using Amazon Mechanical Turk, we posited three predictions. The first is that Democrats would disagree with politically conservative statements, agree that political news is being reported fairly, and disapprove of former President Donald Trump's actions. We confirmed these predictions. We also predicted that the more conservative statements participants agreed with, the lower their rationality score and this hypothesis was confirmed. Lastly, we predicted that participants who relied on right-leaning media would score lower on a rationality test than those individuals who relied on center or left-leaning media. We found that voters who relied on Fox News scored lower on the rationality test than those who relied on National Public Radio.
... For instance, extreme liberals and extreme conserva tives engage in similar levels of motivated reasoning to defend their pre-existing views Gampa et al., 2019) and similarly may avoid exposure to dissenting opinions (Frimer et al., 2017; although this may not be the case on Twitter, where liberals are more likely to share ideologically incongruent information, Barberá, Jost, Nagler, Tucker, & Bonneau, 2015). Likewise, some research suggests that the low integrative complexity typically associated with conservatism is, in fact, domain specific (Conway et al., 2016), and that liberals and conservatives engage in equal amounts of motivated disbelief depending on the topic (Campbell & Kay, 2014). Those with extreme political identities show greater cognitive inflexibility (Zmigrod et al., 2020) and both extreme liberals and conservatives show lower metacognitive awareness (Rollwage et al., 2018). ...
Full-text available
During the 2020 U.S. Presidential primary season, we measured candidate support and cognitive and interpersonal variables associated with political ideology among 831 U.S. participants. Cognitive style variables included openness to experience, active open-minded thinking, dogmatism, and preference for one right answer. Interpersonal variables were compassion and empathy. We modeled candidate support across the political spectrum, ranging from the most conservative to the most liberal (Trump, Bloomberg, Biden, Warren, Sanders), testing competing pre-registered predictions informed by the symmetry and asymmetry perspectives on political ideology. Specifically, we tested whether mean levels on the variables of interest across candidate supporters conformed to patterns consistent with symmetry (i.e., a curvilinear pattern with supporters of relatively extreme candidates being similar to each other relative to supporters of moderate candidates) vs. asymmetry (e.g., linear differences across supporters of liberal vs. conservative candidates). Results broadly supported the asymmetry perspective: Supporters of liberal candidates were generally lower on cognitive rigidity and higher on interpersonal warmth than supporters of conservative candidates. Results and implications are discussed.
... In political science, memorization and recollection of facts are frequent predictors of preference, issue support, and sophistication (Druckman & Lupia, 2000). Further, the premise that different attitudes reflect differences in structured thinking (analogical mental models) is consistent with research in public opinion (Delli Carpini & Keeter, 1996;Goren, 2004;Luskin, 2002;Sniderman et al., 1993;Zaller, 1992), political behavior (Arceneaux & Vander Wielen, 2017), and ideology (Carney et al., 2008;Conway et al., 2016;Dodd et al., 2012;Eidelman et al., 2012;Hibbing et al., 2014;Hinze et al., 1997;Jost et al., 2003;Jost et al., 1999;Kanai et al., 2011;Schreiber et al., 2013;Talhelm et al., 2015;Tetlock, 1984) demonstrating a correlation between how people think and what people think. ...
A bstract We test a method for applying a network-based approach to the study of political attitudes. We use cognitive-affective mapping, an approach that visually represents attitudes as networks of concepts that an individual associates with a given issue. Using a software tool called Valence, we asked a sample of Canadians ( n = 111) to draw a cognitive-affective map (CAM) of their views on the carbon tax. We treat these networks as a series of undirected graphs and examine the extent to which support for the tax can be predicted based on each graph’s emotional and structural properties. We find evidence that the emotional but not the structural properties significantly predict individuals’ attitudes toward the carbon tax. We also find associations between CAMs’ structural properties (density and centrality) and several measures of political interest. Our results provide preliminary evidence for the efficacy of CAMs as a tool for studying political attitudes. The study data are available at
... Evidence of authoritarian support/ideology in the left-wing, however, has begun to emerge. Conway III et al. (2018) developed the Left-Wing Authoritarianism (LWA) scale, mirroring the RWA scale by Altemeyer (1998) to test the authoritarian symmetry hypothesisthe idea that the same processes which lead to authoritarian personality attributes in right-wing individuals also exist in left-wing individuals (Conway III et al., 2016;Crawford, 2012;Suedfeld et al., 1994). The authors found support for authoritarian symmetry; their LWA measure was positively associated with prejudice, dogmatism, attitude strength, and liberalism, demonstrating the viability of the LWA construct in American samples. ...
Many are concerned that authoritarianism is increasing across the political spectrum. In the current study, we investigated the extent to which dark personality variables (i.e., psychopathy, narcissism, Machiavellianism) predict both right- and left-wing authoritarianism (RWA and LWA) between Republicans and Democrats. We developed profiles of individuals' political and personality characteristics concomitant with endorsement of authoritarianism. Our findings (n = 527) suggest a complex interaction between dark traits and political views. Using latent profile analysis, we uncovered underlying profiles characterizing distinct groups of individuals across party identification, LWA, RWA, and dark traits. Four latent profiles emerged: (1) a typical Democrat, low in dark traits and higher in LWA; (2) a typical Republican, low in dark traits and higher in RWA; (3) a dark Democrat, high in dark traits and high in both RWA and LWA; (4) a dark Republican, high in dark traits and high in both RWA and LWA. Together, these data suggest that authoritarianism manifests differentially across the political spectrum and is influenced by emotional style.
Misinformation related to COVID-19 is a threat to public health. The present study examined the potential for deliberative cognitive styles such as actively open-minded thinking and need for evidence in deterring belief in misinformation and promoting belief in true information related to COVID-19. In addition, regarding how responses to the pandemic have been politicized, the role of political orientation and motivated reasoning were also examined. We conducted a survey in South Korea (N = 1466) during May 2020. Participants answered measures related to demographics, open-minded thinking, need for evidence, and accuracy perceptions of COVID-19 misinformation and true information items. Multi-level analyses of the survey data found that while motivated reasoning was present, deliberative cognitive styles (actively open-minded thinking and need for evidence) decreased belief in misinformation without intensifying motivated reasoning tendencies. Findings also showed a political asymmetry where conservatives detected COVID-19 misinformation at a lesser rate. Overall, results suggest that health communication related to COVID-19 misinformation should pay attention to conservative populations. Results also imply that interventions that activate deliberative cognitive styles hold promise in reducing belief in COVID-19 misinformation.
Democracy was forged in the furnaces of oppression, whether combatting tyranny or affirming the rights of the individual. As democracy is under threat in many parts of the world, there has never been a more urgent need to understand political thoughts and behaviours. This lucid and accessible book brings together a global group of scholars from psychology, political science, communication, sociology, education and psychiatry. The book's structure, based on Abraham Lincoln's well-known phrase 'Of, by and for' the people, scrutinises the psychological factors experienced by politicians as representatives 'of' the electorate, the political institutions and systems devised 'by' those we elect, and the societies that influence the context 'for' us as citizens. From trust to risk, from political values to moral and religious priorities, from the personality and language of leaders to fake news and anti-democratic forces, this book provides vital new insights for researchers, politicians and citizens alike.
Social categorization, the process of mentally placing others into a group, is a universal aspect of daily life. Researchers have long been interested in understanding the consequences of social categorization and have more recently turned their attention to determining the processes of how people categorize others into social groups. In this chapter, I present the efficient categorization framework (ECF), which integrates research in social cognition and political psychology to understand the role of a perceiver's political ideology (i.e., whether a person is more liberal or conservative) in social categorization processes. The ECF proposes that political conservatives prioritize efficient categorization—expending few cognitive resources to make a correct judgment—more so than do liberals. Drawing from this framework, I review evidence indicating that liberals and conservatives diverge in their beliefs about which strategies contribute to accurate social category judgments, as well as how they process available cues during social categorization. I also outline findings that highlight how ideological differences in the social categorization process contribute to evaluations, policy attitudes, and political behaviors. I discuss how the ECF gives novel insight into variability in social categorization processes and offers unique perspective into why liberals and conservatives commonly fail to see “eye-to-eye” in their perceptions of the world.
Full-text available
Individuals operating at complex and at simple levels of conceptual structure played a tactical game for three ½-hr. periods. There was a negative relationship between information input and subsequent information search. Conceptually simple Ss, while generally requesting more information, wanted feedback about ongoing events; complex Ss requested information about new aspects of the game.
Full-text available
And the war began, that is, an event took place opposed to human reason and all human nature. (Tolstoy, cited in Huberman & Huberman, 1964, p. 391.) From the dawn of history down to the sinking of the Terris Bay, the world ech-oes with the praise of righteous war…I am almost tempted to reply to the Pacifist as Johnson replied to Goldsmith, "Nay Sir, if you will not take the universal opin-ion of mankind, I have no more to say. (C. S. Lewis, 1949, pp. 64–65.) As suggested by comparing the above reflections, a striking duality about war is that it is at once both seemingly aversive to humans and yet nearly universally accepted and practiced. Virtually all humans would agree that war is, if not inherently bad, at least highly disagreeable and the cause of much suffering. Indeed, the act of killing another human being—even in war—may 1 During the writing of this paper, Peter Suedfeld was a Visiting Scholar at the Mershon Center, The Ohio State University.
Items were generated to explore the factorial structure of a construct of fundamentalism worded appropriately for Judaism, Christianity, and Islam. Results suggested three underlying dimensions: (a) External versus Internal Authority, (b) Fixed versus Malleable Religion, and (c) Worldly Rejection versus Worldly Affirmation. The three dimensions indicate that religious fundamentalism is a personal orientation that asserts a supra-human locus of moral authority, context unbound truth, and the appreciation of the sacred over the worldly components of experience. The 15-item, 3-dimension solution was evaluated across Mexican (n = 455) and American (n = 449) samples. Fit indexes point out the viability of the new inventory across these two samples henceforward referred to as the Multi-Dimensional Fundamentalism Inventory (MDFI). Additional validity tests supported that the new inventory was negatively correlated with participants' integrative complexity in a religious domain-specific way.
Previous research on democratic political leaders has revealed systematic relationships between the content (ideological orientation) and structure (integratively complexity) of political thought. Conservatives tend to be less integratively complex than liberals and moderate socialists-although this effect is qualified by the existence of role-by-ideology and issue-by-ideology interactions. The present study explores the relationship between ideology and integrative complexity in a very different political and cultural context: the Soviet Union in the mid-1980s. Systematic coding of politcy statements of key Soviet leaders revealed a number of effects: (a) Communist Party leaders classified as reformers (pro-Gorbachev) had more integratively complex styles of policy reasoning than traditionalists; (b) this difference was significant in both time periods examined (the last six months of Chernenko's tenure in office and the first six months of Gorbachev's) and was much more pronounced in the Gorbachev period. The article concludes by noting parallels between the data on Soviet and Western political leaders and by considering alternative cognitive, impression management, and institutional explanations for these data.
Individuals and groups can be viewed as information-processing systems which respond in a curvilinear fashion to three components of input load: complexity of information, noxity (unpleasantness) and eucity (pleasantness). An optimal input load is postulated, at which each system is expected to achieve maximum complexity in information-processing. At similar input levels, some systems are expected to show more complex information-processing than other systems. Research is reviewed which suggests that the model holds for perception, information search, decision-making, and innovation. When productivity criteria are associated with complex information-processing, the model predicts productivity. A more complex phasic theory is then advanced, which argues that perceptual and decision-making junctions are separate and not synchronous.
Research suggests that the integrative complexity of political rhetoric tends to drop during election season, but little research to date directly addresses if this drop in complexity serves to increase or decrease electoral success. The two present studies help fill this gap. Study 1 demonstrates that, during the Democratic Party primary debates in 2003–2004, the eventual winners of the party nomination showed a steeper drop in integrative complexity as the election season progressed than nonwinning candidates. Study 2 presents laboratory evidence from the most recent presidential campaign demonstrating that, while the complexity of Obama's rhetoric had little impact on college students' subsequent intentions to vote for him, the complexity of McCain's rhetoric was significantly positively correlated with their likelihood of voting for him. Taken together, this research is inconsistent with an unqualified simple is effective view of the complexity-success relationship. Rather, it is more consistent with a compensatory view: Effective use of complexity (or simplicity) may compensate for perceived weaknesses. Thus, appropriately timed shifts in complexity levels, and/or violations of negative expectations relevant to complexity, may be an effective means of winning elections. Surprisingly, mere simplicity as such seems largely ineffective.
Integrative complexity is a conceptually unique and very popular measurement of the complexity of human thought. We believe, however, that it is currently being underutilized because it takes quite a bit of time to score. More time-efficient computer-based measurements of complexity that are currently available are correlated with integrative complexity at fairly low levels. To help fill in this gap, we developed a novel automated integrative complexity system designed specifically from the integrative complexity theoretical framework. This new automated IC system achieved an alpha of .72 on the standard integrative complexity coding test. In addition, across nine datasets covering over 1,300 paragraphs, this new automated system consistently showed modest relationships with human-scored integrative complexity (average alpha = .62; average r = .46). Further analyses revealed that this relationship consistently remained significant when controlling for superficial markers of complexity and that the new system accounted for both the differentiation and integration components of integrative complexity. Although the overlap between the automated and human-scored systems is only modest (and thus suggests the continued usefulness of human scoring), it nonetheless provides the best automated integrative complexity measurement to date.