University of Southern California Dissertations and Theses

How "reducing" affective polarization increases politically motivated reasoning

(USC Thesis Other)

How "reducing" affective polarization increases politically motivated reasoning

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content RUNNING HEAD: How “Reducing” Affective Polarization Increases Politically Motivated
Reasoning

HOW “REDUCING” AFFECTIVE POLARIZATION INCREASES POLITICALLY
MOTIVATED REASONING

by

Paul L. Sparks

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(COMMUNICATION)

May 2023

Copyright 2023 Paul L. Sparks

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

ii
Dedication

To Brennen Lee Mulligan, Aabria Iyengar, Brian Murphy, and Keith Ayala, without whom I
would have found little respite or community these last several years. To all my fellow
adventurers, Julia, Sterling, Emily, Marv, Phil, Tim, Nathan, Becky, Mark and Matt, who gave
me the support I needed, often by doing the unexpected. And finally, to Tiger, my constant
companion and most trusted confidant. Thank you.

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

iii

Table of Contents
Dedication ....................................................................................................................................... ii
List of Tables .................................................................................................................................. v
List of Figures ................................................................................................................................ vi
Abstract ......................................................................................................................................... vii
Introduction ..................................................................................................................................... 1
Chapter 1 – The Origins of Affective Polarization ......................................................................... 5
What is causing affective polarization? ...................................................................................... 7
Issue-Based Polarization ......................................................................................................... 7
Selective Exposure ................................................................................................................ 10
Ideological Sorting ................................................................................................................ 13
Social Identity Sorting .......................................................................................................... 18
Chapter 2 – Identity-Based Motivated Reasoning ........................................................................ 24
Believing the Lie ................................................................................................................... 24
Brief Review of Confirmation Bias ...................................................................................... 25
Dual Process Models and Social Reward ................................................................................. 27
Identity Protective Cognition ................................................................................................ 31
Chapter 3 – “Reducing” Affective Polarization............................................................................ 36
Depolarization ....................................................................................................................... 38
Using an Economic Game to “Reduce” Affective Polarization ........................................... 39
Participants and Survey Iterations ........................................................................................ 44
Piloting .................................................................................................................................. 46
Procedure .............................................................................................................................. 52
Experiment 1: Party Endorsement and Policy Support............................................................. 55
Experiment 2: Support for Undemocratic Practices ................................................................. 62
Experiment 3: Identity-Based Motivated Learning .................................................................. 66
Measures ................................................................................................................................... 71
Manipulation Checks ............................................................................................................ 71
Covariates ............................................................................................................................. 72
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

iv
Dependent Variables ............................................................................................................. 73
Results ....................................................................................................................................... 73
Sample Characteristics .......................................................................................................... 73
Manipulation Checks ............................................................................................................ 75
Experiment 1 Results: Support For Legislation .................................................................... 76
Experiment 2 Results: Support for Undemocratic Practices ................................................. 83
Experiment 3 Results: Identity-Based Motivated Learning .................................................. 85
Chapter 4: Discussion ................................................................................................................... 91
Lessons Learned.................................................................................................................... 91
What is the effect of Affective Polarization? ........................................................................ 95
Limitations and Future Research ........................................................................................ 101
Thank you, David Broockman ............................................................................................ 105
References ................................................................................................................................... 107
Appendix: Tables ........................................................................................................................ 118

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

v
List of Tables
Table 1: Survey 2 Sample Demographics ................................................................................... 119
Table 2: Survey 1 Sample Demographics ................................................................................... 120
Table 3: Partisan Demographics (Survey 2) ............................................................................... 121
Table 4: Great American Outdoors Act (Experiment 1) ............................................................. 122
Table 5: Gun Violence Prevention Act (Experiment 1).............................................................. 123
Table 6: Support for Undemocratic Practices (Experiment 2).................................................... 124
Table 7: Proportion of Correct Responses and Contrasts for Democrats ................................... 125

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

vi
List of Figures
Figure 1: Trends in Affective Polarization (American National Elections Studies, 2021) ............ 6
Figure 2: Issue Polarization Over Time .......................................................................................... 9
Figure 3: Partisan-Ideological Sorting Over Time........................................................................ 14
Figure 4: Extreme Issue Positions................................................................................................. 18
Figure 5: Belief by Ideology Strength (Kahan, Jenkins-Smith et al., 2011) ................................. 33
Figure 6: Party Perceptions of the Relationship Between Illegal Immigration and Crime .......... 66
Figure 7: Belief that Illegal Immigration Causes Crime by Political Identity Strength ............... 67
Figure 8: Refugee Housing and Crime Covariance Detection Task ............................................. 69
Figure 9: Support for the Great American Outdoors Act by Party and Treatment ....................... 80
Figure 10: Support for the Gun Violence Prevention Act by Party and Treatment ...................... 82
Figure 11: Support for Undemocratic Practices by Party and Treatment ..................................... 85
Figure 12: Republicans on Refugee Housing and Crime Covariance Task.................................. 87
Figure 13: Democrats on Refugee Housing and Crime Covariance Task .................................... 89

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

vii

Abstract

Democrats and Republicans dislike and distrust each other now more than ever — a trend called
affective polarization. While cause for concern in its own right, relatively little is known about
how affective polarization influences political decision-making. Recently Broockman, Kalla and
Westwood (2022) conducted a series of experiments designed to manipulate levels of affective
polarization independent of other factors in order to observe effects on a broad array of political
opinion and intention measures. While levels of out-party animus appeared to be reduced through
their intervention, this reduction had no impact on the variables of interest, leading the authors to
claim that affective polarization has effects exclusively in the interpersonal domain and does not
influence political attitudes. This dissertation critically examines the Broockman et al. (2022)
study, pointing out theoretical and methodological issues and providing alternative hypotheses for
the null findings. The general consensus among depolarization scholars is that affective
polarization has its roots in identity-based motivated reasoning, here referred to as politically
motivated reasoning. By failing to articulate a coherent theory of affective polarization’s causes,
previous research was unable to account for the information processing changes that occurred
when participants had their political identities primed. As such, apparent “reductions” in affective
polarization occurred in measurement while information processing changes from identity priming
produced unexamined effects. In three experiments, this dissertation explicates these effects,
showing that participants who had their levels of affective polarization “reduced” processed
information in a more, not less, partisan manner.
RUNNING HEAD: How “Reducing” Affective Polarization Increases Politically Motivated
Reasoning

Introduction

Over the last several decades, Democrats and Republicans in the general population have
shown a growing tendency to not only disagree across party lines on the substance of policy, but
to express increasing levels of dislike and distrust toward their opposing political party — a trend
commonly known as affective polarization (Iyengar et al., 2012). Some evidence suggests that
these hostile feelings are “ingrained and automatic in voters’ minds” (Iyengar & Westwood,
2015, p. 690), the product not of policy disagreements, but the increasing salience of partisanship
as a social identity (Huddy et al., 2017). In a trend detailed in Chapter 1, party identification has
become a better predictor of several other social identities than it was in the past, with race,
religion, and locality being increasingly associated with party affiliation (Mason, 2016). Several
studies have shown evidence of discrimination across party lines on socially and economically
consequential decisions, such as who to hire for a job, how to distribute resources in an economic
game and several other interpersonal decisions (Iyengar & Westwood, 2015; Iyengar et al., 2019;
Gift & Gift, 2015; McConnell et al., 2018; Carlin & Love, 2013). This evidence is concerning in
its own right, but the fear expressed by many political observers and scholars is that out-party
animus may threaten the continued functioning of democratic systems of governance. For
example, scholars have suggested that affectively polarized partisans might be more willing to
uncritically accept their party’s policy positions (Iyengar et al., 2019); cross-party disdain could
derail legislative bipartisanship (Levendusky, 2018); animus could undermine democratic norms
(Kingzette et al., 2021); even perceptions of objective conditions might be distorted by
partisanship (Iyengar et al., 2012).
1
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

2
Surprisingly, however, only a small number of studies (e.g., Druckman et al., 2020;
Broockman et al., 2022) have attempted to identify possible connections between out-party
animus and political decision-making. While the interpersonal consequences are well
established, there is some debate over the political consequences of so-called “affective
polarization” (Lelkes & Westwood, 2017). The problem scholars must overcome is the inherent
endogeneity between issue positions and out-party animus (Druckman et al., 2021). Given the
seemingly close relationship between issue positions and opinions of the out-party, determining
the impact of affective polarization on political decisions independent of policy attitudes is
methodological challenging. Added to this is the issues that common measures of affective
polarization diverge somewhat on the phenomenon ultimately being captured. For example,
Druckman and Levendusky (2019) compared the most common indicators of affective
polarization and found that measures intended to capture feelings toward “Parties” better
captured attitudes toward party elites, not out-party voters. While many measures of affective
polarization are highly correlated, social distance measures, which emphasize interpersonal
decisions about how to interact with out-partisans, are more weakly correlated with trust and
feeling thermometer scores (Druckman & Levendusky, 2019). As the most commonly accepted
measure of affective polarization, the difference between own and out-party feeling thermometer
scores provides the best available longitudinal measure of the trend, but provide no insight into
the underlying psychological process driving it.
Among the intervention designed to reduce affective polarization, projects that help out-
partisans connect on a personal level show some promise (Volkel et al., 2022). However, we also
see some signs that interventions designed to emulate positive inter-partisan interactions, despite
reducing affective polarization by most measures, do not show an impact on support for
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

3
undemocratic practices, undemocratic candidates, or partisan violence (e.g., Voelkel et al.,
2021). One such experimental intervention, conducted by Broockman et al. (2022), provides the
case upon which the present research is centered.
Broockman, Kalla and Westwood (2022) conducted a series of experiments designed to
manipulate levels of affective polarization exogenously (independent of other factors) in order to
observe effects on a broad array of political opinion and intention measures. Surprisingly, people
who had their levels of out-party animus reduced through a simulated positive economic
interaction with out-party members, expressed opinions no different from participants who had
experienced a discriminatory economic interaction with out-party members. This was true for
measures of electoral accountability, party loyalty, political desensitization, convergence to party
positions, support for bipartisanship, support for democratic norms, and perceptions of objective
conditions. The authors concluded that reducing affective polarization only matters
interpersonally and does not have an impact on political outcomes. The present research
examines this intervention closely an offers an alternative explanation for this null findings.
While the idea that affective polarization has strictly interpersonal implications is
intriguing, the authors’ conclusion that affective polarization does not influence political
attitudes seems to defy common sense and stands in contrast to finding from large-scale multi-
intervention depolarization research by Volkel et al. (2022). That study showed that many
interventions reduce both affective polarization and have an impact on political opinions. In the
present study, three experiments utilizing a modified version of Broockman et al.’s intervention
demonstrate that while affective polarization was reduced as measured by feeling thermometers
scores, this effect may have come with unintended consequences that could explain the null
finding in prior research. Because the cognitive mechanism through which affective polarization
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

4
would influence political decision-making has never been clearly articulated, the assumption that
lower affective polarization could alter information processing is open to interrogation. It will be
shown here that the manipulation used by Broockman et al. (2022) to simulate a positive out-
partisan social interaction (a modified version of the behavioral economic trust game) actually
primed politically motivated reasoning. While this priming effect cannot be demonstrated on all
of the outcomes measured previously, it is clearly present when participants engage in tasks
designed to capture differences between heuristic and systematic information processing.
Though Broockman et al.’s intervention appeared to reduce affective polarization, it also seems
to cause participants to process political information in a more (not less) politically biased
manner.
Chapter 1 summarizes the research on affective polarization, its measurement, origins
and consequences, assisted by analysis of the most recent cumulative American National
Election Studies dataset. Chapter 2 delves into the research on identity-based motivated
reasoning, information processing and the effects of social identity priming. Chapter 3
summarizes current work in the multidisciplinary field of depolarization research, discusses the
theoretical and methodological flaws of the Broockman et al. (2022) study, and presents
evidence from three original survey experiments to show that efforts to reduce affective
polarization can increase politically motivated reasoning.
RUNNING HEAD: How “Reducing” Affective Polarization Increases Politically Motivated
Reasoning

Chapter 1 – The Origins of Affective Polarization

Affective polarization — the tendency for partisans to dislike and distrust members of the
opposing party — is most commonly demonstrated through decades-long trends in out-party
animus, as measured with feeling thermometer scores from American National Election Studies
(ANES) surveys (Iyengar et al., 2019). The ANES has been surveying representative samples of
Americans age 18 and above in at least every presidential election year since 1948. The main
questions used to track affective polarization, “How would you rate the Democratic (Republican)
party?” on a scale of 0 to 100, were first recorded in 1978, though similar measures preceded it.
Figure 1 charts responses to these questions from 1978 to 2020, recoded to show respondents’
feelings toward their own party (green) and out-party (purple) (American National Election
Studies, 2021a).
1
The difference between in-party and out-party scores (the yellow line) is the
most commonly referenced indicator of affective polarization. Now crossing the 50 point
threshold, the 2020 average of 51.09 points (SD = .41) indicates that partisans, as of the last
election, disliked the opposing party more than they like their own.
Narrowing in on the period from 2000 to 2020, average levels of affective polarization
increased 21.40 points (Robust SE = 1.09, p < .001).
2
In a trend that will become familiar
throughout this analysis, this growth in affective polarization occurred within both parties but
asymmetrically, with Republicans reporting average affective polarization levels 6.89 points
higher than Democrats (Robust SE = 2.19, p = .002).

1
In keeping with standard practice for this type of analysis, independents who lean toward one party or the other are
included, but pure politically neutral independents are not.
2
Analysis in this section utilized linear regression with robust standard errors or mixed effects models with
observations clustered in survey year. Sample weights were applied where available from 2000 to 2020.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

6

Figure 1: Trends in Affective Polarization (American National Elections Studies, 2021)

The remainder of this chapter explores a few of the most common explanations for the
current trend toward affective polarization. Because a great deal of affective polarization
research (e.g., Mason, 2015, 2016, 2018) utilizes data from the ANES, I conducted analysis with
the updated 2020 cumulative ANES dataset to demonstrate trends and replicate findings from
prior research. The explanations for affective polarization discussed here are issue-based
ideological polarization, selective exposure to partisan media, and social identity sorting are all
reviewed as potential causes of affective polarization. These provide a backdrop that will
eventually bring us to identity-based motivated reasoning, which is the emphasis of Chapter 2.

0
20
40
60
80
100
1980 1990 2000 2010 2020
Year of Study
Affective Polarization Out-Party Rating Own-Party Rating
Affective Polarization Over Time
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

7
What is causing affective polarization?
Issue-Based Polarization
To win elections, it is helpful to identify and emphasize the issues that best distinguish
voters from each other. Political parties help organize citizens into ideologically consistent
groups, connect them to their representatives, help them advocate for common interests and
provide useful cognitive heuristics that make it possible for citizens to feel comfortable
supporting a candidate without requiring them to be experts on a wide array of issues (Bolsen et
al., 2014; Taber & Lodge, 2006). The classic definition of polarization refers to the increasing
difference between partisans on the issues, referred to as issue-based ideological polarization
(Iyengar & Westwood, 2015a; Mason, 2018a). Issue-based polarization may have grown for a
number of reasons over the last several years, but the most obvious reason we might see it is one
we do not discuss very often: rationally self-interested citizens assessing political issues and
gravitating toward the parties that best represent their attitudes (see Lelkes, 2018; Webster &
Abramowitz, 2017). As parties come to take on a more consistent set of issue positions,
cleavages between the parties reflect real differences in the policy preferences and values of
constituents, thus making polarization a sign of more, not less, representative government.
Unsurprisingly, the trend in affective polarization is correlated with the trend in issue-
based polarization. Replicating and extending the work of Mason (2015), I constructed an issue
position index based on responses to the (only) five policy issues on which respondents have
been consistently surveyed from 1982 to 2020 in the cumulative ANES dataset (n = 42,462).
These include support for abortion rights, government spending on social services, government
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

8
backed health insurance, aid to Black Americans,
3
and government support for jobs and a good
standard of living.
4
Figure 2 charts the mean difference between Democrats and Republicans on
the issue index. For comparison, a version of the affective polarization trend line is included as
well, rescaled to run from 0 to 1.
5
From 1982 to 2020, the absolute difference between
Democrats and Republicans on the issue index increased 25 percentage points (F(3, 8486) =
2451.94, R
2
= .46, 95% CI[.23, .27], p < .001), compared to under 12 percentage points increased
in affective polarization (F(1, 8507) = 538.67, R
2
= .06, 95% CI[.11, .13], p < .001), showing
that both types of polarization have grown over this period of time, but also suggesting the
possibility that issue polarization has out-paced affective polarization.

3
The language of this item in the ANES is “Some people feel that the government in Washington should make
every effort to improve the social and economic position of blacks. Suppose these people are at one end of a scale, at
point 1. Others feel that the government should not make any special effort to help blacks because they should help
themselves. Suppose these people are at the other end, at point 7.” Respondents have been asked this question in this
language since 1970 and it has not been updated despite the terminology being inappropriate by todays standards.
4
With the updated ANES 2020 data, some measures selected by Mason have changed. There is no longer an item
tracking support for defense spending, which was included in her original index. The present version of ideology
index includes the following items: (1) when should abortion be allowed by law (4-point scale from never permitted
to never forbidden); (2) government services-spending (7-point scale from government should provide many fewer
services: reduce spending a lot to government should provide many more services: increase spending a lot; (3)
government health insurance (7-point scale from government insurance plan to private insurance plan); (4) aid to
Black Americans (7-point scale from government should help minority groups/blacks to minority groups/blacks
should help themselves); (5) guaranteed jobs and income (7-point scale from government see to job and good
standard of living to government let each person get ahead on his own). All items were recoded to run from 0 to 1.
5
The mean of the raw affective polarization measure includes negative numbers, indicating higher rating for the out-
party than one’s own party. When normed to run from 0 to 1, negative values are no longer included in the average,
which is why an average of 50 points, for example, with the raw score does not correspond to an average of .5 with
the transformed measure.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

9

Figure 2: Issue Polarization Over Time
If the alignment of parties with issues is driving issue polarization, and issue polarization
is driving affective polarization, we should expect to see more rapid polarization of both kinds as
more information about party positions becomes more widely accessible. In other words, as
rationally self-interested citizens gain more information about party positions, they should be
more strongly drawn to the party that represents their interests. Lelkes, Sood and Iyengar (2017)
explored this link by examining the differences in polarization brought on by the expansion of
broadband internet access. They exploited the clever instrument of so-called “Right of Way”
laws which were designed to support the expansion of internet access by reducing taxes for
providers. With increased access to broadband internet in 2004 and 2008 polarization also
increased, not simply as a function of access to information, but, the authors opine, specifically
due to increased consumption of partisan media. Comparing homes that had broadband internet
to those that had dial-up, the authors stated that, “people with broadband access consume a lot
0
.1
.2
.3
.4
.5
.6
.7
.8
.9
1
1982 1986 1990 1994 1998 2002 2006 2010 2014 2018
Election Year
Issue Polarization Affective Polarization
Note. Mean difference on issue positions includes estimates created with incomplete data in 2012 and 1998, which
may effect the accuracy of averages for those years.
Issue Polarization and Affective Polarization
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

10
more partisan media” (Lelkes et al., 2017, p.16, emphasis mine). Utilizing the 2012 cumulative
ANES, Lelkes et al. (2017) also showed that, over the same period of time, members of both
parties reported consuming more partisan media in households with broadband relative to dial-up
internet. Broadband internet access increased political knowledge broadly, but it did so with a
bias toward partisan media and news. This might seem obvious, though it was not always, but
citizens do not simply select political parties based on the issues, political parties, through
political media, influence the issue positions taken by their constituents (Ditto, Liu, et al., 2019).
Selective Exposure
The firehose of information coming through the internet makes it necessary for citizens to
attend selectively to the media they consume. Selective exposure, also referred to as the selection
aspect of confirmation bias (Taber & Lodge, 2006), is an old concept in the social sciences that
has garnered renewed interest in the information age. Humans tend to seek out information that
confirms their prior beliefs and attitudes. Selective exposure to pro-attitudinal media is
associated with increases in polarization, but polarization also contributes to selective exposure
by strengthening the attitudes that brought about selection in the first place. Steady pressure to
reject counter-attitudinal political information in preference for pro-attitudinal information drives
the mass public into ever more self-fulfilling information silos (Stroud, 2010). Our online social
networks, themselves self-selected and homophilous, exacerbate this by creating environments
that are ripe for information cascades and the preferential spread of false, but engaging news
(Centola, 2010; Vosoughi et al., 2018). Humans have a very well documented tendency to view
information as more valid when it comes from sources identified as political or ideological
affiliates, whether they be political elites or friends on Facebook, which makes countering
misinformation derived through this process challenging for members of the opposing party
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

11
(Ditto, Liu, et al., 2019; Knobloch-Westerwick et al., 2020; Knobloch-Westerwick & Lavis,
2017; Pennycook & Rand, 2019; Stroud, 2010; Stroud & Muddiman, 2013; Taber & Lodge,
2006).
Here, briefly, it is important to contemplate a potential asymmetry between the two
parties. Garrette and Stroud (2014) found that, while individuals across the political spectrum are
more receptive to pro-attitudinal news, Republicans are significantly more likely to avoid
counter-attitudinal information and also more likely to select pro-attitudinal information than
those of other party affiliations (Garrett & Stroud, 2014). Affective polarization may act as a
mediator of misperceptions, especially about Democratic candidates for Republicans who
consume conservative media, but there is not always an observable equivalent effect for
Democrats (Garrett et al., 2019). Ditto and colleagues (2019) conducted a 51 study meta-analysis
and found overwhelming evidence for the “tendency for participants to find otherwise identical
information more valid and compelling when it confirmed rather than challenge their political
beliefs” (Ditto et al., 2019, p. 282), but also made the strong claim that this bias was bipartisan.
The latter point was immediately challenged by Baron and Jost (2019) who criticized several
methodological flaws in Ditto et al.’s symmetry analysis and pointed out, correctly, that
conservatives score consistently lower than liberals on measures of “integrative complexity,
cognitive reflection, need for cognition, and uncertainty tolerance” (p. 293) and higher on
motivated social cognition and need for epistemic closure (Baron & Jost, 2019). These
psychological characteristics have their own robust support and indicate substantial differences
in the way partisans attend to information.
Several lab-based experiments have shown that partisan selective media exposure is
associated with polarized attitudes and bias in the evaluation of information, but to see the real
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

12
world influence of selective exposure, we would need a longitudinal study in which pro-
attitudinal media consumption habits were replaced with an alternative. Near the end of the 2020
campaign season, David Broockman and Joshua Kalla (in pre-print), randomly assigned 304
Republican participants to watch CNN (a network that was critical of President Trump) instead
of Fox News (a network that promoted Trump’s agenda) for an average of 5.8 hours per week,
and paid them $15 an hour to do so (while also tracking a control group for comparison). They
tracked actual viewing habits through the television provider and conducted surveys every week
from late August to late September 2020. Their findings, though not yet published, provide
support for effects from selective exposure. Treatment group participants learned more favorable
information about Democrats and less favorable information about Republicans. They became
less likely to believe that “If Joe Biden is elected President, we’ll see many police get shot by
Black Lives Matter activists,” which was apparently something sufficiently believed by the
control group to show a significant difference. Those watching CNN became significantly more
skeptical of Fox News, less likely to agree that “If Donald Trump did something bad, Fox News
would discuss it.” However, when the $15/hour incentive was removed, participants went back
to watching Fox and their attitudes became polarized once again. Notably, while there were
effects on knowledge, this experiment had no impact on affective polarization. It only
temporarily shifted political attitudes among a peculiar group of people who were willing to be
paid to pay close attention to a news source they did not trust. The immediate return to Fox
viewership, after their skepticism of its coverage was significantly increased, demonstrates that
these partisans were driven by a motivation other than accuracy in their information selection
process. Selective exposure is part of what’s driving polarization, but something deeper is going
on.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

13
One of the key debates among scholars in this literature is the question of whether
exposure to partisan media is a cause or a symptom. That is, do people seek out news that
confirms their prior beliefs or does partisan news generate these beliefs to begin with? Some
scholars have even argued that exposure to political media, congenial or crosscutting, has no
significant effect on affective polarization at all (Wojcieszak et al., 2021). Whether online or off,
a fairly limited amount of an average citizen’s time is spent consuming news media. Belief in
misinformation, one of the most commonly cited concerns about how partisan media could harm
democracy, may be better explained by laziness than bias (Pennycook and Rand, 2019). The
more an individual is willing to think analytically, regardless of their political ideology, the more
accurately they judge both real and fake news.
Ideological Sorting
Members of the two dominant political parties have increasingly diverged in their views
on the issues over time, but issue positions are actually a relatively uncommon way to measure
political ideology. In the ANES, and commonly in other research, the primary ideology question
asks survey respondents to place themselves on a single-item 7-point political ideology scale
ranging from 1 “Extremely Liberal” to 4 “Moderate, middle of the road,” to 7 “Extremely
Conservative” with two gradations between each of these points. Over the last few decades, this
measure of ideological self-identification has become an increasingly strong predictor of political
party identification, in a process known as ideological sorting (Mason, 2018a; Weber & Klar,
2019). Figure 3 plots the expected probability of identifying as a liberal Democrat or
conservative Republican in each Presidential election year since 1982 based on a mixed effects
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

14
linear regression model controlling for several factors
6
with observations clustered within years
(F(40, 29451) = 137, p < .001, R
2
= 16).

Figure 3: Partisan-Ideological Sorting Over Time

As shown previously, polarization based on issue positions has increased steadily and,
many scholars argue, asymmetrically since the 1970s, with Republicans taking ever more
extreme policy positions relative to Democrats over time (Hacker and Pierson, 2005). However,
ideological sorting has remained fairly constant for Republicans, while it has varied substantially
for Democrats with each passing administration according to the cumulative ANES data. The
line in Figure 3 for Liberal-Democrats follows an interesting pattern. Only 55% of Democrats
identified as liberal (somewhat, liberal, or extremely) at the end of the Reagan administration in

6
Control variables include age, gender (male / female), race (white or not), southern (or not), Christian evangelical
(or not), and frequency of church attendance.
.4
.5
.6
.7
.8
.9
1
Linear Prediction
1980 1984 1988 1992 1996 2000 2004 2008 2012 2016 2020
Election Year
Conservative Republican Liberal Democrat
Note. N = 29,451. Proportion of Democrats identifying as Liberal and Republicans identifying as conservative by
election year. Estimates are marginal predictions from a mixed effects model controlling for education, age, gender,
race (white), southern background, evangelicalism, and church attendance.
Liberal Democrats and Conservative Republicans
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

15
1988. Liberal identification ticked up slightly in the George H. W. Bush years until Clinton’s
election in 1992, then corrected in the midterms, but shot back up in 1998 (midterm estimates are
not shown in Figure 3). With the election of George W. Bush in 2000 and the terrorist attacks on
9/11, being a Liberal-Democrat became less popular, slumping back down until the 2008
election of Barack Obama. From 2008 to 2020 Democrats’ identification as liberal increased
rapidly and to its highest point since tracking began. While the campaign messages of hope and
change probably had a something to do with this, in this same time period Americans also saw
the rise of the far right-wing Tea Party movement and the eventual election of Donald Trump.
Social identities are defined by two functions: (1) they identify the members of an in-
group; (2) they identify the members of an out-group (Oyserman, 2015). When Republican elites
stoked the anger, fear and enthusiasm of their most radical constituents, they energized their base
to come out and vote, but by leveraging racial prejudice and extremist ideology to do so, they
also propagated highly threatening rhetoric that energized their opposition as well (Mason et al.,
2021). In the short run, their strategy might have paid off, but it also gave Democrats something
to rally against. This led to the situation we found ourselves in 2020, when ideological alignment
with the liberal label among Democrats reached its highest level on record.
Once again following the approach taken by Mason (2015), I recoded the issue index to
indicate issue extremity rather than the relative average position in each party (n = 11,533). Each
item was center folded, such that more moderate positions were scored lower and positions at
either end of the scale were scored higher. For example, if a respondent indicated either that they
believed abortion should be legal in all circumstances or abortion should never be legal, the
response was coded at the highest score for that item, 1 on a scale from 0 to 1. In this analysis,
instead of looking at the average difference between Democrats and Republicans on the available
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

16
issues, plotted points indicate the average issue extremity by party identification while
controlling for ideological sorting and demographic differences, based on linear regression
analysis with observations clustered within years from 1984 to 2020 (n = 11,533, Waldχ
2
(28) =
777.47, p < .001). Figure 4 shows the beginnings of a potentially major departure from previous
trends. For the first time on record, in 2020, Democrats took more extreme positions on the
issues than Republicans. Based on separate regression analysis restricting the data set to the
period from 2016 to 2020 (n = 5,024, F(10, 5013) = 22.04, p < .001), relative to Republicans
who moderated their issue position extremity slightly (b = -.04, 95% CI[-.06, -.02], p < .001)
Democrats expressed 12 point more extreme issue position (95% CI[.10, .15] p < .001).
Considering the events leading up to the 2020 election, the pandemic, the green new deal,
protests against police brutality, growing calls to defund the police, and growth in the movement
for black lives, it makes some sense that this would be the moment in history when Democratic
political extremity would eclipse that of Republicans. However, this trend should be interpreted
with caution, keeping in mind the index includes only the five issue items that have been tracked
for the longest period of time and not the specific set of issues that might be most relevant at this
moment. Nevertheless, these shifts on the Democratic side bring up an important question. Going
forward, will we continue to see ever increasing political extremism, or, as the impressive
showing by Democrats in the 2022 mid-term elections might indicate, are these trends about to
abate?
One note of interest here is that position extremity topped out for Republicans in 2012,
the year of President Obama’s re-election. This followed the passage of the Affordable Care Act
and the zenith of the Tea Party movement’s influence. While ostensibly a group of grassroots
activists who rallied behind decreasing taxes, reducing federal debt, and opposition to healthcare
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

17
reform, the Tea Party was in reality a so-called “Astroturf” operation, created and funded largely
by billionaire activists David and Charles Koch (Monbiot, 2010). While no causal claim can be
made based on the evidence presented here, the increase in political extremism among
Republicans in 2012 may reflect the power of elites manipulate of information in order to garner
support for causes that go against the interests of the very Americans who are mobilized. Among
the issues included in this index is support for government-back health care, which reached its
lowest level of support among Republicans in 2012, predicted M = .28, SE = .002, on an item
ranging from 0 to 1. At that time, the Koch brothers were funding right-wing think tanks and
projects that channeled misinformation into the Sinclair broadcasting group, the owner of the
largest number of local broadcast television stations in the U.S. (Graves, 2017). Nothing about
these stations would give the impression that they were explicitly partisan. The prevalence of
extreme positions, driven at least partially by elite control over a broad swath of television
stations, lends credence to the claim that polarization is heavily influenced by both overt and
covert partisan news media, which pushes out news stories specifically designed to exacerbate
out-group animus in order to galvanize partisans behind policy positions that primarily benefit
said elites.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

18

Figure 4: Extreme Issue Positions
Anger and enthusiasm, especially in response to perceived out-group threat, are powerful
motivators of political engagement (Mason, 2015). Mason (2015) used a matched sample of
respondents from the ANES to show that, other factors held constant, being ideologically sorted
predicted greater affective polarization, anger directed at the out-party candidate, and political
activism. Rather than the issues driving the strength of political and ideological identification,
partisanship and ideology as an identity is what scholars increasingly agree drives affective
polarization (Dias & Lelkes, 2021).
Social Identity Sorting
There was a time, not long ago in the history of the United States, when political
divisions were far less prominent than they are now. Americans have always disagreed on the
political and social issues of the day, but political parties were not the primary signifier of those
.45
.5
.55
.6
.65
Linear Prediction
1984 1988 1992 1996 2000 2004 2008 2012 2016 2020
Year of Study
Republican
Democrat
Predictive Margins of party with 95% CIs
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

19
divisions for much of our past. Before these trends began in earnest, a conservative, white,
Republican candidate, and a liberal, white, Democratic candidate might have discussed the issues
of the day without rancor. They might even have come to a unified vision, ushered into clarity by
their shared protestant Christian faith (Mason, 2018d). But it would have been no accident that
they were both white. When Truman sent his civil rights proposals to congress in 1948, or when
the Supreme Court ruled against segregation in 1954, or when the Democratic convention
adopted a pro-civil rights platform in 1960, but most assuredly when Lyndon Johnson, a
Democratic President from a southern state, shepherded in the Civil Rights Act of 1964 and
Southerners, for the first time, voted overwhelmingly for Republican candidate Barry Goldwater,
that is when this started (Mangum, 2013). As with so much of the story of this country, the way
that political parties came to be synonymous with political ideologies, was, inevitably,
anticipated by racial resentment (Westwood & Peterson, 2020).
Starting in the late 1960s, Tajfel and colleagues conducted a series of experiments aimed
at understanding the extent to which conflicts over resources and other interests were necessary
to create out-group discrimination. In a famous experiment, participants were asked to estimate
the number of clustered dots projected on a screen during short time intervals. In the baseline
condition subjects were told that people tend to fall into one of two value neutral groups — over-
estimators and under-estimators, neither being more accurate. After estimating the number of
dots on several slides, participants were informed that they were members of one of these two
groups. In reality, these group labels had been randomly assigned. To ensure that the randomly
assigned groups did not develop a social identity on any other basis, assignment was anonymous,
subjects were isolated in a laboratory and unable to see, hear or interact with each other. When
subjects were then given the option to distribute monetary rewards between in-group (similar dot
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

20
estimator) and out-group (different dot estimator) members, they showed significant favoritism
for their in-group as well as willingness to deprive resources from their out-group. In a follow-up
experiment, participants were asked to express their opinions on a series of works of art and then
randomly assigned to conditions in which they were informed that they preferred either the artist
Klee or Kandinsky. Notably, in the second experiment, subjects were given the option to
maximize the benefit for all subjects regardless of group, but still chose significantly more often
to take a lower payout in order to ensure that their group would “win” in relative terms (Tajfel et
al., 1971).
This line of research came to be known as the minimal group paradigm, used to express
the idea of the minimum distinction required to produce discrimination, and an enormous
number of studies have replicated these findings. Tests under ever more stringent conditions and
in populations where alternative explanations are hard to fathom present a strong case that out-
group bias is deeply rooted in human psychology. For example, Dunham, Baron and Carey
(2011) conducted a series of minimal group experiments with five-year-old children and found
significant effects on attitudes, resource allocation, behavioral attribution, reciprocity and
information processing when participants were randomly assigned to either the “red” or “blue”
group. Notably, the children in these experiments knew that this assignment was random because
it was determined by asking them to pick one of the experimenter’s hands in which the coin
selected assigned their group (Dunham et al., 2011). If such biases can be demonstrated among
children on the basis of the most minimal of group distinctions, what might be the effects of
more stable, deeply held social identities among adults?
While even the most minimal distinction in social identity can induce intergroup conflict
(Tajfel & Turner, 1979), more highly sorted identities are especially prone to anger and
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

21
enthusiasm in response to threats and reassurances (Mason, 2016, 2018a, 2018c). Expanding
beyond ideological sorting, Mason’s more recent work emphasizes the role of social sorting.
This is a process in which our social identities, such as race, religion, region, as well as a large
number of tastes and preferences have become more closely aligned (Mason, 2016). For more
highly sorted individuals, political party has become a “mega” identity (Mason, 2016, p. 18). The
more sorted, the more perceived threats to one social identity become threats to all of one’s
identities. Mason points out that individuals who have more overlapping identities, that is, not
just liberal-Democrat / conservative-Republican, but liberal-black-agnostic-female-Democrat /
conservative-white-evangelical-male-Republican, express the most out-party animus and the
most extreme issue positions (Mason, 2018d). A white, male, conservative, regular church
attender is far more likely to be a Republican today than he was in the 1980s (Mason & Wronski,
2018). More problematically, though, cross-cutting social identities (e.g. conservative-Democrat)
have declined as a proportion of the population, thereby reducing the number of people who
might be willing to compromise. Individuals who embrace more complex representations of their
multiple social identities tend to have greater tolerance for outgroup members (Mason, 2018).
When an ingroup is under threat, those who embrace complexity are able to shift the locus of
their identities to cope. However, individuals who have simpler representations of their ingroup
identities, tend to express high need for closure — motivation to achieve finality and
absoluteness in their judgement — are more likely to be intolerant, angrier and more biased
against out-groups, particularly when those identities are under threat (Roccas & Brewer, 2002).
As our social identities become more aligned, complexity is reduced, which may lead to greater
intolerance of outgroups.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

22
It is important here to note that we perceive these divisions much more strongly than the
true extent of them. One of the consequences of affective polarization is that it causes partisans
to view members of the out-group as more divergent from the in group than they actually are.
Ahler and Sood (2018) asked a representative sample of Americans “to estimate the percentages
of Democrats who are black, atheist or agnostic, union members, and gay, lesbian or bisexual,
and the percentage of Republicans who are evangelical, 65 or older, Southern, and earn over
$250,000 per year” (Ahler & Sood, 2018, p. 966). On average, respondents overestimated the
party composition of these groups by 342%. For example, respondents thought that 31.7% of
Democrats were gay, lesbian, or bisexual, roughly five times the 6.3% who actually are. They
estimated that 38.2% of Republicans make over $250,000 per year, the correct percentage is
2.2% of GOP supporters. More importantly, participants randomly assigned to have these
misperceptions corrected expressed significantly lower levels of preference for partisan social
distance, perceived out-party extremity and out-party animus. These findings show that
perceptions of opposing partisans are strongly influenced by representativeness bias (Kahneman,
2011), or the tendency to view recognizable stereotypes and public figures as representative
prototypes of the whole group.
In an important follow-up to Ahler & Sood (2018), Druckman and colleagues (2022)
showed that Americans misestimate the ideological extremity and political engagement of
opposing partisans. When answering standard affective polarization questions, e.g., social
distance and feeling thermometer measures, partisans rely on these misperceptions.
“Consequently, when scholars, pundits, and journalists use these measures to characterize
affective polarization, they inadvertently reinforce an inaccurate image of extreme difference
between members of the two parties” (Druckman et al., 2022). While it is true that Americans
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

23
have become more socially sorted, it may be as much the perception of sorting as the thing itself
that ultimately drives affective polarization.
While the debate over how strongly sorting predicts affective polarization is an active and
important discussion, researchers tend to agree that the influence of social identity factors on
cognition are at the root of our political divide. Identity-based politically motivated reasoning is
the core issue to tackle if we wish to find common ground. In the next chapter, I explore how
motivated reasoning and identity-protective cognition bias our perceptions of political issues and
reality.

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

24
Chapter 2 – Identity-Based Motivated Reasoning

Tajfel’s minimal group paradigm exposes the human tendency to discriminate on the
basis of the most arbitrary of identity-based group coalitions. These tendencies can be much
more intense for identity groups we are socialized into from an early age. Culture itself likely
originated as a way of encouraging group cooperation for survival (Oyserman & Yan, 2019).
Rejection of an identity-based belief thus conflicts with an evolved need for belonging, as
expulsion from the group historically meant death (Clark et al., 2019). Political identity is an
especially powerful sociocultural identity for the reasons discussed in Chapter 1. Even though
official party membership comes later in life, our political ideologies “have a strong genetic
basis, emerge early in life, and manifest in brain structure” (van Bavel & Pereira, 2018, p. 214).
The value of holding a group-aligned political belief is driven by a need for social belonging,
epistemic closure, status, morality, and many other culturally associated goals. “Value” in this
sense is not necessarily a conscious association, but an automatic weighting of conclusions that
minimize threat in the brain’s decision circuitry. This chapter delves more deeply into research
on the political ramifications of social identity to explore how our identities alter not only the
way we see each other, but the process by which we make decisions and learn political
information.
Believing the Lie
American Democrats and Republicans lack a shared perception of reality. There may be
no better way to demonstrate this than by pointing out the one thing they do tend to agree on:
64% of Americans believe that democracy in the United States is failing. That is 68% of
Democrats and 79% of Republicans, controlling for gender, race, generation and region of the
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

25
country (NPR / Ipsos, January 2022). The reason democracy is failing, we think, is because the
other team is destroying it. Interestingly, however, if you ask Americans to describe the policy
positions driving this, many will be hard pressed to explain them (Fiorina & Levendusky, 2006).
This is in part because, while the absolute difference between Democrats and Republicans on
policy issues and the proportion of the electorate taking more extreme policy positions has
increased (see Chapter 1), Americans still share overlapping opinions across party lines (Dunn &
Cerda, 2022). It is also often argued, especially for Trump supporters, that Americans tend to be
“low information voters” (Fording & Schram, 2017), who sometimes believe misinformation
simply because they are cognitively lazy (Pennycook & Rand, 2019). Blaming our political woes
on laziness, though, tends to under-represent the cognitive burden of making complex decisions.
Our brains evolved to make life and death decisions as efficiently as possible, on the basis of
prior experience and a large number of natural and social cues. Before exploring how politically
motivated reasoning influences decision making, it is helpful to review how motivated reasoning
differs from confirmation bias.
Brief Review of Confirmation Bias
There are multiple ways to think about how our political identities influence our beliefs.
Chapter 1 briefly discusses confirmation bias in terms of selective exposure, which is also
referred to as confirmatory bias. Due to the popularity of this concept across the social sciences
there is some inconsistency in how the concept is defined. Generally speaking, confirmation bias
refers to the tendency for people to favor information that confirms their prior beliefs —
emphasis on prior. Communication research often focuses on confirmation bias as a trigger for
selective exposure, or the tendency to search for information that supports prior beliefs
(Knobloch-Westerwick et al., 2020; Knobloch-Westerwick & Lavis, 2017; Wojcieszak et al.,
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

26
2020). Evidence contradicting prior beliefs is often avoided, while evidence confirming prior
beliefs is consciously or unconsciously sought out. Somewhat distinct from this search bias is the
approach commonly taken by behavioral economists, who view confirmatory bias as a tendency
to “misinterpret ambiguous information as confirming current hypotheses about the world”
(Rabin & Schrag, 1999, emphasis mine). This second type of confirmation bias is closely
associated with anchoring effects, which operate off of an availability heuristic (Tversky &
Kahneman, 1973). The ambiguous evidence is interpreted as conforming to a prior belief
because that is the easiest, least costly, process of interpretation. For example, Rabin & Schrag
(1999) refer to a study by Bruner and Potter (1964) in which subjects were shown blurred
pictures that were gradually brought into focus. The difference between the experimental and
control groups was the initial intensity of blurring. Among those who started with a more
intensely blurred image, less than 25% eventually correctly identified the image, compared to
more than 50% in the less blurred condition, even though the final image quality was the same
for both groups. Subjects in the high blur group had formed opinions based on less information
and then stuck with those beliefs despite new information being presented. For this type of
confirmation bias, the bias occurs not at search or selection, but in the weight given to prior
beliefs. Someone presented with valid information will still fail to correct mistaken perceptions if
their priors are strong enough (Rabin & Schrag, 1999).
The difference between the two types of confirmation bias is subtle but important
because they lead to different interventions and methods of evaluation. If we believe that search
bias is the dominant force driving false beliefs, then it might be possible to correct false beliefs
by controlling the information to which people are exposed. However, if the problem is not with
selection but with the crediting of information, false beliefs will not be corrected merely by
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

27
introducing or restricting information. Indeed correcting misinformation often fails and can even
backfire, leading people to become more entrenched in their false beliefs (Nyhan & Reifler,
2010).
Both types of confirmation bias are related to, but distinct from the key concept of
interest here: motivated reasoning. Motivated reasoning “refers to the tendency of individuals to
unconsciously conform assessments of factual information to some goal collateral to assessing
the truth” (Kahan, 2016, p.2, emphasis original). Motivated reasoning does not operate by
overweighting prior beliefs, nor on the selection of information, but by aligning the interpretation
of new information with a motivation other than accuracy. This means that false beliefs can
persist regardless of the information environment or prior knowledge if individuals are motivated
to interpret new information inaccurately.

Dual Process Models and Social Reward
To understand how social identity influences motivated reasoning, it is helpful to know a
little bit about what our brains are doing when making a decision. To begin with, we need to
conceptualize the brain not as a single entity, but as modular, a system of systems with
competing goals. Some systems within the brain can substitute for others, but such substitution
may come at a cost. The primary functional unit of the brain is the neuron and the fundamental
unit of information in the brain is the action potential (Brocas & Carrillo, 2014). Neurons
communicate with other neurons by “firing” or “spiking” in all-or-nothing electrochemical
events. Somewhat like bits of code, neurons in the decision-making parts of the brain only have
an “on” state and an “off” state. When a particular choice is preferred, more neurons activate.
Whether or not neurons fire in a particular region is a function not only of the neurological
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

28
process being undertaken, but the availability and distribution of the fuel that allows neurons to
fire. Oxygen and glucose, carried through the bloodstream and observable through functional
magnetic resonance imaging (fMRI), along with neurotransmitters, such as dopamine, are the
fuel of the brain, but each system within it varies in its fuel consumption requirements. This
means that resources in the brain are scarce and expensive systems only activate if resources are
available and attention is directed toward their function.
Modularity and resource constraint help to explain many observed biases in decision-
making. Psychologists began modelling cognition as a process involving multiple, often
competing brain systems in the 1970s. These dual-process models have been described using
several metaphors. Some psychologists refer to central versus peripheral routes of information
processing (Petty & Briñol, 2014), others refer to fast and effortless (System 1) versus slow and
effortful (System 2) processing (Kahneman, 2011), still others refer to automatic versus
controlled processing (Schneider & Shiffrin, 1977; Shiffrin & Schneider, 1977). In neuro-
behavioral research, using one type of processing or the other has known associations with
neuronal activation in particular regions of the brain (though it’s considerably more complex
than two systems). The value of a choice, broadly speaking, is represented by neuronal activation
in the ventral medial prefrontal cortex (vmPFC). If a choice is being made carefully and
systematically (controlled, slow, effortful), neuronal activation will also occur in the dorsal
lateral prefrontal cortex (dlPFC) and thereby modulate activation in the vmPFC (Brocas &
Carrillo, 2012). The problem is that the dlPFC is much more resource intensive than the more
ancient systems that project reward value onto the vmPFC. As such, activation of the dlPFC is
neurologically, metabolically and attentionally more demanding (Rangel & Clithero, 2013). To
be clear, while activation of the dlPFC is associated with systematic processing, this does not
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

29
imply that it is associated with better decisions, since we can systematically process our way to
making errors (e.g., Kahan, 2017a). Superior reasoning proficiency is often used to conform
interpretations of evidence to the positions of a cultural group (Bolsen et al., 2014; Kahan, 2016).
Interestingly, our brains appear to use more or less the same “currency” (the firing of
neurons in the vmPFC) to represent the value of a decision regardless of the decision domain.
Whether we are selecting primary stimuli, such as food, or higher order representations of value,
such as money, neurons in the same approximate region of the brain appear to be activated
proportionally to the corresponding value of a choice. For example, when presented with two
options, such as tempting or healthful food items, the brains of dieters who demonstrate self-
control by picking the healthier option show activation in the dlPFC, whereas the brains of
dieters who choose the candy do not (Hare et al., 2009). The idea here is that the dlPFC activates
when we are exerting self-control, modulating the value signal in the vmPFC when we engage in
an effortful cognitive process.
Value signals originate in the reward center of the brain, broadly speaking the striatum.
Activity in the striatum is associated with the expected magnitude and probability of primary
rewards such as food, as well as economic and social expected rewards (Bhanji & Delgado,
2014). Dopamine projections to the dorsal striatum encode a reward prediction error signal when
people make a choice and receive a benefit (Ruff & Fehr, 2014).
7
Because many types of
decisions involve similar neural value computations, the brain uses the same systems in decision-
making when comparing alternatives, regardless of the type of decision being made. This makes

7
Because there is often some confusion about this in the general public, I wish to clarify that dopamine is not
associated with pleasure. Neurologically, there is a difference between wanting and liking. Dopamine is associated
with wanting and plays a major role in reinforcement learning, that is, training the brain to do a thing. This is helpful
in explaining addictive behavior because narcotics such as cocaine have limited pleasurable effects over prolonged
use, but are powerful because they hack the dopaminergic reward (wanting) signal. This helps explain why it is
possible for addicts to hate the substance they are addicted to, find no pleasure in it and still be desperate to consume
it.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

30
sense based on our current understanding of human evolution. Evolution tends to provide “good
enough” solutions by creating biological systems that are sufficiently complex to achieve an
outcome, but not so specialized that accumulating complexity threatens survival. Thus,
combinations of elements are capable of producing equivalent output in multiple domains
(Marder & Goaillard, 2006; Weiss et al., 2012).
Activity in the brain’s reward center has also been observed in individuals who are
making socially influenced judgements (Bhanji & Delgado, 2014). Specifically, activity in the
ventral striatum increases when one’s choice aligns with popular opinion, such as when there is
consensus among others that a financial risk is a good option. Similar activation has been
observed when people think the advice they are receiving is good, when they value an opinion
more strongly because it has been endorsed by an expert, and when they evaluate new
information as agreeing with advised decisions. Likewise, when a reward is shared with a friend
rather than a stranger or computer, striatal activity increases. This is also true in situations where
people are given the option to share an endowment with a charity of their choice (Bhanji &
Delgado, 2014). On an even more basic level, neural activity in the brain’s reward circuitry has
been elicited when participants are presented with approving faces from attractive others (as well
as erotic photos) (Ruff & Fehr, 2014), demonstrating that the apparatus underlying a great deal
of learning and decision-making is innately attuned to the social world.
Even decisions about objective information can be influenced by competing demands on
the brain’s reward and decision value architecture. This is especially the case when information
might be perceived as threatening to an ideological position held by affiliates. The brain attempts
to satisfy cognitive constraints — which maximize “goodness of fit to the data,” or more simply,
the efficient computation of accuracy (Western et al., 2006, p. 1947) — as well as emotional
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

31
constraints — which maximize positive affect while minimizing negative affect. Western et al.
(2006) showed that when presented with information about a political candidate that would lead
to emotionally aversive conclusions, the brains of partisans who arrived at a different conclusion
— i.e., avoided the aversive conclusion — did not show activation in the dlPFC (the value
modulation circuit). Rather, the brains of these individuals showed activation in the vmPFC
(medial orbital PFC), as well as the insula, anterior cingulate cortex, and other systems
associated with the experiences of punishment, pain, and negative affect (Westen et al., 2006). In
other words, the decision value of an accurate assessment was overcome by the value of averting
a socially divergent belief. Rather than controlling the impulse to default to a party aligned
position (with self-control, activation of the dlPFC), participants in this experiment experienced
an aversive reaction to the threatening information and ignored it. To reiterate, this does not
mean that partisans who adopt a party-aligned belief will always process information
heuristically. The way that information is presented and the accessibility of alternative schemata
play an important role. Nevertheless, when identity is implicated in a decision, our brains tend to
favor conclusions that conform to the norms of our social groups.
Identity Protective Cognition
We tend to think of our identities as stable, intuitive and requiring no empirical support.
To some extent we can reflect on our own patterns of behavior and note situational consistencies.
A person might think of themselves as independent, nurturing, contemplative, a habitual
procrastinator or an anxious perfectionist (not mutually exclusive, in my experience). These sorts
of personality traits can be categorized under the umbrella of personal identities and often do
represent a relatively stable, repeatedly observed self-concept. Alternatively, we can think of
ourselves as inhabiting several social identities. Social identities are coalitional and relational in
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

32
nature, ranging from constructed social categories such as gender, racial-ethnic identity, culture,
and religion to temporally narrow identities such as grad student, teenager, tourist. These
identities often feel stable, but are actually highly sensitive to situational cues. The particular
social identity that is salient in a decision-making process can subconsciously alter the weight
given to the value signal projected in to the vmPFC.
As Oyserman (2015) puts it “rather than being stable, which identities come to mind and
what they mean are dynamically constructed in context” (p. 1). At the same time, identities that
are frequently activated build up neurological pathways that become chronically accessible.
Repeated use of a set of neurons strengthens the connections among them creating an easily
traversable electrochemical pathway. This is a literal physical process that takes place in a well-
defined network of brain regions referred to as the “default mode network”
8
(Niemeyer, 2013).
The default mode network is the identity center of our brains and when a particular identity is
brought to mind, the accessibility of attitudes associated with that identity can alter the
interpretation of information in a manner conforming to that identity. Interpretations that are
congruent with identity-aligned attitudes take on a quality of metacognitive fluency, the feeling
that beliefs are natural, familiar and true. That which is unexpected takes on a metacognitive
quality of difficulty, which may cause us to process information more systematically in order to
find reasons to reject, ignore or rationalize disfluent information (Oyserman & Yan, 2019; Petty
& Briñol, 2015).
This type of information processing was illustrated in the Western et al. (2006) study
discussed above and is termed by Dan Kahan and others “identity protective cognition” (Kahan,

8
The default mode network is thought to include the posterior cingulate cortex, anterior medial prefrontal cortex,
posterior inferior parietal love, medial temporal lobe, ventral medial prefrontal cortex, retrosplenial cortex, regions
of the hippocampus, dorsal medial prefrontal cortex, temporal pole, lateral temporal cortex, and temporal parietal
junction (Niemeyer, 2013).
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

33
2017, p. 2). In identity-protective cognition, which is a form of identity-based motivated
reasoning, information is processed in a manner that conforms to identity rather than accuracy
goals. For example, Kahan and colleagues (2011) conducted an experiment in which participants
were assigned to judge the expertise of a scientist with regard to global warming. Democrats and
Republicans were assigned to conditions in which the information provided by the scientist made
global warming sound like it was either high risk or low risk. Perceptions of the scientist’s level
of expertise were associated with the strength of the respondent’s Liberal-Democrat or
Conservative-Republican self-reported identity strength. Recalling from Chapter 1, this measure
is the combination of the standard 7-point ideology scale with a political identity scale, increases
in which are referred to in the affective polarization literature as ideological sorting. Highly
sorted Democrats were more likely to agree that the scientist was an expert when they were
assigned to the high risk condition, but were unlikely to say he was an expert in the low risk
condition (the inverse was true for Republicans). The following graphic provides an excellent
illustration of the study design and effects. It has been republished by Kahan in multiple reports
and articles (Kahan, 2016, 2017b) and is his and his colleagues sole authorship.

Figure 5: Belief by Ideology Strength (Kahan, Jenkins-Smith et al., 2011)

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

34
Importantly, later research by Kahan et al. (2012, 2013) demonstrate that individuals who
have a higher capacity for the processing of complex information (e.g., numeracy, cognitive
reflectivity) tend to direct that capacity toward adopting party-aligned positions. Systematically
processing political information does not make people more likely to process it accurately, rather
it is an indication that the individual is reasoning their way to their preferred answer.
Methodologically, research into the influence of social identity on decision-making
frequently utilizes psychological priming to increase the salience of a social identity and observe
the resulting effect on information processing and decision-making (Cohn & Maréchal, 2016).
The interventions used in these priming experiments can be very subtle. For example, in a study
about the effects of student-athlete status on test performance, participants assigned to have their
athlete identity primed received test booklets with a cover page that required them to indicate
their status as an athlete by writing the sentence, “I am an athlete” (Stone et al., 2012). Unsworth
and Fielding (2013) assigned participants to have their political identities primed by writing out
three words that characterize people from various political parties and then indicating the party
they personally support (Unsworth & Fielding, 2014). That simple priming exercise had a large
and significant impact on the perception that humans contribute to climate change. In an example
that is more relevant to the particular interests of this study, Levendusky (2018), primed
American social identity by having participants write a short paragraph about “what people like
best about America and why they are proud to identify as American.” Over three experiments,
this prime, as well as similar primes designed for the same purpose, were effective at reducing
affective polarization. This example echoes a finding from the earliest experiments in social
identity theory, showing that the priming of a superordinate identity (i.e., national identity over
party identity) can overcome some of the biases of our defaults. However, as argued in the next
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

35
chapter, priming party identity, even in an effort to reduce affective polarization, can cause us to
process information in a more, not less biased manner.

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

36
Chapter 3 – “Reducing” Affective Polarization

While a considerable body of evidence has shown that affective polarization “plays a role
in how much time we spend with our families, where we want to work and shop and whom we
want to date and marry” (Druckman et al., 2021, p. 28) the political ramifications of affective
polarization are surprisingly unclear (Iyengar et al., 2019a). This is due to an issue of
observational equivalence. Individuals who express high levels of out-party animus tend also to
express more extreme political positions, as such it is difficult to determine if affective
polarization causes ideological extremism or if ideology drives animus. Broockman, Kalla, and
Westwood (2022) point out that there has been little discussion given to the theoretical basis for
why affective polarization would effect political judgements. Presumably, other influences, such
as the perceived benefit of a policy, or the economic performance of the party in power, or the
alignment of one’s values with those promoted by a candidate all contribute to how we make
political decisions and it is unclear why affective polarization would have a distinct role in such
processes.
One of the strongest counterpoints to the argument that affective polarization strictly
influences interpersonal decision-making was a study that utilized the unanticipated occurrence
of the COVID-19 pandemic and the coincidence of having a baseline survey in the field prior to
its outbreak. Given that no priors about the pandemic existed before the survey, the baseline
measure of affective polarization allowed researchers to estimate the effects of previously
recorded out-party animus on political opinion formation about the pandemic (Druckman et al.,
2020). The pandemic forced unavoidable conflict between exercising individual choice and
protecting public health, which then unfolded into a political debate over protective measures
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

37
mandated by the government. Especially controversial were public health recommendations to
wear masks in public settings. Despite consensus among public health experts that mask
mandates significantly decreased COVID rates (Lyu & Wehby, 2020), deaths and
hospitalizations (Adjodah et al., 2021), among Republicans, mask mandates were viewed with
considerable skepticism, not only for being inconvenient, but supposedly ineffective (Van Kessel
& Quinn, 2020). People in counties where Donald Trump saw strong support in 2016 were
significantly less willing to wear masks in public, even after controlling for a large number of
socio-demographic factors (Kahane, 2021). Similar effects were shown for several other
prevention behaviors (e.g., working from home, using hand sanitizer, avoiding large gatherings
of people, etc.). Druckman et al. (2020 & 2021) showed that partisans with high levels of out-
party animus, recorded prior to the outbreak of the pandemic, were motivated to distinguish
themselves from their political opponents by taking positions that differed from their out-party
and match those of their preferred party. However, these effects were muted in parts of the
country where COVID case counts were especially high, revealing that the rejection of public
health measures was due not to a lack of the capacity to understand the problem, but resistance to
information that threatened partisan identity (Druckman et al., 2021).
While a compelling contribution to the field, research demonstrating the relationship
between out-party animus and support for pandemic public health policies is still ultimately
correlational rather than causal. Moreover, while political opinions were clearly predicted by out-
party animus in this case, the research could not address the question of how reducing animus
might effect political opinions. This chapter reviews the burgeoning literature on the most
effective depolarization efforts, digs more deeply in to the potential connection between animus
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

38
and political opinions, and presents experimental evidence demonstrating the impact of
“reducing” affective polarization on political information processing.
Depolarization
Depolarization is a catch-all term given to a broad array of research directed at reducing
out-party animus and encouraging productive, evidence-based assessment of political topics.
There have been several attempts to experimentally out-party animus, but depolarization as an
interdisciplinary field has only recently developed to the point where systematic evaluation of
the best interventions is possible. Most recently, a report on a “megastudy” by Voelkel, Stagnaro,
Chu, et al. (2022), examined the effects of 25
9
simultaneously run experiments aimed at
depolarization with a high-powered national sample (N = 32,059). The report has not yet been
published in an academic journal, but it provides the most comprehensive look so far at the
available interventions and the relative strength of their effects in several domains.
10
The first
thing to note from this study is that participants in the control group, those who received no
intervention, “expressed concerning levels of partisan animosity, support for undemocratic
practices, and support for partisan violence” (Voelkel et al., 2022, p.7). The good news is that 23
out of 25 methods tested significantly reduced out-party animus. The most effective strategies
were those that exposed participants to “relatable, sympathetic exemplars with different political
beliefs” (p. 8) such as the Contact Project, which showed participants a video of people with
strongly different political views working together and coming to a friendly understanding with

9
In total, 252 interventions were submitted to the project for review from across the social sciences. The 25
interventions included in the were selected by an expert panel of scholars and practitioners.
10
See for example interventions: Ahler & Sood, 2018; Clayton & Willer, 2021; Hartman et al., 2022; Huddy &
Yair, 2021; Landry et al., 2022; Lees & Cikara, 2020; Levendusky, 2018a; Levendusky, 2018b; Mernyk et al., 2022;
Moore-Berg et al., 2020; Ruggeri et al., 2021; Santos et al., 2022; Simonsson & Marks, 2021; Voelkel et al., 2018;
Voelkel et al., 2021b; Warner et al., 2020; Wojcieszak & Warner, 2020; Zoizner et al., 2021.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

39
each other (over a beer).
11
Though not generally discussed by depolarization researchers, the
success of the Contact Project and other interventions like it, is theoretically backed up by
research into the interpersonal contact (Allport, 1954) and para-social interaction theories
(Horton & Wohl, 1956). These theories emphasize how interactions between members of divided
groups, or the positive depiction thereof, reduce between-group animosity under the right
circumstances. It would appear that current research in depolarization lends support to these
theories.
Despite the impact of contact-like interventions on affective polarization, manipulations
of this type had much weaker effects on support for undemocratic practices and support for
political violence. These findings at first appear to align with other research suggesting that
affective polarization is mainly about interpersonal rather than political perceptions (Santoro &
Broockman, 2021; Voelkel et al., 2021), but Voelkel et al. (2022) point out that there are strong
correlations among outcomes with partisan animosity at the “center of a cluster of several
societally relevant outcomes, including biased evaluation of politicized facts, general social
distrust, and preferences for social distance from out-partisans” (p.16). Utilizing what could be
considered a similar theoretical approach, but with a very different type of intervention,
Broockman, Kalla, and Westwood (2022, henceforth BKW) employed simulated positive inter-
partisan interactions in their depolarization experiments.
Using an Economic Game to “Reduce” Affective Polarization
In a standard trust game in behavioral economics research, there are two players, an
“Investor” (Player 1) and a “Trustee” (Player 2). Player 1 is endowed by the experimenter with
some amount of money, or something symbolically representing money such as “tokens,” and

11
See https://tinyurl.com/2hd6zyy5 to view the best performing intervention on out-party animus.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

40
has to decide how much of it they would like to “invest” in Player 2. Whatever amount is
invested will be tripled by the experimenter. Player 2 then has the opportunity to return an
amount of their choosing back to Player 1. This amount is not altered by the experimenter. At the
end of play, each participant receives an amount equal to a predetermined proportion of the funds
(or the value of the tokens) retained during the game. The game is played between two players
only once so that there are no effects from reputation in proceeding rounds of play.
To illustrate, consider a trust game initiating with a $10 endowment. If both players trust
each other completely and want to maximize the equal distribution of funds, Player 1 should
invest all $10 in Player 2. This amount is tripled by the experimenter, so Player 2 receives $30.
Player 2 can then return some, all, or none of these funds to Player 1, but if the goal is to be
perfectly fair, would return $15, thereby ensuring that both players receive the largest possible
evenly divided payout ($15 each). This is referred to as the “trust” game because there is no
economic incentive for Player 2 to return any portion of the investment, and knowing this, there
is no reason that Player 1 should transfer any money at all. The only reason to be generous is to
satisfy some self-imposed moral calculation. As such, the unique Nash equilibrium for the
Investor (Player 1) is to keep all of the money and invest nothing (Tzieropoulos, 2013). Of
course, this is not what humans tend to do. In the original trust game experiment, 30 out of 32
Investors invested at least some money and 11 Trustees paid back more than the amount
transferred by the Investor prior to it being tripled (Berg et al., 1995). As the game has been
repeated throughout the world, behavioral economists and psychologists have found that people
are usually at least somewhat generous, evincing a generalized social norm of trust and
reciprocity.
12

12
In a now famous article Henrich, Heine, and Norenzayan (2010) provide an excellent summary as well as several
cross-cultural comparisons of play in this and other economic games.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

41
The trust game, as well as the closely related dictator game, have been used previously to
measure out-partisan discrimination and the effects of party identification strength (Iyengar &
Westwood, 2015). Carlin and Love (2013) showed that investors (Player 1s) sent 34% larger
transfers to co-partisans than rival partisans. The amounts transferred to out-partisans declined
further as partisan identification strength increased. There was a caveat to this finding, however,
Player 1’s political party only mattered if their transfers were miserly. If Player 1 was generous,
Player 2 tended to reciprocate with large return transfers in kind, rather than relying on partisan
bias to guide their decision on the transfer amount.
BKW conducted five highly powered surveys of Democrats and Republicans (total N =
12,341) in an effort to understand the effects of affective polarization on political decision-
making. In surveys 1 through 4, participants played a mock economic trust game in which they
were randomly assigned to have a positive or negative experience with three members of their
opposing political party (no control group). These interactions were simulated, but participants
were told they were playing the game live with other (real) survey participants. Notably,
participants were told that they would receive bonus compensation equal to .05 times the amount
they “won” in the game. Participants randomly assigned to the negative experience condition
received miserly economic transfers from simulated out-party players ($0 out of $10 each round)
and received an end of game bonus of $0; participants in the positive experience condition
received generous transfers ($8 out of $10 on average each round) and an end of game bonus of
$1.20. After completing these games, participants in both groups were told that the reason for the
amounts transferred was their political party affiliation. The positive experience condition caused
a large reduction in affective polarization as measured by social distance items and feeling
thermometer scores relative to the negative experience condition. Despite this, BKW found no
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

42
corresponding effect on measures of electoral accountability, party loyalty, political
desensitization, convergence to party positions, support for bipartisanship, support for
democratic norms, or perceptions of objective conditions. They concluded that affective
polarization has uniquely interpersonal implications and does not impact political decisions.
BKW suggest that the approach taken in the trust game intervention is a valid
depolarization intervention based on the evidence that it reduced affective polarization, as
commonly measured. However, they also point out that a theoretical linkage between affective
polarization and political decision-making has not been well established in the literature thus far
(Broockman et al., 2022), and because their claim is that affective polarization does not affect
political opinions, their treatment of it as a theoretically important mediator is somewhat
superficial. Their discussion minimizes the importance of a key theoretical question: why would
we expect this particular intervention to have an impact on political decision-making?
Presumably, out-party animus mediates the value representation of a political choice at
the time of decision, thereby causing participants to make politically motivated selections that
align with their party position. It is here assumed that political decisions derived from affect
would necessarily be heuristically driven. Instead of making a careful systematic evaluation of
the political question they are presented with, affectively polarized citizens might rely on the
feeling of animus to guide them to a party-aligned position. To be clear, BKW present evidence
suggesting that this is not the case. Instead, they claim that animus directed at a social group,
such as a political party, is relevant interpersonally, but does not factor into the process of more
significant political decisions such as which candidate to vote for based on their policy positions.
Quoting Ottati and Wyer (1993) BKW lay out their first reason for being skeptical about
the link between affective polarization and political decision-making:
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

43
When forming judgments, individuals seek to use heuristics that are both easily accessible
and as relevant to the judgment as possible. As Ottati and Wyer (1993, p. 298) note,
‘evaluations of an object might not be based on affective reactions at all when individuals
have more relevant dimensions to rely on.’ When forming judgments about objects
related to but distinct from parties, such as candidates or norms, there may be other easy-
to-access attributes (and competing heuristics) that are usually more relevant than one’s
affect towards a party. One could therefore imagine partisan affect manifesting in
abstract survey questions about whether one would be unhappy having an outpartisan
neighbor (as there are no other dimensions of judgment available) but not evaluations
of an actual outpartisan neighbor or an actual outpartisan politician — cases when other
information is available. (Broockman et al., 2022, p. 4)
This quotation offers an important insight as well as a potential counterpoint to BKWs
interpretations of their own findings. If the only impact of the trust game is the reduction of
negative affect and a reduction in social distance preferences in the positive experience
condition, as BKW claim, then the trust game intervention shows that reducing out-party animus
does not impact political judgements. However, if the positive experience trust game cues a more
relevant heuristic or a more systematic mode of processing, one that is unobserved in the data
and primes politically motivated reasoning, then the effects of the experiment are confounded
and do not offer conclusive evidence for the exclusively interpersonal, non-political, influence of
affective polarization.
To determine if the trust game intervention truly has no impact on political decision-
making or if the intervention cued an unobserved cognitive process, I conducted three
experiments utilizing a modified version of the trust game intervention, with outcome measures
designed to capture the influence of politically motivated reasoning. The overarching hypothesis
of these experiments is that the trust game primes partisan social identity and thereby primes
politically motivated reasoning thus prompting participants to respond to outcome measures in a
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

44
manner that aligns with their partisan identity. This identity priming occurs in both the positive
and negative experience conditions of the trust game, which is why there was no observable
difference between the two groups. As with numerous effects reviewed in previous chapters, the
effects of political identity priming on political decision-making are moderated by partisan
ideological sorting. Effects are asymmetrical across party lines, such that partisans process and
respond to information differently depending on the topic. Effects on feeling thermometer scores
and social distance items in the BKW experiments are better understood as the result of demand
characteristics, rather than an actual lowering of affective polarization.
Participants and Survey Iterations
Data for this study come from a series of experiments conducted between March 29
th
and
July 12
th
2022. Data collection initiated with a series of pilot experiments designed to closely
emulate the intervention approach of BKW, in which the intervention included a negative
experience condition. Three relatively small pilot studies conducted largely with Democrats
(combined n = 165, n_Democrat = 125, n_Republican = 40) and a larger survey experiment (N = 235,
n_Democrat = 100, n_Republican = 135) were eventually combined into a single dataset for analysis,
Survey 1 (N = 400, see Table 2 for participant demographic information). All of the data
included in Survey 1 were generated through an intervention that included the negative
experience trust game and the positive experience trust game, but did not include a control
group. The dependent variable of interest for Survey 1, support for undemocratic practices, was
measured in a consistent manner throughout these pilots. One-way Analysis of variance of the
key dependent variable of interest showed no significant differences between pilot iterations for
Republicans F(4, 175) = .18, p = .95 or Democrats F(4, 225) = .65, p = .65.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

45
Survey 2, the primary research sample, included several modifications to the original
intervention discussed below (N = 661, n_Democrat = 333, n_Republican = 328), including the
replacement of the negative experience condition with a control group (see Table 1 in the
Appendix). All data were collected utilizing the online panel provider Prolific. Prolific has been
shown to provide higher quality data than Qualtrics or Amazon’s Mechanical Turk and is
comparable to panel providers such as Dynata in terms of data quality for non-probability
samples (Peer et al., 2021).
The target population was male (cis or trans), U.S. citizens, based in the U.S., age 19 to
80, who had participated in at least one recent national election (Presidential 2016,
Congressional 2018, or Presidential 2020) and who identified as members of either the
Democratic or Republican party. Independents who identified as leaning Democratic or
Republican could also complete the survey, however Prolific’s screening process ensured that
this group was small (n_Independent = 44 out of a combined N = 1,042 across all studies). Men only
were selected to participate in order to reduce variance in the sample population. Additionally,
participants needed to pass a knowledge check demonstrating their comprehension of the trust
game in order to complete the survey. Participants who failed the knowledge check were
provided partial compensation for their time. All checks and screening occurred prior to random
assignment. In all n = 34 incomplete observations, including n = 20 respondents who failed the
required comprehension check, and n = 2 participants who said they voted for candidates who
were not on the ballot in 2016 or 2020 (after they had a chance to correct the error) were dropped
from the study prior to analysis.
The target population for this study is a considerably more narrow sample than that of the
BKW study, which targeted a more broadly representative sample, included women, and did not
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

46
require participants to be voters. This difference is in part due to funding constraints which
limited the sample size, thus making it a priority to minimize variance in the sample. The present
study also more strictly stratified the sample on partisan identification because between party
asymmetries were anticipated. BKW do not report analysis involving between-party differences
for most outcomes.
Piloting
In the 5
th
survey conducted by BKW, the authors attempted to address several potential
concerns over the design of their intervention. They note, “perhaps a positive experience in the
trust game also affected constructs other than affective polarization and the effects of these other
constructs cancelled out the salutary effects of reducing affective polarization” (p. 26). Despite
acknowledging this as a motivation for the 5
th
survey, the authors did not sufficiently address this
possibility in the revised intervention design. The 5
th
iteration of the survey included a negative
experience condition, a handful of alternative depolarization interventions, and a no exposure
control group, but did not include a positive experience condition, making it impossible to
directly assess the noted concern.
Open Science Forum preprints show that BKW transitioned from claiming that the
negative experience condition increased affective polarization to saying that the positive
experience condition reduced it in the draft manuscript posted after the 5
th
survey. This might
indicate a null effect on affective polarization from the negative experience condition, which was
previously unobservable without a control group. This interpretation aligns with findings from
the pilot rounds of the present research and highlights an important question which is ultimately
not addressed by BKW. If the positive experience condition reduced affective polarization and
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

47
the negative experience condition had no effect on it, what was the negative experience condition
actually doing?
The pilot studies were used to test variations in survey instrument design and collect
qualitative feedback about participant experiences in the survey. Because substantial changes to
the manipulation and alternative hypotheses were generated based on the pilot data, a somewhat
detailed discussion is warranted here. The pilot Trust Game manipulation closely emulated that
of the game utilized by BKW, in which all participants played as the Trustee (Player 2), those
assigned to the negative experience condition received $0 in transfers, and the game played out
over a series of text-only screens. In the very first pilot, participants assigned to the negative
experience condition received transfers of $0 for all three rounds of play. This should
hypothetically have generated anger toward members of the opposing party. Manipulation checks
based on the feeling thermometer measure of affective polarization and social distance items did
not show significant differences between experimental conditions in the relatively small initial
pilot (n = 90). However, when asked if they felt they were treated “fairly” during the trust game,
being in the negative experience group caused a 91 percentage point reduction in perceived
fairness (SE = .04, p < .001, based on a linear probability model).
13
As such, the manipulation
seemed to work, in the sense that participants identified the transfers as unfair, but this did not
lead to a corresponding negative perception of the out-party. One potential explanation for this is
that the participants did not believe the simulated players were real and felt that they were being
treated unfairly by the researcher.

13
For binary outcomes, I generally utilizes logistic regression models and reports odds ratios. In this case, however,
the two groups were so small and unbalanced on the outcome that it was not possible to fit a model using maximum
likelihood estimation.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

48
This explanation was supported by open response comments as well as questions about
perceived deception at the end of the pilot survey. When asked, “Did you feel at any point in this
survey that the experimenter was trying to trick or deceive you in order to influence your
responses?” participants in the positive experience condition (in Study 1, N = 400) had
significantly lower odds of answering “yes” (OR = 0.50, 95%CI [.30, .83], p = .007). Following
this question, any participant who indicated they thought there might have been deception was
invited to elaborate with the prompt, “Please describe any times when you felt like there was a
deception involved.” The following is a sampling of responses from negative experience group
participants:
“The Trust Game seemed not real - like there wasn't another human actually
playing against me.”

“Getting my hopes up and giving me a bogus chance for a bonus only to have
‘players’ who gave me $0 and making me angry.”

“Researchers should just stop with deceptions. Nobody is buying it. The other
players are fake.”

“After being withheld money by 3 Republicans I feel like it was obvious that
would change my feelings about Republicans.”

“I don't believe I was playing with other real participants”

“I don't believe there were real people involved in the first part (money) task”

“I don’t believe there was another person in the game”

“I assumed there were no other participants for the bonus sharing task.”

“I do wonder if the amount I received in the trust game was fixed.”

There were several more comments along these lines. Notably, these questions were
presented at the end of the questionnaire, prior to the debriefing page where it was revealed that
participants in the negative experience condition would still receive a $1 bonus payment despite
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

49
not incurring winnings in the trust game (a bonus BKW seems not to have provided). These
comments offer some insight into ways the negative experience condition might have had
unforeseen effects. These participants did not believe they were playing the game with other real
survey participants and correctly perceived that the experimenter had (seemed to) select them to
receive no bonus compensation in order to elicit an angry reaction against out-partisans. Instead
of increasing animus toward the opposing political party, however, their anger was directed at the
experimenter. It has been shown elsewhere that individuals in “hot” states of mind may be
motivated to process information systematically (Moons & MacKie, 2007). In this case,
participants may have been specifically motivated to process information in a manner that defied
their correctly assessed anticipation of what the experimenter wanted them to do. In other words,
instead of increasing out-party animus, some proportion of negative experience participants were
primed to think more systematically about downstream political questions, and may have even
been motivated to pushed back against the perceived attempt of the experimenter to control their
thought process, a form of psychological reactance (Dillard & Shen, 2005; Scherr & Müller,
2017). This would help explain BKW’s null results, as it suggests that the manipulation might
have primed negative experience participants to throttle their partisan bias when confronted with
substantive questions. At the same time, comments from pilot participants reveal awareness of
the experimenter’s intentions, thus revealing a demand characteristic. If BKW’s participants had
a similar experience, it would help explain why direct measures of affective polarization and
social distance showed the expected increases or reductions, despite there being no
corresponding effect on political attitudes. It might be the case the participants in the positive
experience condition also identified the deception, but rather than being upset by it, knowing that
they were receiving more money, they were happy to go along with what the experimenter
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

50
seemingly wanted them to think and therefore expressed relatively lower levels of affective
polarization, again, despite the intervention having no impact on political attitudes.
BKW attempted to address these concerns in Survey 5, commenting that “an obtrusive
manipulation such as the trust game may raise respondent suspicion and produce demand. For
this to have led to null results, a positive trust game experience would need to have led
respondents to misrepresent their attitudes in a manner consistent with less accountability and
being more anti-democratic, cancelling out the salutary effects of the reduced affective
polarization” (Broockman, et al., 2022, p.26). The premise of the present research is that this is
indeed what happened. More specifically, the positive experience condition primed party aligned
politically motivated reasoning.
Since three of the experiments in the present study rely on some participants processing
information heuristically, anger can induce systematic processing, piloting revealed demand and
priming effects that could confound the analysis, and the record of BKW’s research suggests that
the negative experience condition did not cause the increase in affective polarization they
anticipated, the negative experience condition was not included in the main survey (Survey 2)
and participants in the present study were assigned to either a positive experience condition or a
control group in which participants did not play the trust game. For participants assigned to the
control condition, following the instructions for the trust game and the completion of a
comprehension test, a screen with the headline “Matching Players Unavailable” was displayed. It
went on to explain that “You have been routed to this page because we were unable to find other
players with whom you could play the Trust Game. You will still receive a 50 cent bonus
payment. Please press the arrow below to continue the survey.”
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

51
Another important difference between this study and the one it partially replicates is the
effort put into verisimilitude. Broockman, Kalla, and Westwood (2022) used simply designed
text boxes to convey the information occurring in the trust game. In the present study, over
multiple rounds of piloting, several details were added to make the mechanics of the game more
convincing. For example, when beginning the trust game, the look and feel of the survey
instrument changed. The university logo was removed from the page and a loading bar spun on
the screen with a message saying, “This survey is now linking with an external server where you
will be connected with other survey participants to play the Trust Game. This may take up to 30
seconds. Please do not leave this screen while being connected. This page will advance
automatically once connected.” The screen automatically advanced after 17 seconds of “linking.”
Similarly, participants were not allowed to move through the game at their own pace. When it
was time for the other “player” to make a decision, 9 to 14 seconds passed while they were
“deciding how much to transfer.” At the beginning of the first round and after each round of play,
a screen with the loading bar came up and informed participants that “You are being connected
with a new Player 1 for the next round. Please wait, this screen will advance automatically.” The
wait times at these “connecting” screens varied from 3 to 8 seconds. A 500 milliseconds screen
flashed up once a new player was “found” with the message “New Player Found! Connected,”
after which the round of trust game play commenced. Because these changes were made
incrementally and in conjunction with alterations to the core manipulation — variations in the
amounts transferred and eventually the removal of the negative experience condition — it is not
possible to attribute outcomes to any particular change. However, at the end of every round of
piloting as well as the main survey (Survey 2), participants were asked two questions to assess
the effectiveness of the deception:
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

52
(1) Did you feel at any point in this survey that the experimenter was trying to trick or
deceive you in order to influence your responses?
(2) While playing the economic Trust Game, which one of these statements best
represents how you thought of it?
a. I was playing the game with another survey participant
b. I was not really playing the game with another survey participant
c. I suspected that something was not real, but I was unsure what it was
Grouping the pilots studies into those that included the negative experience condition
(and were also less realistic) compared to the final version of the survey instrument (which did
not include the negative experience condition), the probability of a respondent stating that they
believed the experimenter tried to trick them declined by 29%, based on a bivariate logistic
regression (OR = 0.66, 95% CI[.48, .92], p = .013). Similarly, the probability of a participant
reporting that “I was not really playing the game with another survey participant” dropped by
around 25% (OR = 0.33, 95%CI[.25, .43], p < .001). It is also important to point out that
participants assigned to the negative experience condition in Survey 1 were 33% more likely to
indicate that they believed the experimenter was trying to trick or deceive them (OR = 0.50,
95%CI[.30,.83], p = .007) compared to control participants in Survey 2. These differences are
important to note because if the trust game manipulation check in the previous study indicated a
change in affective polarization due to a demand characteristic, i.e., participants knew what the
experimenter was trying to do and responded compliantly, the changes made to increase
verisimilitude might result in smaller and/or fewer statistically significant manipulation checks.
Procedure
The questionnaire initiated with a set of screening and demographic items. Because
Prolific makes survey links available only to panel members who qualify according to their
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

53
online database, rejection due to screen-outs was low in Survey 1 (n = 7) and nonexistent in
Survey 2. Participants who failed a required comprehension check designed to ensure an
understanding of rules of the Trust Game were screened out after three opportunities to answer
practice questions correctly (n = 20 across all studies). In addition to the trust game manipulation
experienced by the treatment, but not the control group, participants completed three survey
experiments designed to capture the effects of the positive experience condition on political
information processing, following the trust game and manipulation checks. These experiments
were presented in a random order to counterbalance potential priming effects, recency, and
primacy bias from their position in the instrument relative to the intervention and each other. All
participants were debriefed following the survey and informed that they would receive bonus
compensation even if they were assigned to the control group and had not participated in the trust
game. All results presented here refer to findings from Survey 2 (N = 661) unless otherwise
specified.
Following screening and demographic items, all participants were led through
instructions on how to play the trust game and answered a series of practice questions to ensure
that they understood the rules. As a reminder, in a standard (live) trust game there would have
been two players, an “Investor” (Player 1) and a “Trustee” (Player 2). Play would proceed as
follows: Player 1 is endowed with $10 by the experimenter and has to decide how much of it he
would like to “invest” in Player 2. Whatever amount is invested is tripled by the experimenter
(e.g., $8 x 3 = $24). Player 2 then has the opportunity to return an amount of his choosing back
to Player 1. This amount is not altered by the experimenter and can be some, all, or none of the
invested amount. At the end of play, each participant receives an amount equal to a
predetermined proportion of the funds retained during the game (in the case of this study, .025
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

54
multiplied by winnings). The game is played between two players only once, so that there are no
effects from reputation.
In keeping with the version of the trust game adapted by Broockman, Kalla and
Westwood (2021), participants in the treatment condition (positive experience trust game) were
told that they would be matched with another survey participant and assigned a role at random.
In reality, all treatment condition participants were assigned to play as Player 2 (the Trustee) for
three rounds of the game, while the actions of three separate Player 1s were simulated.
Participants were presented with some basic demographic information about the supposed other
player: their age (random between 20 and 42), income range (random range of less than $25,000,
$25,000 to $50,000, or $50,000 to $75,000), and partisanship (that of respondent’s out-party).
Partisanship remained consistent for all three rounds of the game. The BKW experiment also
included a random gender for the simulated player, but because the present study included men
only, gender was not part of the demographic profile of simulated players. Participants in the
trust game received generous economic transfers for all three rounds of play, $8 (multiplied to
$24) in the first round, $7 (multiplied to $21) in the second round, and $10 (multiplied to $30) in
the third round. This order of transferred amounts was consistent for all participants. By
maintaining an average of $8 per round from prior research, but ending the game with the
highest possible transfer, the intention was to make the highest value payout most accessible as
participants answered the questions that followed. After each simulated Player 1 transfer, as
Player 2, participants had the opportunity to return some, all, or none of the now tripled amount
back to Player 1.
14
While making this decision a timer counting down from 20 seconds was

14
Twelve percent of participants (n = 39) returned $0 for all three rounds. The average total amount returned was
$31 total (SD = 17.3). Eleven participants returned $75, the maximum possible, and thereby reduced their own
bonus to $0. This is in line with prior research demonstrating a general tendency toward reciprocity in real trust
games.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

55
displayed on the screen. The urgency created by the timer was intended to add a layer of realism
to the simulated interaction by conveying the idea that someone was on the other side awaiting
their response. After entering a return amount, participants were asked to indicate which factors
(age, income, partisanship, the amount sent to me, or something else) were part of the decision
making process. This was again to add realism to the simulation at the end of the game, when
participants were shown a page summarizing their winnings and the supposed reason for the
transferred amount. For two of the three rounds, partisanship was indicated as the reason for the
transfer amount, while “something else” was indicated in two of the three as well. This is another
small departure from the BKW intervention. In that case, political party was listed as a reason for
all three transferred amounts. This change was made to prioritize realism and reduce the potential
for demand characteristics while maintaining the major elements of the previous intervention.
Following the trust game, participants completed a short series of manipulation checks based on
common measures of affective polarization: feeling thermometers and social distance items. At
this point, participants were randomly assigned to conditions for the three embedded
experiments.
As described above, participants assigned to the control group learned the rules of the
trust game, but when it came time to play, were presented with a screen stating that no matching
participant could be found at that time. They were also informed at this point that they would still
receive a 50 cent bonus, so as to avoid any potential effects from having their bonus denied.

Experiment 1: Party Endorsement and Policy Support
In one of the most frequently cited experiments looking at the effects of partisan
motivated reasoning, Bosel, Druckman and Cook (2014), demonstrated that political party
endorsement can cause citizens to support (oppose) policies that they would otherwise oppose
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

56
(support) in the absence of an endorsement. Modelled on that experiment, after exposure to the
trust game manipulation (or control), participants were assigned to read about two pieces of real
bipartisan legislation: the Great American Outdoors Act (GAOA) and what became the
Protecting Americans from Gun Violence Act of 2022, which was passed by Congress while the
experiment was ongoing. Both of these achieved bipartisan support, but they varied substantially
in terms of their divisiveness, novelty, and salience in the public debate. Whereas the GAOA was
passed in 2017, had relatively stronger bipartisan support, and was about a topic that does not
have a strong attachment to partisan identity (maintenance of national parks), the Gun Violence
Act was being actively debated at the time of data collection, represents “the most significant
revision to the nation’s gun laws in decades” (Stolberg, 2022), and is about a topic with a strong
attachment to partisan identity for Republicans (Lacombe et al., 2019).
The contrast between these two pieces of legislation is intended to help clarify the effects
of the negative experience trust game. As suggested earlier, the negative experience game (in the
pilots) succeeded in eliciting anger, but did so in a manner that it was directed at the
experimenter. It is known that “hot” cognitive states can cue systematic information processing.
The Gun Violence Protection Act could be viewed as threatening to some Republicans, thereby
eliciting a similarly “hot” cognitive state. By eliciting this mindset independent of the trust game
itself, it is possible to interrogate the argument that this state of mind made respondents more
psychologically reactant in response to a perceived threat to freedom (Brehm & Brehm, 1981),
i.e., the impression that the experimenter is trying to get them to answer in a particular way.
The stimuli for this experiment were summaries of each bill accompanied by a partisan
endorsement. For the Great American Outdoors Act, the summary read as follows:
Bill Summary:
Through the Great American Outdoors Act (GAOA), the Department of the Interior is
investing $1.9 billion annually for five years ($9.5 billion total), in much-needed
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

57
maintenance for national parks, forests, wildlife refuges, and recreation areas. Specifically,
the law establishes the National Parks and Public Land Legacy Restoration Fund to support
deferred maintenance projects on federal lands. 50% of energy development revenues
credited as miscellaneous receipts from oil, gas, coal, or alternative or renewable energy
development on federal lands and waters are to be deposited into this fund (up to $1.9
billion annually). GAOA also authorizes permanent funding of the Land and Water
Conservation Fund at $900 million annually to improve recreational opportunities on
public lands, protect watersheds and wildlife.

This summary was based on the one provided by GovTrack.us, an independent legislation
tracking service. The way that each summary was written was intended to be comprehensible if
participants read carefully, but also wordy and esoteric enough that a quick skim of the summary
would not necessarily provide a clear understanding of its specific content. In order to assess the
bills’ content accurately, participants needed to engage in at least somewhat attentive and
deliberate (systematic) information processing. However, if they did not want to read the
summaries carefully, they had another way to evaluate their level of support. Each bill
description was accompanied by one of two (own-party or out-party) randomly assigned
endorsements preceding the summary on the same page. The endorsements for the GAOA were
structured as follows:
“Party Endorsement: POLITICAL PARTY. This bill was passed into law during a
period of divided government when fellow partisans voted together nearly all of the time.
It was co-sponsored by NUMBER OF PARTY-IDENTIFIED CO-SPONSORS,
including Ranking Member WELL KNOW LEGISLATOR FROM SAME PARTY,
and supported by a majority of POLITICAL PARTY legislators.

For example, a Republican (Democratic) participant randomly assigned to the party incongruent
(congruent) condition read this endorsement of the GAOA:
Party Endorsement: Democratic. This bill was passed into law during a period of divided
government when fellow partisans voted together nearly all of the time. It was co-
sponsored by 18 Democrats, including Ranking Member John Lewis (D-Georgia), and
supported by a majority of Democratic legislators.

If a Democrat (Republican) was assigned to the party incongruent (congruent) condition, they
read:
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

58
Party Endorsement: Republican. This bill was passed into law during a period of divided
government when fellow partisans voted together nearly all of the time. It was co-
sponsored by 10 Republicans, including Ranking Member Kevin Brady (R-Texas), and
supported by a majority of Republican legislators.

Taber and Lodge (2006) pointed out that political issues are “hot” for most people. Social
identity-relevant political topics trigger affective responses that precede semantic information
processing. Put differently, people often form attitudes or interpret information in a manner that
is consistent with identity group norms regarding a general topic rather than the specific content
of presented information. This is a subconscious process that we experience as an emotional
reaction, which in turn biases our assessment of information. For this reason, less politically
divisive issues — supporting our national parks, for example — are unlikely to elicit strong
counter-arguments compared to issues that are more salient to political identity — such as gun
control. As such, one might expect to see political-identity salient topics treated with greater
skepticism when proposals are endorsed by the opposing party, but more support for out-party
proposals on topics that are less relevant or threatening to identity.
Americans are starkly divided over the topic of gun control. Looking at the ANES survey
data from 2020 on the question “should the government make it more difficult or easier to buy a
gun, or should the rules stay the same as they are now,” there is a 51 point (SE = .01) difference
between the parties comparing those who think the laws should stay the same or make it easier to
buy a gun compared to those who say that it should be more difficult (Waldχ
2
= 2227.57, p <
.0001). Democrats strongly support making it harder to purchase guns (78%), while for
Republicans, gun ownership is a powerful social identity that predicts firearm-related policy
attitudes (Lacombe et al., 2019).
For the Gun Violence Act, which was referred to as “Gun Law Reform” at the time of
data collection, the summary read:
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

59
Bill Summary:
The emerging policy would put extra scrutiny on gun buyers under the age of 21 by
requiring an investigative period to review juvenile and mental health records. Convicted
domestic violence abusers and people subject to domestic violence restraining orders
would now be included in the National Instant Criminal Background Check System. The
proposal would also provide states with incentives to pass so-called "red flag" laws, which
would allow police, family members or even doctors to petition a court to take away
someone's fire arms if they are found to be a danger to themselves or others. The policy
would also crack down on "straw purchases," i.e., the illegal act of buying a gun for
someone who is prohibited by law from possessing one.

This summary was developed based off of a wide variety of news sources (including
NPR, The New York Times, and the Los Angeles Times) and does not represent the precise
content of the final bill. It was accompanied with an endorsement statement constructed as
follows:
Party Endorsement: Democratic (Republican). This policy outline has been proposed
in the current congress, in which fellow partisans voted together nearly all of the time. In
response to the recent mass shootings in Buffalo, New York and Uvalde, Texas, a group
of Senators, including 18 Democrats (10 Republicans) led by Richard Blumenthal, D-
Connecticut (John Cornyn, R-Texas) and Cory Booker, D-New Jersey (Roy Blunt, R-
Missouri) have outlined a package of reforms to America's gun laws.

All of the information presented in the endorsements and summaries were accurate at the
time of data collection. Regardless of the topic of the bill, it is expected from prior research
(Bolsen et al., 2014; Taber & Lodge, 2006) that partisans will express stronger support when the
bill is endorsed by members of their own party relative to members of the out-party.
Hypothesis 1: Support for both pieces of legislation will be higher when they are own-
party relative to out-party endorsed.
While this hypothesis is straightforward, the primary interest of this experiment is a more
complex research question. To reiterate from earlier, the central hypothesis of this study is that
the positive experience trust game condition primes party aligned politically motivated
reasoning. Thus despite any apparent “reduction” in affective polarization, when it comes to
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

60
assessing political information, treatment condition participants will more strongly express views
that align with known party positions.
Research Question 1: How will exposure to the positive experience trust game interact
with party endorsement to affect support for legislation?
While stated as a research question, the direction of effects in RQ1 is predictable, though
potentially unintuitive. If the positive experience trust game primes political identity and thus
politically-motivated reasoning, then we would expect the highest levels of support for
legislation in the treatment condition when the content of the legislation is congruent with known
party positions and is endorsed by one’s own party. This should be the case for the less
controversial piece of legislation, the GAOA. Likewise, it is expected that the lowest levels of
support for the GAOA will be found among partisans who have had their political identities
primed and been exposed to an out-party endorsement.
Hypothesis 2: For members of both parties, the difference between the two endorsement
conditions will be larger in the treatment group than in the control group for the GAOA.
However, for the Gun Violence Act, participants are likely to have much stronger
identity-based beliefs and thus the direction of effects from the endorsement and trust game
treatment (political identity priming) could be at odds, specifically for Republicans. Given the
association between gun ownership and political identity for Republicans, the topic of gun
control legislation alone is likely to elicit identity-protective cognition. The perceived identity-
threat of “having our guns taken away” (in the parlance of Republican talking points) will be
exacerbated by the identity-priming treatment, thus reducing support for the legislation.
Simultaneously, the Republican endorsement should increase support for the legislation, at least
among those who do not have strong prior beliefs or identity-based attachments. If priming
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

61
political identity has a stronger effect than endorsement, or causes participants to engage in
reactive reasoning — pushback against perceived coercion by the researcher –– then
participation in the trust game will reduce Republicans’ support for Gun Violence Prevention
Legislation regardless of the endorsement. This leads to the following hypothesis.
Hypothesis 3: Republicans who have had their political identities primed through the
positive experience trust game will express lower levels of support for the Gun Violence
Prevention Act, regardless of party endorsement.

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

62
Experiment 2: Support for Undemocratic Practices
Utilizing three items from the experiment conducted by Voelkel et al. (2021), which also
used the positive / negative experience trust game as its intervention, participants were randomly
assigned to one of two undemocratic practices conditions. In the first condition, the “unjustified”
condition, participants read a standard set of questions used to measure support for practices such
as redrawing districts to favor their own party and bending voting rules to hurt the out-party’s
chance at winning. In the second condition, the “justified” condition, these questions were
altered to include a brief and non-substantive justification for the practice.
Research into the social psychology of ethical dissonance shows that providing even a
nonsensical justification for a breach of norms is enough to increase willingness to accept the
violation, at least when the stakes are low (Barkan et al., 2015). The non-substantive nature of
these justifications is meant to emulate the manner in which political elites often villainize the
out-party. With the Presidency of Donald Trump, Americans saw the ascendence of a political
strategy that was light on substantive policy, but fervent about the perceived threat posed by out-
group members. A common chant during Trump’s 2016 campaign was to “build that wall,” on
the U.S.-Mexico border, in order to prevent (illegal) immigration, despite numerous experts who
declaimed the wall strategy as ineffective. Similarly, rally goers chanted “lock her up” as a call
to imprison Trump’s Democratic opponent despite the Justice Department’s decision not to
prosecute Clinton for using her own email server during her time as Secretary of State (Politico,
2016; Politico Magazine, 2016). After four years characterized by anti-democratic rhetoric,
Trump’s presidency ended with accusations of seditious conspiracy over his alleged collusion
with white supremacist groups to prevent the certification of the 2020 election by storming the
Capitol building on January 6, 2021. Two years since the end of his presidency, Trump’s strategy
lives on in the conspiracy known as the “great replacement theory,” which is pushed out to the
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

63
viewers of Fox News on a regular basis, as a warning to so-called “legacy Americans.” Here it is
succinctly summarized by Fox host Tucker Carlson:
“Now, I know that the left and all the little gatekeepers on Twitter become literally
hysterical if you use the term “replacement,” if you suggest that the Democratic
party is trying to replace the current electorate — the voters now casting ballots,
with new people, more obedient voters from the third world. … But they become
hysterical because that’s what’s happening, actually. Let’s just say it — that’s
true”
15
(Barbaro, 2022).

The replacement theory and out-group animus have found their most violent expression
in the writings of multiple mass shooters. Among them, the 18-year-old who, in May, live-
streamed himself as he attempted to kill as many Black people as possible at a supermarket in
Buffalo, New York. There can be little doubt that the recent spate of racist violence is motivated
by an animus that Trump and his abettors invited upon this country. It therefore stands to reason
that if such violent extremism can be elicited on the basis of justifications that center on fear of
the outsider that providing similar non-substantive justifications should cause a shift in political
attitudes, at least among Republicans who have been prepared by repeated exposure to accept
this type of argument.
Voelkel et al. (2021) did not find an effect from the trust game affective polarization
reduction on support for undemocratic practices. As such, it is likely that participants will only
respond more favorably to the practice if they have both had their political identities primed (by
the trust game) and received a justification. As an example of how the justification manipulation

15
Excerpt from an archival recording of Tucker Carlson taken from a transcript of the New York Times podcast The
Daily, which aired on May 16, 2022.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

64
operated, participants were assigned to read one version of the following statements and express
the strength of their agreement on a 7-point Likert-type scale.

In this example, the addition of “because the out-party would do worse” is the only
difference between the text of the two conditions. This discussion leads to Hypothesis 4.

Hypothesis 4: Participants exposed to the positive experience trust game will express
higher levels of support for undemocratic practices when they are provided with a non-
substantive justification for doing so.

Given that this particular topic is centered on the inducement of anger directed toward the
out-party, it is worth considering whether the negative experience trust game, which reliably
evokes anger, though it is not reliably targeted, could have an impact on support for
undemocratic practices when a justification is provided. Particularly among Republicans, for
whom fear and anger directed toward out-groups is a mainstay of political rhetoric even in the
absence of substantive reason, it is possible that a non-substantive justification could be enough
to channel animus toward their political opponents and increase support for undemocratic
practices. As such, for undemocratic practices outcomes, in addition to the data collected in
Survey 2, data from Survey 1, which included the negative experience trust game but no control
group, were also analyzed under Hypotheses 5a and 5b.
Unjustified: I think {Own-Party} should do everything they can to hurt the {Out-
Party}, even if it is at the short-term expense of the country.
Justified: I think {Own-Party} should do everything they can to hurt the {Out-
Party}, even if it is at the short-term expense of the country because the {Out-
Party} would do worse.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

65

Hypothesis 5a: Republican participants will express higher levels of support for
undemocratic practices when exposed to the negative experience trust game and a non-
substantive justification for the practices, relative to Republicans in the positive
experience and unjustified conditions.

Hypothesis 5a aligns with the theory underlying the BKW intervention, which assumed
that the negative experience condition increased out-party animus and could thereby alter
political attitudes (though they found this not to be the case). The alternative hypothesis
underlying the present research is that the negative experience trust game triggered reactance
against the experimenter’s attempt to manipulate participants into disliking the out-party. As
such, it is expected that the very opposite of what is proposed in Hypothesis 5a will actually
occur.
Hypothesis 5b: Republican participants will express lower levels of support for
undemocratic practices when exposed to the negative experience trust game and a non-
substantive justification for the practices, relative to Republicans in the positive
experience and unjustified conditions.
The inverse of this relationship, as expressed in Hypothesis 4, is also likely to be true in
the version of the game with the negative experience condition. Republicans exposed to the
positive experience trust game will have had their identities primed, but without corresponding
anger or reactance, thus putting them in a state to accept justification for undemocratic practices
more readily and increase their support for these practices.

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

66
Experiment 3: Identity-Based Motivated Learning
The topic of immigration sharply divides Democrats and Republicans. For both, the
strength of one’s political identity is predictive of significant differences in the perception of
immigrants, but this is a relatively recent phenomenon. On the subject of illegal immigration, as
recently as 1994, Democrats and Republicans had nearly identical views. Analysis conducted by
Pew Research showed that 62% of Democrats and 64% of Republicans agreed that “immigrants
are a burden on our country because they take jobs, housing and health care” (Khalid, 2019). As
of 2019, only 11% of Democrats agreed with that statement, compared to 49% of Republicans.
More relevantly for the present study, in 2020 the ANES began tracking views on the question,
“Does illegal immigration increase, decrease, or have no effect on the crime rate in the U.S.?”
(American National Election Studies, 2021b). On this question as well, views are notably
polarized (Figure 6).

Figure 6: Party Perceptions of the Relationship Between Illegal Immigration and Crime

0
10
20
30
40
percent
Republican Democrat
1. Increase 2. Decrease 3. Have no effect 1. Increase 2. Decrease 3. Have no effect
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

67
Democrats are more likely to believe that immigration has no effect on crime, while
Republicans believe practically the mirror opposite. Furthermore, as shown in Figure 7 below,
the average belief that illegal immigration causes crime increases significantly the more strongly
identified as a Republican one is, while the opposite is true for Democrats, F(3, 4838) = 953.32,
p < .001. Figures 7 helps to illustrate the commonly noted connection between partisan identity
strength and issue position. The more strongly identified as a Democrat or Republican, the more
extreme is one’s positions on immigration.

Figure 7: Belief that Illegal Immigration Causes Crime by Political Identity Strength

In Experiment 3, a task modelled on one developed by Glinitzer, Gummer and Wagner
(2021) was used to investigate if ideological sorting — alignment of identity-based ideology
with partisan identity — strengthens identity-based motivated learning around a political issue.
2
2.5
3
3.5
4
Linear Prediction
Strong Republican Moderate Republican Moderate Democrat Strong Democrat
Party Identification Strength
Illegal Immigration Causes Crime by Party ID Strength
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

68
Participants were randomly assigned to one of two covariance detection conditions, in which
numerical information supported one of two randomly assigned conclusions: crime increased
more in cities with refugee housing, or crime increased more in cities without refugee housing.
Participants were told that data in the table came from a real academic study, though it did not,
and were asked to indicate which of the two conclusions they believed based on the information
presented. The topic was framed as referring to immigration from South and Central America in
order to distinguish it from refugee assistance for those fleeing the war in Ukraine. Prior research
has shown that people tend to process information in a fast-and-frugal (heuristic-driven) way
when the easily computed (but incorrect) answers confirm their prior beliefs (Kahan et al., 2013).
When prior beliefs are threatened, people tend to process information in a more deliberative,
systematic manner. To answer the question correctly, participants needed to engage in systematic
processing of a relatively challenging mental mathematical problem, where the correct answer
was always the less intuitive. See Figure 8 for the full question.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

69

Or

Figure 8: Refugee Housing and Crime Covariance Detection Task

In addition to the tables displayed in Figure 8, two other tables with the order of the rows
flipped were also used to ensure that outcomes were not influenced by order effects. Otherwise
the correct answer was always determined by the column labels. The most direct way to estimate
the correct answer is to sum the columns and then compare the “Crime increased” ratios. For
example, in the final table displayed above,
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

70
Increase with housing: 240 + 80 = 320; 240/320 simplifies to 3/4 or 9/12;
Increase without housing: 100 + 20 = 120; 100/120 simplifies to 5/6 or 10/12;
Because 10/12 > 9/12 crime increased more in cities without refugee housing.
At a glance, participants are likely to focus on the magnitude of the numbers 240 and 80,
and therefore answer incorrectly if they are not motivated to process the information more
deeply. The key expectation for this task is that participants will engage in systematic processing
of information when the fast-and-frugal conclusion threatens their beliefs. More specifically,
highly sorted Democrats are more likely to answer the question correctly when 240:80 is in the
“increase with refugee housing” column, because they will be more skeptical of that answer and
concentrate on the math. The opposite is true for Republicans.
Hypothesis 6: The proportion of correct answers will be associated with the strength of
partisan-ideological sorting, such that highly sorted partisans will more often select the
fast-and-frugal (incorrect) conclusion when it conforms to partisan-ideological positions.
Leading out from Hypothesis 6, its inverse is also proposed here. When the fast-and-
frugal conclusion is threatening, highly sorted partisans will engaged in more systematic
information processing and thereby answer correctly more often. Glinitzer, Gummer and Wanger
(2021) showed that participants who were asked to complete a version of this task tended to
answer in a manner that conformed to their prior stated beliefs, as recorded with an earlier
measured survey item. This indicates not simply politically motivated reasoning, but politically
motivated learning, since no participant had been exposed to information presented prior to the
task, and shows that people who hold different beliefs can look at the same objective information
and come to divergent conclusions. In the present study, this work is extended to show that not
only do prior stated beliefs bias the learning of objective information, but the strength of one’s
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

71
partisan social identity, when that identity is bound up with a polarizing belief, will also bias the
learning of such information.
Because the task in the present study is about the influence of a salient social identity
rather than a reflection on the way that prior beliefs influence motivated learning, there is a
possibility that the strength of ideological sorting will only matter when participants have had
their political identities primed. This leads to the following research question.
Research Question 2: How will partisan identity priming (the positive experience trust
game) interact with partisan-ideological sorting to effect the proportion of correct
answers in the covariance detection task.
As with Research Question 1, the central hypothesis of this study predicts the direction of
effects. When partisan identity is primed, the effect anticipated by Hypothesis 6 will increase,
leading to greater differences in the proportions of correct and incorrect answers when
comparing individuals within the same party who are more or less ideologically sorted.

Measures
Manipulation Checks
Feeling Thermometers. Feeling thermometer scales ranging 0 (unfavorable) to 100
(favorable) recorded perceptions of respondents own and out-party members. Affective
polarization was estimated as own-party minus out-party feeling thermometer score, recoded to
range from 0 to 1 (M = .37, SD = .30). The raw interpretation of this measure is that partisans
favor their own party 37 points higher than the out-party on average. The language used for these
measures is a slight departure from the ANES measure, which asks about feelings toward
“Democrats” and “Republicans,” whereas here the question referred to “People who are”
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

72
Democrats or Republicans. This wording of the questions is identical to that used by Broockman,
Kalla, and Westwood (2022) for comparability.
Social Distance. Preference for social distance from out-party members was measured
with four items recorded on 1 to 4 point scales ranging from 1 = “Not at all comfortable / upset”
to 4 = “Extremely comfortable / upset” with the fourth item reverse coded (alpha = 0.77).
Examples of items include “How comfortable would you be having neighbors on your street who
are [out-party members]?” and “Suppose a son or daughter of yours was getting married. How
would you feel if he or she married a supporter of the [out-] party?” These are commonly used
measures, but this particular set of questions was taken from Druckman and Levendusky’s
(2019) “What do we measure when we measure affective polarization.” Items were averaged and
then the resulting variable was recoded to run from 0 to 1 (M = .59, SD = .16).
Covariates
Ideological Sorting. Ideological sorting is composed of two scales. The first is the
standard 7-point ideology scale used in the ANES and elsewhere that runs from 1 = “Very
Liberal” to 7 = “Very conservative.” The second is a 4-item partisan identity strength index
measured on 5-point Likert scales, which is a validated measure of identity strength created by
Green (2004) and Huddy et al. (2017). It includes the following items: “How important is being a
[own-party] to you?” from “Not important at all” to “Extremely important”; “When talking about
[own-party], how often do you use ‘we’ instead of ‘they’?”, from “Never” to “All the time”;
How well does the term [own-party] describe you?” from “Not at all” to “Extremely well”; “To
what extent do you think of yourself as a [own-party]?” from “Not at all” to “A great deal”
(alpha = .89). After recoding both scales to run from 0 to 1 they were averaged together to
produce a single measure of partisan-ideological sorting (M = .55, SD = .19).
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

73
Dependent Variables
Policy Support. Policy support, for the party endorsement experiment, was measured with
three items for each bill captured on 7-point scales ranging from low to high on each item: “To
what extent do you oppose or support the [bill]?” “To what extent is the [bill] unimportant or
important for America?” “To what extent do you think the [bill] will be ineffective or effective at
[restoring national parks and federal lands / reducing gun violence]?” These items were average
for each bill and the resulting variable was recoded to range from 0 to 1. Both sets of items had
high reliability coefficients (alpha GAOA = .86, alpha Gun Violence Prevention = .89). When
submitted to exploratory factor analysis, for both bills the items loaded on to a single factor, in
all cases with loading scores above .7, without rotation.
Support for Undemocratic Practices. Support for undemocratic practices was measured
as described previously in this chapter based on items adapted from V oelkel et al. (2021). Each
of three items referred to a particular practice and was measured on a 7-point scale ranging from
1 strongly disagree to 7 strongly agree. Though the wording of questions differed depending on
the assigned experimental condition, the reliability alpha for both sets of scales as well as the
combined measure were above .84. Items were averaged and the resulting variable was recoded
to range from 0 to 1.
Results
Sample Characteristics
Referring to Survey 2, the trust game and control groups were broadly balanced across
demographic factors with no significant differences between the two groups in terms of race
(Black or White), religion (Christianity), income, education, or suburban/urban locality as
assessed through logistic and linear regression analysis (see Table 1 in the Appendix). The two
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

74
groups did differ somewhat in terms of age, as participants assigned to the positive experience
trust game condition were 2.13 years older on average (M = 36.52, SD = 13.20 versus M =
34.38, SD = 11.46, t(659) = -2.22, p = .027). As such, age is included as a covariate in the
forthcoming models. There were no significant demographic between group differences in any of
the three experiments that followed the priming intervention. The numbers of individuals
identifying with each party did not differ significantly across treatment conditions either, as
reported above.
Comparing Democrats and Republicans to each other, there were no differences in terms
of age, but, as expected, the two parties did differ on several other demographic characteristics
(see Table 2). Over 50% of Democrats identified as atheist or agnostic while 67% of Republicans
identified as Christian (Catholic or Protestant) (OR = 0.19, 95% CI[.14, .27], p < .001). Notably,
this is a much higher number of Atheists and Agnostics than the proportion in the Democratic
electorate (14% of Democrats in the ANES 2020 dataset). Democrats were around 36% less
likely to live in a rural area than Republicans (OR = 0.56, 95% CI[.36, .88], p = .012), while
Democrats were 68% more likely to live in an urban area than Republicans (OR = 2.13, 95%
CI[1.49, 3.03], p < .001). Majorities of both parties lived in the suburbs (59% overall). The
overall level of education was higher than the 2020 electorate, with 54% of participants holding a
Bachelor’s Degree or above (compare to 45% in the 2020 ANES). However, 60% more
Democrats than Republicans had a Bachelor’s degree or higher (OR = 1.52, 95% CI[1.12, 2.07],
p = .008). The difference in education between the two parties is almost identical in magnitude
to that of the 2020 electorate according to ANES data.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

75
Manipulation Checks
Broockman, Kalla, and Westwood (2022) note that when the trust game intervention was
run with the negative and positive experience conditions, its impact on the feeling thermometer
measure of affective polarization was a 14.3 degree difference. In the present study, the
equivalent effect was smaller, 4.8 degrees (95% CI[0.18, 9.46], p = .04); however, this is the
difference between the positive experience condition and a no exposure control group, for whom
there should be no demand effects. In the versions of the experiment where the negative
experience condition was included (Survey 1), this gap widened to 6.9 degrees, still well below
the effect shown by BKW, but perhaps understandably so given the differences between the
target populations and the effort put into reducing demand characteristics through verisimilitude
in the present study. Nevertheless, this check certifies that the trust game manipulation had the
hypothesized effect on affective polarization, reducing it in the positive experience conditions in
both cases.
Unlike the BKW study, in the present study there were no significant differences between
the treatment and control groups on social distance items (b = .017, 95% CI[-.01, .04], p = .182).
The explanation for this null between group difference cannot be fully explicated with the
available data, but if effects on affective polarization were, as proposed previously, rooted in
demand characteristics of the survey and changes made based on lessons from piloting the
survey had their intended effect, it is not surprising that no significant difference is shown with
this manipulation check. The effort made to convey the impression that interactions during the
trust game were real likely reduced courtesy bias in the responses.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

76
Experiment 1 Results: Support For Legislation
Hypothesis 1 stated that participants exposed to endorsements by their own party, relative
to the out-party, would express greater support for the legislation. This hypothesis was supported
for both pieces of legislation. Overall, the Great American Outdoors Act (GAOA) received 5
points more support when it was presented as own-party relative to out-party endorsed (N = 661,
F(3, 657) = 26.6, p < .001, R
2
= .10, 95% CI[-.08, -.03], p <.001).
16
The Gun Violence
Prevention Act received 6 points more support with own-party endorsement (F(3, 657) = 96.5, p
< .001, R
2
= .31, 95% CI[-.09, -.02], p = .002). Majorities of both parties supported the GAOA,
but Democrats expressed 11 points more support overall (M = .77, SD = .15) than Republicans
(M = .68 , SD = .20, 95% CI[.08, .13], p < .001). As expected, the Gun Violence Prevention Act
was much more divisive with a 31 point difference between the two parties (95% CI[.28, .35], p
<.001). Eighty-three percent (SD = .30) of Democrats supported the proposal, while a slim
majority, 52% (SD = .15), of Republicans did so, pooling all experimental conditions.
Research Question 1 asked how exposure to the positive experience trust game would
interact with party endorsement to affect support for legislation. A linear regression model
testing the interaction between the treatment condition and the endorsement frame across the
entire sample while controlling for party identification and age was used to address this question
(N = 661, F(5, 660) = 15.26, p < .001, R
2
= .10). Pairwise comparison of between group
differences using Tukey’s correction revealed that participants in the treatment condition
(positive experience trust game) who were exposed to an endorsement from their own party
expressed no greater support for the GAOA on average than those in the control condition (diff =
.009, 95% CI[-.04, .06], p = .970), all else constant. This indicates that the additive effect of

16
Endorsement effects estimated based on linear regression controlling for political party and age. Between party
contrasts were based on Student’s t-tests.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

77
priming on top of party endorsement was non-significant for the less controversial piece of
legislation.
17
Keeping in mind that the positive experience condition was hypothesized to prime
party-aligned motivated reasoning, the more interesting comparison is between the effect of
being in the treatment condition versus control when there is an own-party frame compared to an
out-party frame. If the positive experience trust game primes politically motivated reasoning, we
would expect the difference between own-party and out-party frames to be larger in the treatment
condition (Hypothesis 2). The effect of an own-party endorsement among the control group was
non-significant (diff = -.04, 95% CI[-.09, .01], p = .19). The effect of own-party endorsement in
the treatment group was a 7 point increase in support for the legislation (95% CI[-.12, -.02], p =
.001). Similarly, being in the treatment condition when legislation was framed as own-party
endorsed compared to being in the control condition when it was out-party endorsed caused a 6
point increase in support for the GAOA (95% CI[-.11, -.01], p = .007). This indicates that the
endorsement only mattered among participants who had their political identities primed in the
positive experience trust game. As predicted in Hypothesis 2, the positive experience trust game
increased party-aligned motivated reasoning around support for the GAOA.
How the magnitude and direction of these effects might differ between Democrats and
Republicans for each bill is another interest of this experiment, though the difference between
the two parties’ support for the legislation is not. Adding partisanship to the regression model as
part of a three-way interaction rather than a covariate allows for a more detailed interrogation of
the differences between Democrats and Republicans across the other dimensions. Figure 9 plots
the effect of own vs. out-party framing over control vs. treatment on support for the GAOA, with

17
Statistical tests for a difference between treatment and control within the own-party endorsement group for a
given party are also underpowered. For example, post hoc power analysis for a mean comparison between treatment
and control among Democrats resulted in power (1 – β) = .29, which is an approximately 30% chance of detecting
an effect if there is one. This is well below the standard power target of .80.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

78
plots separated by party (Republicans on the left). This three-way interaction model was run for
display purposes only. Because there are no hypotheses related to how members of the two
parties will differ in their opinions of the legislation, linear regression models were run
separately for each party, controlling for age, whenever within party comparisons across the
other dimensions were made; Tukey’s correction for multiple comparisons was utilized in post
hoc analysis.
Because they were not primed with the positive experience trust game, among
Republicans in the control condition, there was no significant difference between the own- and
out-party frames on support for the GAOA (F(4, 323) = 2.03, p = .09, R
2
= .01, diff = .04, 95%
CI[-.12, .04], p = .655). However, among Republicans whose political identity was primed
through the positive experience trust game, the own-party endorsement increased support for the
legislation by 8 points (95% CI[-.14, -.02], p = .01) relative to out-party endorsement. This
contrast remained significant at the 90% confidence level after correction for multiple
comparisons (95% CI[-.16, .001], p = .055). To be clear about this effect, the positive experience
trust game primed Republican political identity. In doing so, it caused Republicans who read
about an uncontroversial own-party endorsed piece of legislation to support it slightly more (71
points) compared to Republicans in the control group (68 points). It also caused the opposite
effect among Republicans who read about the same piece of legislation with an out-party
endorsement, slightly less support in the treatment group (63 points) compared to control (65
points). This increase and decrease were too small to be statistically significant in themselves,
but were enough to show that the positive experience trust game caused support for the party line
to increase, regardless of the implied position taken by the party. When an uncontroversial piece
of legislation was supported by Democrats, Republicans liked it less. When it was supported by
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

79
Republicans, they liked it more, but this difference only mattered if their political identities were
primed through an intervention designed to reduce out-party animus.
For Democrats, the pattern of effects is similar. Among Democrats in the control
condition, there was again no difference between the own- and out-party frames (F(4, 328) =
3.35, p = .01, R
2
= .04, diff = .04, 95% CI[-.10, .02], p = .283). Among Democrats who
participated in the positive experience trust game, the out-party frame caused a 6 point reduction
in support for the GAOA (95% CI[-.12, -.002], p = .038), after correction for multiple
comparisons. For both Democrats and Republicans, the effect of having a positive simulated
interaction with an out-party member was a decrease in support for out-party supported
legislation. This is the opposite of what we would expect to see from a reduction in affective
polarization if affective polarization were indeed the only attitude being manipulated by the
intervention.

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

80

Figure 9: Support for the Great American Outdoors Act by Party and Treatment

Turning to the Gun Violence Prevention Act, the pattern of results was expected to
change due to the topic of the bill. As above, Figure 10 plots predicted means for each group
based on a linear regression with a three-way interaction term (treatment x endorsement x party),
controlling for age (N = 661, F(8, 652) = 39.7, p < .001, R
2
= .24). Comparisons within parties
were again estimated based on independent regression models with Tukey’s correction for
multiple comparisons. Broadly speaking the pattern of results was similar to that found for the
GAOA. Looking first at a model containing all observations while controlling for party
identification (N = 661, F(5, 655) = 60.56, p < .001, R
2
= .24) comparing treatment + own-party
endorsement to control + out-party endorsement resulted in an 11 point difference in support for
the Gun Violence Prevention Act (95% CI[-.17, -.04], p < .001). However, in contrast to the
finding for the less controversial piece of legislation, where differences between the two
.6
.7
.8
.9
C o n t ro l
Treatment
C o n t ro l
Treatment
Republican Democrat
Out-Party Endorsement Own Party Endorsement
Linear Prediction
Control v. Trust Game Treatment
Support for the Great American Outdoors Act
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

81
endorsement conditions increased when contrasted across the treatment group, in this case the
difference shrunk to non-significance overall (95% CI[-.12, .01], p = .179). In another departure
from the findings for the GAOA, for Gun Violence Prevention, exposure to the positive
experience trust game significantly reduce support for the legislation by 4 points (95% CI[-.08, -
.01], p = .01), controlling for endorsement, and party. This effect is driven by Republicans, for
whom participation in the positive experience trust game was associated with a 10 percentage
point decrease in support for the gun violence prevention legislation (95% CI[-.17, -.04], p =
.002), controlling for endorsement framing (N = 328, F(3, 324) = 6.37, p <.001, R
2
= .06). As
predicted in Hypothesis 3, Republicans who had their political identities primed through the
positive experience trust game expressed lower levels of support for the Gun Violence
Prevention Act, regardless of party endorsement.
Examining this effect in more detail, the largest difference among Republicans in support
for gun legislation occurred between the treatment + out-party endorsement condition and the
control + own-party endorsement condition, a 19 point difference (95% CI[-.32, -.07], p < .001).
However, the contrast between the treatment + out-party endorsement condition and the
treatment + own-party endorsement condition narrowed to a non-significant 10 point difference
(95% CI[-.23, .16], p = .113). In other words, unlike the GAOA, for the Gun Violence
Prevention Act, priming party identity for Republicans reduced the difference between own and
out-party endorsement, enough so that there was no longer an detectable effect from the
endorsement framing. Gun laws are an especially “hot” political topic for Republican with strong
social identity implications, as such this finding supports the conclusion that inducing a
cognitively “hot” state, in this case through a controversial piece of legislation, can cause
reactive motivated reasoning. This helps explain how the null results in prior research may have
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

82
been produced. By making priming political identity at the same time as making partisans angry
participants became more likely to reason in a manner that conforms with well-known party
positions. As such, the null results after correction in this case, as well as in the BKW study, are
potentially explained by the same underlying cognitive process.
Democrats in the treatment condition (M = 85%, SD = .14) were no more likely to
support the gun violence legislation than Democrats in the control group (M = .85, SD = .12, p =
.777).

Figure 10: Support for the Gun Violence Prevention Act by Party and Treatment

.4
.6
.8
1
C o n t ro l
Treatment
C o n t ro l
Treatment
Republican Democrat
Out-Party Endorsement Own Party Endorsement
Linear Prediction
Control v. Trust Game Treatment
Support for the Gun Violence Prevention Act
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

83
Experiment 2 Results: Support for Undemocratic Practices
Participants were asked to rate their support for three undemocratic practices with or
without a justification. Hypothesis 4 stated that only participants in the positive experience
condition, relative to control, would increase their support for these practices when provided
with a justification. Because the negative experience trust game was specifically designed to
raise partisan animus and undemocratic practices items directly list harming the other party as a
goal, in the present analysis, data from both Survey 1 is analyzed here to address Hypotheses 5a
and 5b.
Looking first at Survey 2, with observations from both parties pooled a linear model
predicting support for undemocratic practices on the interaction between treatment and receiving
a justification while controlling for party and age showed no effects from the variables of interest
(N = 661, F(5, 655) = 2.89, p = .01, R
2
= .02). Pairwise comparison of means likewise showed
no significant effects for between-group differences. For the version of the trust game including
the positive experience and control conditions, there was no effect on support for undemocratic
practices with or without a justification for them. Apropos of no specific hypothesis or research
question, this model did show that Democrats expressed significantly more support for
undemocratic practices than Republicans (b = .06, 95% CI[.03, .10]), p = .001). This finding
cannot be generalized, but it does defy the commonly held perception that Republicans are more
likely to hold anti-democratic attitudes. Hypothesis 4, stating that the positive experience trust
game, relative to control, would increase support for undemocratic practices, was not supported.
The competing Hypotheses 5a and 5b present the most direct test of the true underlying
mechanism through which the trust game manipulation operates. If Hypotheses 5a is supported,
then the negative experience condition operates as BKW assumed it would in the early drafts of
their manuscript, by causing an increase in affective polarization. However, if Hypothesis 5b is
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

84
supported, then there is support for the conclusion that the negative experience trust game
primed systematic reasoning through anger directed at the experimenter and led participants to
respond in a more considered manner. Figure 11 presents the findings from an linear model with
all possible comparisons using data from Survey 1 (N = 400, F(8, 391) = 1.01, p = .427, R
2
=
.02). As shown, Republicans in the positive experience condition who were provided with a
justification for undemocratic practices increased their support by 13 points relative to
Republicans who had not received a justification (95% CI [.03, .23], p = .014). There were no
significant differences for Democrats. However, in a model including Republicans only (N =
175, F(4, 170) = 1.49, p = .207, R
2
= .03), after correcting for multiple comparisons, this result
was no longer significant at the .05 level (b = .13, 95% CI[-.01, .27], p = .087), though it would
be considered significant at the .1 level. While technically only significant to a 91.3% certainty,
the practical significance of this effect is still notable. Providing a non-substantive justification,
compared to providing no justification, when participants had their political identities primed,
had a striking effect on support for undemocratic practices as shown in Figure 11. The fact that
there was no observable difference between justified and unjustified in the negative experience
condition lends credence to the notion this condition cue systematic processing and reactance to
the (correctly) perceived intentions of the experimenter. While effects from the justification in
the positive experience condition did not remain statistically significant at the conventional level
following the correction, this is more likely due to limited statistical power than the absence of
an effect. For this comparison a total of 83 observations were available, the statistical power
calculated post hoc was 1 – ß = .33, or a 33% chance of finding the observed result if there was
one. As such this interpretation should be accepted preliminarily at this time, awaiting an
experimental manipulation with greater statistical power.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

85

Figure 11: Support for Undemocratic Practices by Party and Treatment

Experiment 3 Results: Identity-Based Motivated Learning
Hypothesis 6 stated that the proportion of correct answers in the refugee housing and
crime covariance detection task would be higher if the fast-and-frugal conclusion was associated
with greater threat to a strongly held political identity. While this might be relevant regardless of
whether or not participants have been primed to interpret the data through the lens of their
political identity, it was hypothesized that the effect would only occur when political identity
was primed. Because the threatening conclusion is hypothesized to differ dependent on political
party, logistic regressions were run separately for Democrats and Republicans.
This analysis utilized logistic regression predicting the proportion of correct answers for
partisans as a function of level of ideological sorting (in standard deviations) interacted with a
.1
.2
.3
.4
Negative
Positive
Negative
Positive
Republican Democrat
Justified Unjustified
Linear Prediction
Negative v. Positive Experience Trust Game
Support for Undemocratic Practices
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

86
binary variable indicating whether the correct answer was an increase in crime in cities with or
without refugee housing. Looking first to Democrats (N = 333, LR-χ
2
(5) = 13.49, p = .019), a
standard deviation increase in ideological sorting was associated with a 29% decrease in the
likelihood of answering correctly when the correct conclusion was that crime increase in cities
that provided refugee housing (OR = 0.41, 95%CI[.22, .77], p = .005), controlling for age and
education. This result is consonant with previous findings (Glinitzer et al., 2021), but extends
them to show that not only prior stated beliefs, but the strength of partisan-ideological sorting
can influence the interpretation of objective information.
Turning to Republicans (N = 328, LR-χ
2
(4) = 3.65, p = .455), no equivalent effect from
sorting was observed (OR = 0.96, 95% CI[0.49, 1.86]). Overall, Republicans selected the
incorrect answer to the question more often than Democrats (an 8.5 percentage point difference,
OR = 1.44, 95% CI[1.05, 1.99], p = .023), however, neither party’s members answered the
question correctly more than half of the time. There was a non-significant negative association
between ideological sorting and obtaining the correct answer (OR = 0.83, 95% CI[0.26, 2.67], p
= .748) controlling for party, age and college education. While not significant in this model, it is
known that tasks such as this tend to be performed differently by individuals with different levels
of numeracy, such that more analytically minded individuals tend to reason their way to the
identity protective conclusions (Kahan, 2017a). Given the lower average level of education
among Republicans in this sample, it may be the case that the null result, on par with selecting at
random, is a consequence of this group being less conversant with mathematics. While
controlling for an education with a dummy variable indicating a bachelor’s degree or higher had
no effect on the contrast between the two parties, this measure may be too imprecise to draw out
differences in numeracy.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

87
Research Question 2 asked how partisan identity priming would interact with partisan-
ideological sorting to effect the proportion of correct answers in the covariance detection task.
Once again, logistic regressions were run separately for Democrats and Republicans. As noted
above, the proportion of Republicans who answered the question correctly was not effected by
which answer was correct. Predicting the proportion of correct answers on the basis of a three-
way interaction between ideological sorting, treatment condition and a dummy for the correct
interpretation of the table while controlling for age and college education also produced no
significant effects among Republicans (N = 328, LR-χ
2
(9) = 8.02, p = .532).

Figure 12: Republicans on Refugee Housing and Crime Covariance Task

For Democrats the picture was notably different. As Democratic party identity strength
increased, Democrats tended to select the correct answer more often, but only when the correct
answer was that crime increased in cities without refugee housing, as described above. However,
-.5
0
.5
1
1.5
-3
-2
-1
0
1
2
3
-3
-2
-1
0
1
2
3
Control Treatment
Crime increased WITHOUT refugee housing Crime increased WITH refugee housing
Pr(Correct)
Strength of Partisan Ideological Sorting (SD)
Republicans on Refugee Housing and Crime
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

88
when the three-way interaction term was included in the logistic regression (N = 333, LR-χ
2
(9) =
22.22, p = .008), only the positive experience (treatment) condition showed the sorting effect
(OR = -1.47, 95% CI[-2.80, -0.16], p = .028, for the three-way interaction term, all else held
constant). For a standard deviation increase in partisan-ideological sorting, Democrats in the
treatment condition were 21% less likely to answer the question correctly when the fast-and-
frugal interpretation was threatening (OR = .27, 95%CI[.07, .97], p = .045). Figure 13 displays
the distribution of correct responses across six standard deviations of sorting, with the mean
centered at zero and covariates held at the means (age and college education). The figure shows
that the lines representing conditions where crime increased with and crime increased without
refugee housing cross at one standard deviation below the mean level of partisan sorting. Weakly
sorted Democrats, those 3 standard deviations below the sorting mean, were significantly more
likely to answer correctly when the correct conclusion was threatening to Democratic ideals and
significantly less likely to answer correctly when the correct conclusion was aligned with
Democratic party ideals (dif. = -.79, χ
2
(1) = 40.41, Sadik Adj. p < .001). This difference remains
significant for democrats 2 standard deviations below mean sorting (dif. = -.53, χ
2
(1) = 14.32,
Sadik Adj. p < .001). The opposite is true at the other end of the sorting scale. One standard
deviation above the mean level of sorting, the estimated average difference between the two
correct answer conditions was 29 percentage points wide (χ
2
(1) = 7.39, Sadik Adj. p = .007). This
grows to an estimated 90 percentage point difference by the third standard deviation above the
mean (χ
2
(1) = 99.86, Sadik Adj. p < .001). All comparisons across the range of sorting for
Democrats in the treatment condition are displayed in Table 7 in the Appendix.
This experiment provides the strongest evidence presented in this study that the so-called
“reduction” in affective polarization was accompanied by the simultaneous priming of politically
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

89
motived reasoning through positive experience trust game. Only among Democrats who had
participated in the trust game did the strength of ideological sorting matter, but among this
group, the effects were large and significant at both ends of the sorting distribution.

Figure 13: Democrats on Refugee Housing and Crime Covariance Task

-.5
0
.5
1
1.5
-3
-2
-1
0
1
2
3
-3
-2
-1
0
1
2
3
Control Treatment
Crime increased WITHOUT refugee housing Crime increased WITH refugee housing
Pr(Correct)
Strength of Partisan Ideological Sorting (SD)
Democrats on Refugee Housing and Crime
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

90

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

91
Chapter 4: Discussion
Lessons Learned
The key insight provided by the preceding experiments is that research conducted with
underdeveloped theory can easily lead to results that only partially represent effects. As political
scientists and communication scholars delve into realms that are the traditional purview of
psychology and behavioral economics, it is vital to build upon the foundations provided by these
fields. It is especially important to articulate the cognitive mechanisms and when possible the
neuroscience underlying hypothesized effects in order to ensure that what we think we are testing
is actually what we are testing. Brookman, Kalla, and Westwood’s (2022) treatment of the
mechanism underlying the supposed reduction in affective polarization was based solely on the
observation that the positive experience trust game reduced out-party animus on feeling
thermometer scores and social distance items. The specific theory they overlooked, identity-
based motivation, comes to us from psychology (Oyserman, 2015b), including work on
stereotype threat (Bargh et al., 1996; Gustafsson & Björklund, 2008), culture as situated
cognition (Oyserman, 2015a), and several works at the juncture between neuro-psychology and
political decision-making (Hackel et al., 2017; Kahan, 2016, 2017b; Kahan et al., 2011; van
Bavel & Pereira, 2018). Indeed, much of the recent work on affective polarization has coalesced
around politically motivated reasoning as the key driver of observed effects (Druckman et al.,
2021).
Broockman, Kalla, and Westwood (2022) claimed that their experiments demonstrate that
affective polarization does not impact political decision-making. Based on the research presented
here and the theory used to develop it, a very different conclusion could be reached, though not a
total contradiction of this finding. The results of these three experiments suggest that
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

92
interventions designed to modulate affective polarization may also produce unforeseen changes
that are only observable when context, priming effects, and the distribution of identity-based
attachments are accounted for. The case made by Fiorina (e.g., Fiorina & Levendusky, 2006),
that political attitudes are not chronically accessible for most Americans, helps explain many of
the findings in this field. Most people do not go around continuously assessing their position on
political issues. Instead, we rely on prototypical exemplars of party members or leaders who
embody party-stereotypical characteristics (Hogg, 2001) and unself-consciously consumed
partisan news media to guide our views (Pennycook & Rand, 2019). Unless one’s social identity
is implicated, by political rhetoric from one of these sources or a simple increase in accessibility
through an experimental manipulation, most people have little reason to process information in a
politically biased manner. However, when one of these sources does raise the salience of a
political identity, the effect is likely to be unobservable without measures designed to detect
changes in information processing. Minimally, experiments on the relationship between social
identity and information processing should account for how effects change when participants are
presented with different cues.
For example, in the legislation framing experiment (Experiment 1), it was shown that the
party endorsement only mattered when political identity had been primed by the positive
experience trust game and the policy was uncontroversial. The positiveness of the positive
experience game is not what caused the observed effects. This is clear from the fact that between
control and treatment support for the out-party endorsed legislation decreased marginally for
both parties, while support for the own-party endorsed legislation increased for Republicans.
However, when the legislation was on a controversial topic, with well-known party positions and
major social-identity implications for Republicans, the positive experience trust game reduced
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

93
support for the legislation among Republicans, regardless of the party endorsement. The
hypothesized reason for this effect, despite the statistically significant reduction in affective
polarization caused by the positive experience condition, is that political identity salience, which
made politically motivated reasoning more accessible, was the primary cognitive attribute
manipulated by the intervention. Reflecting back upon BKW’s experiments, the most obvious
explanation for the null results is that politically motivated reasoning was activated in both the
positive and negative experience trust games thereby washing out any between group
differences.
Another important contribution from this research is the reminder to assess outcomes
across the distribution of important independent variables, especially when populations are
expected to be diverse. A substantial body of research has grown up around the effects of
political identity strength and ideological sorting (Huddy et al., 2017; Mason, 2015, 2016, 2018a;
Mason & Wronski, 2018). In Experiment 3, participants were presented with a covariance
detection task in which obtaining the randomly assigned correct answer required systematic
processing of information and correct answers were randomly assigned to be threatening or non-
threatening to party-aligned political beliefs. One limitation of this experiment, which has been
noted in prior research utilizing the same task (Glinitzer et al., 2021), is that it was not possible
to account for participants’ level of numeracy or cognitive reflectivity. While level of education
was controlled for in the analysis, other work (e.g. Kahan, 2017) has shown that scores on
cognitive reflection tasks and numeracy tasks are associated with the likelihood of correct
answers on similar covariance detection tasks. This may help to explain why Republicans, are
less likely to be college educated and tend to express higher need for closure, selected the fast-
and-frugal (incorrect) answer more often than Democrats. Among Democrats, however, the
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

94
strength of partisan-ideological sorting had an unexpected effect. It was not merely the case that
more sorted individuals were more likely to select party-aligned answers. When, and only when,
Democrats had their political identities primed through the trust game, the least sorted Democrats
actually got the question correct significantly more often when it went against the party-
congenial position. These individuals represent a group of moderates we often perceive as
shrinking in the U.S., those who identify with a political party but do not believe that their
partisan affiliation describes them well. When these individuals had their political identities
made salient, they appeared cued to perceive refugees with a greater degree of skepticism. When
this same group was presented with the scenario in which crime increased more in cities without
refugee housing, they responded correctly far less often. This effect does not appear in the
control group. This finding highlights previous findings that our political parties are far less
homogeneous than we perceive them to be (Ahler & Sood, 2018). It has been shown previously
that partisans tend to “misestimate the ideological extremity and political engagement of typical
out-party members” (Druckman, et al., 2022, p.1), but this is an important lesson for researchers
as well. When we make cross-party comparisons, we are aggregating a large number of
characteristics within each party. Especially when considering how cognition and decision-
making are influenced by the salience of a social identity, it is important to allow for the
possibility of variation in the strength of one’s attachment to the implicated identity. Measures
such as the partisan identity strength scale (Huddy et al., 2017) are valuable tools that can be
usefully applied to other identity categories as well (e.g., Mason & Wronski, 2018).
There are several lessons that can be taken from the covariance detection task
(Experiment 3) in particular. This experiment adds to the growing body of evidence showing that
the way we reason to our political beliefs is deeply rooted in identity. The dominant process
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

95
through which we arrive at our beliefs is by reasoning from who we are to what is acceptable. It
is not always the goal of social movements to persuade members of out-groups or to change the
minds of the opposition. Political organizing based on identity-affiliation can be very powerful
when the goal is to mobilize likeminded citizen to advocate for a cause. However, polarization is
likely to be exacerbated by such approaches, since they often rely on pointing out differences
between an own-party aligned us and an out-party aligned them. In those times when it is
necessary to persuade people with whom we do not share a common political identity based on
the facts, it is vital to avoid increasing the salience of political identity through messaging. The
content of the message only matters if the audience is in a state of mind where they will be
willing to hear and consider it. Our perceptions of objective information can be significantly
altered if there is any sense that accepting a belief could endanger our connection to an identity
group.
What is the effect of Affective Polarization?
The effect of affective polarization, independent of one’s stance on political issues,
remains difficult to parse. It seems obvious that individuals with the strongest issue positions
would also express the highest levels of out-party animus. This prompts the question, are these
two constructs truly separable or are they simply two different measurements of the same
underlying phenomenon? Broockman, Kalla and Westwood attempted to answer this question by
manipulating affective polarization exogenously, i.e., independently of other factors, or so they
thought. They concluded that affective polarization has effects exclusively in the interpersonal
domain, but does not influence political opinions. The results of the present study do not prove
this finding incorrect, but do show that the question of separability is unresolved. Reducing
affective polarization, as measured through feeling thermometer scores, primed partisan
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

96
motivated reasoning and caused participants to express political attitudes in the opposite
direction of that anticipated by BKW. One way to resolve this apparent contradiction, the
perspective taken throughout this study, is to assume that participants responded to the feeling
thermometer questions in a manner biased by demand characteristics of the experiment, but it is
also hypothetically possible that the intervention caused both effects. If this is the case it would
strengthen BKW’s argument, because it would suggest that affective polarization truly is
independent of changes in political attitudes. However, it would still be difficult to argue that the
observed effect is representative of affective polarization as a construct and not the peculiarities
of this particular intervention. It is thus far more reasonable to conclude that demand drove the
significant manipulation check.
Other experiments that utilize intergroup contact or the representation thereof to reduce
affective polarization have been shown to influence political decision-making in the expected
direction (Voelkel et al., 2022). Why not in this case? The most substantive difference between
this set of experiments and other contact-like studies might be specificity. For example,
Brookman and Kalla themselves helped to publicized a method known as “deep canvassing,” in
which canvassers go door-to-door and engage in approximately 10-minute long conversations
with residents. These conversations are structured to utilize perspective taking, listening and
empathy to help change minds about often controversial topics. This approach was shown to be
effective at durably reducing transphobia while also increasing support for a nondiscrimination
law (Broockman & Kalla, 2016). This illustrates that reducing out-group animus can have an
impact on policy attitudes, but it might be the case that such effects are only significant when the
policies in question have a specific connection to the out-group. It may be the case that the
approach of targeting affective polarization as a construct independent of other attitudes is
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

97
simply too broad to be a useful lever on specific attitudes. Similarly, specific interventions
appear to have broader effects. Several of the interventions tested in Voelkel et al.’s (2022)
mega-study were focused on topics other than out-party animus. One of the most effective
interventions for reducing effective polarization among Americans was a video depicting
conversations among people in the United Kingdom. The political issues brought up were
relevant to an American audience, but there was no throughline to animus between Democrats
and Republicans. The following observation may seem absurd at first blush, but it is possible that
designing interventions to reduce affective polarization is a bad way to reduce affective
polarization. This is because such an approach misunderstands the root causes, which the
preponderance of evidence now shows is identity-based motivated reasoning, the effects of
which are exacerbated by partisan-ideological sorting. The best way to reduce affective
polarization is probably piece by piece dispelling misconceptions of the opposing party, which is
an approach that has already shown impressive results (Druckman et al., 2022).
The positive experience trust game reduced affective polarization according to a feeling
thermometer, but what do feeling thermometers really measure? Affective polarization itself a
psychological state. There is no uniquely specialized polarization region of the brain. As BKW
point out, very little theorizing has been done around what specifically is being manipulated
when we see a change in a feeling thermometer score or set of social distance items. Particularly
when it comes to lab-based experiments that presume an underlying cognitive mechanism,
BKWs nulls results are a good demonstration of the maxim, if it isn’t happening in the brain, it
isn’t happening. As currently conceptualized, none of the more commonly used indicators of
affective polarization meaningfully map on to any particular neurological system. In the case of
the trust game manipulation, feeling thermometer scores probably only captured participants’
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

98
acknowledgement of what the experimenter was trying to get them to do. However, in Chapter 2,
I provided a detailed explanation for what likely is going on in the brain when people make
political decisions. When policy information is presented but no particular political identity or
processing style (i.e., systematic or heuristic) has been made salient, the signal value associated
with a political belief as represented in the vmPFC is unlikely to systematically altered across
subjects. In other words, various individuals will bring their idiosyncratic contemplations into
consideration of any policy, but the particular neurological substrate activated is unlikely to be
consistent across trials. However, future fMRI research should look at the activation pattern
when political decisions are being made after political identity has been primed. In particular,
increased activation in the default mode network and subsequent dlPFC modulation of the value
signal in the vmPFC would more explicitly demonstrate the link between identity salience and
politically motivated reasoning. It would likewise be interesting to see if activation in areas such
as the insula (associated with disgust) becomes stronger when individuals are presented with
aversive political information after having their identities primed.
It is a measurable trend in the disposition of the country, and in that sense a potentially
useful diagnostic or dependent variable, but it is probably not a causal factor in the determination
of political attitudes. In this regard, I agree with Broockman, Kalla and Westwood, but based on
a different set of theories and evidence. Identities are themselves contextual and dynamically
constructed (Oyserman, 2015b). When the salience of a political identity is manipulated, one of
the possible effects may be a change in reported out-party animus. It is also important to realize
though, that for most people, political opinions are also contextual and dynamically constructed.
Humans have an extremely well-known and long-documented habit of conforming their beliefs
to contextually relevant social groups (Diehl, 1990; Maeda & Hashimoto, 2020; Tajfel, 1959,
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

99
1974; Tajfel et al., 1971; Tajfel & Turner, 1979). When we reason our way to a position, whether
it be our perceptions of the out-party or our position on a policy, identity-protective goals often
outweigh other factors (Kahan, 2017a). If we accept that out-party animus and policy attitudes
are both ultimately derived from identity-protective cognition, what practical purpose is there in
separating the two? To reiterate, I take the position that affective polarization is a problem in and
of itself, but the feared effects of affective polarization on democratic institutions misunderstands
root causes.
One of the key challenges with experimental research is that it abstracts processes from
real-life experience to such an extent that findings can sometimes have no compelling connection
to the real world. For example, when we encounter information that might sway our political
opinions, what is the channel and framing through which we receive it? For most of us, the
source of information is pre-selected in support of our existing views. This means that we do not
simply hear the facts, but also the justification for the interpretation of the facts that aligns with
our beliefs. In Experiment 2, justifications designed to emulate those voiced by political elites
over partisan news media, were used to test partisans’ willingness to support undemocratic
practices. In line with findings from the psychology of ethical dissonance (Barkan et al., 2015), it
was shown that providing a non-substantive justification increased support for undemocratic
practices, but only in the positive experience condition and only among Republicans. This
provides further evidence that the effect of the positive experience trust game was political
identity priming. It also highlights an often observed asymmetry between Democrats and
Republicans. In every experiment conducted within these surveys, outcomes diverged along
party lines. While there has been much debate over the asymmetrical effects of polarization
(Bakshy et al., 2015; Baron & Jost, 2019; Ditto, Clark, et al., 2019; Ditto, Liu, et al., 2019;
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

100
Enders et al., 2021; Hiaeshutter-Rice et al., 2021)
18
there seems to be a desire among many
researchers in the field to pretend as if members of both political parties are equally polarized. I
take the position that we are all polarized on some wedge issues, but that Republicans have a
well-documented psychological disposition (Baron & Jost, 2019; van Bavel & Pereira, 2018) that
makes them far more prone to appeals to out-group threat than Democrats. Bias is not bipartisan,
as Ditto and others claim. The results of Experiment 2 support this conclusion. Only among
Republicans is a non-substantive justification perceived as sufficient to support undemocratic
practices. This is likely due to several factors, among which are lower levels of educational
attainment and the dominance of political elites in news media consumed by Republicans. The
lesson provided here for future research is that it is essential to assume the existence of this
asymmetry and alter the construction of interventions accordingly.
As for the actual effect of reducing effective polarization, the question remains unsettled.
Perhaps the observed changes in affective polarization over time have led us to reverse the
causality between identity-based processes and political attitudes. The work done to separate
affective polarization from other types of political decision-making in order to see how one
effects the other may in reality be comparing two outcomes from the same underlying process.
While the BKW study does not ultimately prove that affective polarization has no impact on
political decision-making, their hypothesis may still be correct. The present study does not
address whether or not affective polarization influences political decision-making. Instead it
emphasizes the inseparability of identity-based motivations and political information processing.

18
I just want to briefly highly the work of Hiaeshutter‑Rice, Neuner and Soroka (2021) who have a fantastic piece in
Political Behavior that I did not find a good opportunity to discuss. It shows how ostensibly non-political objects are
becoming more partisan and it is absolutely fascinating.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

101
In the final analysis the only theory generally supported is the one developed by Tajfel and
Turner many decades ago.
Limitations and Future Research
The major limitation of the present study is statistical power. This was shown in at least
two cases in which effects shown from simple linear comparisons were no longer significant at
the 95% confidence level once correction for multiple comparisons were applied. Other issues
related to sample size are also important to note. Due to funding constraints, it was not possible
to match the sample utilized by BKW in their experiments. In fact, the panel provider for the
BKW experiments, Dynata, was approached for this study. However, after reviewing the content
of the experiment, a representative at Dynata concluded that the deception involved in the
negative experience trust game was unethical. This may be due to feedback they received on the
previous studies, but that is speculation. More importantly, the piloting for this research led me
to conclude that the negative experience trust game was indeed unethical as BKW presented it.
Randomly assigning participants into a condition that ensured they would receive no bonus
payment might have worked if respondents believed they were playing the game with real
people, but a large proportion of them did not believe this deception. This fact is at the root of
the issue with this intervention –– it is simply unbelievable. As such, while it might be fruitful to
conduct research with a larger, more representative, sample of participants in the negative
experience condition, I cannot recommend that such a project be taken up without serious
contemplation of the ethical ramifications.
The findings around support for undemocratic practices are some of the most intriguing,
but also the most problematic. It is unclear why these effects appeared when the negative
experience condition was included, but not when it was replaced by a control group. The effects
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

102
come entirely from the positive experience trust game. Post hoc analysis was performed pooling
the positive experience conditions together over both surveys and while the direction of the
effect remained, it was not statistically significant. Between the two surveys several changes
were made, including changes to the trust game and the wording of these questions.
Nevertheless, future research should continue to explore the relationship between non-
substantive justifications and political beliefs for a few reasons. In real life we seldom encounter
a political position absent a justification. Particularly for Republicans, these justifications are
often unempirical, promote the agenda of elites, and are designed to stoke fear and anger of
outsiders. Meanwhile, on a survey instrument, participants are usually asked to express their
political opinions without having any justification presented. This may lead to external validity
issues, where survey responses do not reliably reflect the manner in which political opinions are
activated in real-world contexts that matter. Keeping in mind that, aside from a handful of
chronically accessible wedge issues, most people do not have strong political opinions or a
sophisticated ability to analyze policy, in order to predict the effects of something like support
for undemocratic practices, it should be measured in a way that reflects the reality in which those
opinions would be activated.
There is a great deal of research, particularly that produced by Mason (e.g. 2018),
emphasizing identity sorting as a primary cause of affective polarization. However, asking
survey respondent to indicate the strength of their attachment to a social identity is a potentially
unreliable way of measuring it that may also under-represent the effect of being embedded in
identity affirming social networks. Experiment 3 showed that self-reported partisan-ideological
sorting, which is composed of a party identity strength scale and an ideological position item,
can have a significant impact on perceptions of politically relevant information, however, there
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

103
may be less obtrusive and more nuanced measures by which to measure social identity sorting
from a network standpoint. Social identity affiliations and the accessibly of these identities are
likely to be strengthened by network dynamics. For example, research looking to extend the
measurement of social sorting might benefit from asking survey respondent to describe the
demographic and ideological characteristics of their social network. An item asking simply for
the political party affiliation and demographics of a respondent’s closest friends could provide
rich data with which to parse the effects of having more or less homogenous social groups, a
potential proxy for social identity sorting.
Issues with the timing of data collection may have an impact on the results, though how
so is unclear. Data were collected starting in the spring of 2022. However, large modifications
were made, resulting in Survey 2, following the official publication of the BKW study in AJPSA
and the revelation of null effects from the negative experience game compared to control. While
abandoning the approach taken in Survey 1 was ultimately a wise decision, it did mean a
considerable delay in data collection, which pushed the survey period closer to the fall 2022 mid-
term elections. A related issue that compounded this was the fact that Republicans were a much
smaller pool of panel participants than Democrats on the Prolific platform. After excluding all
those Republicans who participated in Survey 1, the rate at an adequate number of Republicans
completed Survey 2 relative to Democrats was very slow. To compensate for this, I recruited
Democrats in smaller batches, spread over the data collection period, to minimize potential
effects from the narrowing of the window between the survey and the election. In spite of this,
Republicans started the survey around 6 days later on average than Democrats, even though all
data were collected between the 18
th
of June and the 9
th
of July for both parties. It is unclear what
effect, if any, this might have had on the results. Researchers who utilize anonymous online
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

104
survey panels should be cautious about how the overrepresentation of Democrats in these panels
may influence both data collection and the distribution of sample characteristics. If instead
setting a party quota for participant recruitment I had simply run data collection until the
required sample size was met, the overwhelming majority of participants would have been
Democrats. Researchers who use these platforms, but do not specifically target proportions of
partisans, will drastically misestimate baseline averages relative to the broader population. The
distribution of sample characteristics in the present study was also not representative of U.S.
voters and had it been possible to recruit a representative sample, this would have been strongly
preferred. This limits the extent to which findings from this study should be considered
representative of Democrats and Republicans broadly. However, since this research is focused
more narrowly on the information processing effects of partisan identity salience, replication of
the core findings is unlikely to be encumbered by sample characteristics.
Considerably more research should be conducted on the topics brought up in this study.
The question of whether or not affective polarization has an effect on political decision-making
is still unresolved, though I suspect this is the wrong question to ask. The growing field of
depolarization research offers a wide array of perspectives and interventions that are ripe for
exploration. While currently dominated by political scientists, this field is inherently
interdisciplinary and an excellent candidate for multi-method, multi-disciplinary collaboration.
As the research on depolarization grows, we are likely to see the establishment of institutions and
collaborative networks that have the potential to be far less siloed than what we see throughout
much of academia. At the same time, some caution is warranted. Multidisciplinary research
should not mean selecting the interesting pieces of theory from one discipline and applying them
out of context in another. Some of the most interesting research currently being conducted is led
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

105
by scholars who do not simply borrow ideas from other disciplines, but seek to master them. This
is especially clear from the emergence of many neuro-hyphenate disciplines, such as neuro-
psychology, neuro-economics, political-neuroscience, etc.. As illustrated here, taking a
piecemeal approach to theory by drawing only from those sources that support our findings or
contribute to the circumscribed conversation occurring within our disciplines can lead to
misconceptions about what is actually going on.
Thank you, David Broockman
Throughout this dissertation I have been tough on BKW, but I want to acknowledge here
that these researchers are worthy exemplars for anyone doing academic research. Much of the
insight I have into the process by which their experiments were conducted is due to pre-
registration of their studies on the Open Science Forum. Broockman is a major proponent of this
group and the pre-registration of academic research generally. He originally made a name for
himself by uncovering the misconduct of a researcher who had fabricated data. I do not think
Broockman, Kalla and Westwood have done anything rising to the level of misconduct here, but
they definitely did changed their initial hypotheses about the trust game intervention in order to
make them fit the data. I followed these experiments, over multiple drafts and pre-prints, for over
a year before the final study was published in the American Journal of Political Science. In that
time, they switch from claiming that the negative experience trust game increased affective
polarization to (once they had run a version of the experiment with a control group) claiming that
the positive experience game reduced it. However, they never actually ran a version of the
experiment with the positive experience condition and a control group. It was after this change
that I decided to change the design of the intervention for Survey 2, in order to fill in that gap.
While this led to a critical review of the previous study, it was taken on very much in the spirit of
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

106
the example and inspiration provided by this group of scholars, who have been diligent in their
efforts to promote transparency and academic integrity.

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

107
References
Adjodah, D., Dinakar, K., Chinazzi, M., Fraiberger, S. P., Pentland, A., Bates, S., Staller, K.,
Vespignani, A., & Bhatt, D. L. (2021). Association between COVID-19 outcomes and mask
mandates, adherence, and attitudes. PLoS ONE, 16(6 June), 1–26.
https://doi.org/10.1371/journal.pone.0252315

Ahler, D. J., & Sood, G. (2018). The parties in our heads: Misperceptions about party
composition and their consequences. Journal of Politics, 80(3), 964–981.
https://doi.org/10.1086/697253

Allport, G. W. (1954). The nature of prejudice. Addison Wesley.

American National Election Studies. (2021a). ANES 2020 Time Series Study Full Release
[dataset and documentation. www.electionstudies.org

American National Election Studies. (2021b). ANES 2020 Time Series Study Full Release
[dataset and documentation]. In [dataset and documentation].
http://www.electionstudies.org/

Bakshy, E., Messing, S., & Adamic, L. A. (2015). Exposure to ideologically diverse news and
opinion on Facebook. Science, 348(6239), 1130–1132.
https://doi.org/10.1126/science.aaa1160

Barbaro, M. (2022, May 16). The Racist Theory Behind So Many Mass Shootings - The New
York Times. The Daily. New York Times.
https://www.nytimes.com/2022/05/16/podcasts/the-daily/buffalo-shooting-replacement-
theory.html

Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of Social Behavior: Direct Effects
of Trait Construct and Stereotype Activation on Action. Journal of Personality and Social
Psychology, 71(2), 230–244. https://doi.org/10.1037/0022-3514.71.2.230

Barkan, R., Ayal, S., & Ariely, D. (2015). Ethical dissonance, justifications, and moral behavior.
Current Opinion in Psychology, 6, 157–161. https://doi.org/10.1016/j.copsyc.2015.08.001

Baron, J., & Jost, J. T. (2019). False Equivalence: Are Liberals and Conservatives in the United
States Equally Biased? Perspectives on Psychological Science, 14(2), 292–303.
https://doi.org/10.1177/1745691618788876

Berg, J., Dickhaut, J., & McCabe, K. (1995). Trust, reciprocity, and social history. In Games and
Economic Behavior (Vol. 10, Issue 1, pp. 122–142).
https://doi.org/10.1006/game.1995.1027

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

108
Bhanji, J. P., & Delgado, M. R. (2014). The social brain and reward: Social information
processing in the human striatum. Wiley Interdisciplinary Reviews: Cognitive Science, 5(1),
61–73. https://doi.org/10.1002/wcs.1266

Bolsen, T., Druckman, J. N., & Cook, F. L. (2014). The Influence of Partisan Motivated
Reasoning on Public Opinion. Political Behavior, 36(2), 235–262.
https://doi.org/10.1007/s11109-013-9238-0

Brehm, J. W., & Brehm, S. S. (1981). Psychological Reactance: A theory of freedom and
control. In Contemporary Psychology (Vol. 27, Issue 12). Academic Press.
https://doi.org/10.1007/BF00290976

Brocas, I., & Carrillo, J. D. (2012). Dual-process theories of decision-making: a selective survey.

Brocas, I., & Carrillo, J. D. (2014). Value computation and value modulation : a dual-process
theory of self-control. Los Angeles, CA, USA: University of Southern California,
Department of Economics.

Broockman, D. E., Kalla, J. L., & Westwood, S. J. (2022). Does Affective Polarization
Undermine Democratic Norms or Accountability? Maybe Not *. American Journal of
Political Science, Forthcoming. https://doi.org/10.7910/DVN/3AOBP2

Broockman, D., & Kalla, J. (2016). Durably reducing transphobia: A field experiment on door-
to-door canvassing. Science, 352(6282), 220–224. https://doi.org/10.1126/science.aad9713

Centola, D. (2010). The spread of behavior in an online social network experiment. Science,
329(5996), 1194–1197. https://doi.org/10.1126/science.1185231

Clark, C. J., Liu, B. S., Winegard, B. M., & Ditto, P. H. (2019). Tribalism Is Human Nature.
Current Directions in Psychological Science, 28(6), 587–592.
https://doi.org/10.1177/0963721419862289

Cohn, A., & Maréchal, M. A. (2016). Priming in economics. In Current Opinion in Psychology
(Vol. 12, pp. 17–21). Elsevier B.V. https://doi.org/10.1016/j.copsyc.2016.04.019

Dias, N., & Lelkes, Y. (2021). The Nature of Affective Polarization: Disentangling Policy
Disagreement from Partisan Identity. American Journal of Political Science, 0(0), 1–16.
https://doi.org/10.1111/ajps.12628

Diehl, M. (1990). The minimal group paradigm: Theoretical explanations and empirical findings.
European Review of Social Psychology, 1(1), 263–292.
https://doi.org/10.1080/14792779108401864

Dillard, J. P., & Shen, L. (2005). On the nature of reactance and its role in persuasive health
communication. Communication Monographs, 72(2), 144–168.
https://doi.org/10.1080/03637750500111815
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

109

Ditto, P. H., Clark, C. J., Liu, B. S., Wojcik, S. P., Chen, E. E., Grady, R. H., Celniker, J. B., &
Zinger, J. F. (2019). Partisan Bias and Its Discontents. Perspectives on Psychological
Science, 14(2), 304–316. https://doi.org/10.1177/1745691618817753

Ditto, P. H., Liu, B. S., Clark, C. J., Wojcik, S. P., Chen, E. E., Grady, R. H., Celniker, J. B., &
Zinger, J. F. (2019). At Least Bias Is Bipartisan: A Meta-Analytic Comparison of Partisan
Bias in Liberals and Conservatives. Perspectives on Psychological Science, 14(2), 273–291.
https://doi.org/10.1177/1745691617746796

Druckman, J., Klar, S., Krupnikov, Y., Levendusky, M., & Ryan, J. B. (2020). The Political
Impact of Affective Polarization : How Partisan Animus Shapes COVID-19 Attitudes. 0–52.

Druckman, J. N., Klar, S., Krupnikov, Y., Levendusky, M., & Ryan, J. B. (2021). Affective
polarization, local contexts and public opinion in America. Nature Human Behaviour, 5(1),
28–38. https://doi.org/10.1038/s41562-020-01012-5

Druckman, J. N., Klar, S., Krupnikov, Y., Levendusky, M., & Ryan, J. B. (2022).
(Mis)estimating Affective Polarization. Journal of Politics, 84(2).
https://doi.org/10.1086/715603

Druckman, J. N., & Levendusky, M. S. (2019). What do we measure when we measure affective
polarization? Public Opinion Quarterly, 83(1), 114–122. https://doi.org/10.1093/poq/nfz003

Dunham, Y., Baron, A. S., & Carey, S. (2011). Consequences of “Minimal” Group Affiliations
in Children. Child Development, 82(3), 793–811. https://doi.org/10.1111/j.1467-
8624.2011.01577.x

Dunn, A., & Cerda, A. (2022, November 17). Anti-corporate sentiment in U.S. is now
widespread in both parties. Pew Research Center. https://www.pewresearch.org/fact-
tank/2022/11/17/anti-corporate-sentiment-in-u-s-is-now-widespread-in-both-parties/

Enders, A. M., Uscinski, J. E., Seelig, M. I., Klofstad, C. A., Wuchty, S., Funchion, J. R.,
Murthi, M. N., Premaratne, K., & Stoler, J. (2021). The Relationship Between Social Media
Use and Beliefs in Conspiracy Theories and Misinformation. Political Behavior,
0123456789, 1–24. https://doi.org/10.1007/s11109-021-09734-6

Fiorina, M., & Levendusky, M. (2006). Disconnected: The Political Class versus the People. In
D. Brady & P. Nivola (Eds.), Red and Blue Nation? Characteristics, Causes, and
Consequences of America’s Polarized Politics (pp. 49–71). Brookings Institution Press and
the Hoover Institution Press.

Fording, R. C., & Schram, S. F. (2017). The cognitive and emotional sources of trump support:
The case of low-information voters. New Political Science, 39(4), 670–686.
https://doi.org/10.1080/07393148.2017.1378295

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

110
Garrett, R. K., Long, J. A., & Jeong, M. S. (2019). From partisan media to misperception:
Affective polarization as mediator. Journal of Communication, 69(5), 490–512.
https://doi.org/10.1093/joc/jqz028

Garrett, R. K., & Stroud, N. J. (2014). Partisan paths to exposure diversity: Differences in pro-
and counterattitudinal news consumption. Journal of Communication, 64(4), 680–701.
https://doi.org/10.1111/jcom.12105

Glinitzer, K., Gummer, T., & Wagner, M. (2021). Learning Facts About Migration: Politically
Motivated Learning of Polarizing Information About Refugees. Political Psychology, 42(6),
1053–1069. https://doi.org/10.1111/pops.12734

Graves, L. (2017). This is Sinclair, “the most dangerous US company you’ve never heard of.”
The Guardian.

Gustafsson, U., & Björklund, F. (2008). Women self-stereotype with feminine stereotypical traits
under stereotype threat. Current Research in Social Psychology, 13(18), 219–231.

Hackel, L. M., Zaki, J., & van Bavel, J. J. (2017). Social identity shapes social valuation:
Evidence from prosocial behavior and vicarious reward. Social Cognitive and Affective
Neuroscience, 12(8), 1219–1228. https://doi.org/10.1093/scan/nsx045

Hare, T. A., Camerer, C. F., & Rangel, A. (2009). Self-Control in Decision-Making Involves
Modulation of the vmPFC Valuation System. Science, 324(5927), 646–648.
https://doi.org/10.1126/science.1168450

Hiaeshutter-Rice, D., Neuner, F. G., & Soroka, S. (2021). Cued by Culture: Political Imagery
and Partisan Evaluations. Political Behavior, 0123456789. https://doi.org/10.1007/s11109-
021-09726-6

Hogg, M. A. (2001). A social identity theory of leadership. Personality and Social Psychology
Review, 5(3), 184–200. https://doi.org/10.1207/S15327957PSPR0503_1

Horton, D., & Wohl, R. R. (1956). Mass Communication and Para-Social Interaction.
Psychiatry, 19(3), 215–229.

Huddy, L., Bankert, A., Huddy, L., & Bankert, A. (2017). Political Partisanship as a Social
Identity. Oxford Research Encyclopedia of Politics, May 2017, 1–23.
https://doi.org/10.1093/acrefore/9780190228637.013.250

Iyengar, S., Lelkes, Y., Levendusky, M., Malhotra, N., & Westwood, S. J. (2019a). The origins
and consequences of affective polarization in the United States. Annual Review of Political
Science, 22, 129–146. https://doi.org/10.1146/annurev-polisci-051117-073034

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

111
Iyengar, S., Lelkes, Y., Levendusky, M., Malhotra, N., & Westwood, S. J. (2019b). The origins
and consequences of affective polarization in the United States. Annual Review of Political
Science, 22, 129–146. https://doi.org/10.1146/annurev-polisci-051117-073034

Iyengar, S., Sood, G., & Lelkes, Y. (2012). Affect, not ideology: A social identity perspective on
polarization. Public Opinion Quarterly, 76(3), 405–431. https://doi.org/10.1093/poq/nfs038

Iyengar, S., & Westwood, S. J. (2015a). Fear and Loathing across Party Lines: New Evidence on
Group Polarization. American Journal of Political Science, 59(3), 690–707.
https://doi.org/10.1111/ajps.12152

Iyengar, S., & Westwood, S. J. (2015b). Fear and Loathing across Party Lines: New Evidence on
Group Polarization. American Journal of Political Science, 59(3), 690–707.
https://doi.org/10.1111/ajps.12152

Kahan, D. M. (2016). The Politically Motivated Reasoning Paradigm, Part 1: What Politically
Motivated Reasoning Is and How to Measure It. Emerging Trends in the Social and
Behavioral Sciences, 1–16. https://doi.org/10.1002/9781118900772.etrds0417

Kahan, D. M. (2017a). Misconceptions, Misinformation, and the Logic of Identity-protective
Cognition. Cultural Cognition Project Working Paper Series No. 164, Yale Law School,
Public Law Research Paper No. 605, Yale Law & Economics Research Paper No. 575.
https://doi.org/10.2139

Kahan, D. M. (2017b). Misinformation and Identity-Protective Cognition (No. 587; John M. Olin
Center for Studies in Law, Economic, and Public Policy, Issue 587).

Kahan, D. M., Jenkins-Smith, H., & Braman, D. (2011). Cultural cognition of scientific
consensus. Journal of Risk Research, 14(2), 147–174.
https://doi.org/10.1080/13669877.2010.511246

Kahan, D. M., Peters, E., Dawson, E. C., & Slovic, P. (2013). Motivated Numeracy and
Enlightened Self-Government. SSRN Electronic Journal.
https://doi.org/10.2139/ssrn.2319992

Kahan, D. M., Peters, E., Wittlin, M., Slovic, P., Ouellette, L. L., Braman, D., & Mandel, G.
(2012). The polarizing impact of science literacy and numeracy on perceived climate
change risks. Nature Climate Change, 2(10), 732–735.
https://doi.org/10.1038/nclimate1547

Kahane, L. H. (2021). Politicizing the Mask: Political, Economic and Demographic Factors
Affecting Mask Wearing Behavior in the USA. Eastern Economic Journal, 47(2), 163–183.
https://doi.org/10.1057/s41302-020-00186-0

Kahneman, D. (2011). Thinking, fast and slow. Macmillan.
How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

112
Khalid, A. (2019). Democrats Used To Talk About “Criminal Immigrants,” So What Changed
The Party? National Public Radio. https://www.npr.org/2019/02/19/694804917/democrats-
used-to-talk-about-criminal-immigrants-so-what-changed-the-party

Kingzette, J., Druckman, J. N., Klar, S., Krupnikov, Y., Levendusky, M., & Ryan, J. B. (2021).
How Affective Polarization Undermines Support for Democratic Norms. Public Opinion
Quarterly, 85(2), 663–677. https://doi.org/10.1093/poq/nfab029

Knobloch-Westerwick, S., & Lavis, S. M. (2017). Selecting Serious or Satirical, Supporting or
Stirring News? Selective Exposure to Partisan versus Mockery News Online Videos.
Journal of Communication, 67(1), 54–81. https://doi.org/10.1111/jcom.12271

Knobloch-Westerwick, S., Mothes, C., & Polavin, N. (2020). Confirmation Bias, Ingroup Bias,
and Negativity Bias in Selective Exposure to Political Information. Communication
Research, 47(1), 104–124. https://doi.org/10.1177/0093650217719596

Lacombe, M. J., Howat, A. J., & Rothschild, J. E. (2019). Gun Ownership as a Social Identity:
Estimating Behavioral and Attitudinal Relationships. Social Science Quarterly, 100(6),
2408–2424. https://doi.org/10.1111/ssqu.12710

Lelkes, Y. (2018). Affective polarization and ideological sorting: A reciprocal, albeit weak,
relationship. Forum (Germany), 16(1), 67–79. https://doi.org/10.1515/for-2018-0005

Lelkes, Y., & Westwood, S. J. (2017). The Limits of Partisan Prejudice. The Journal of Politics,
79(2), 485–501. https://doi.org/10.1086/688223

Levendusky, M. S. (2018). Americans, not partisans: Can priming American national identity
reduce affective polarization? Journal of Politics, 80(1), 59–70.
https://doi.org/10.1086/693987

Lyu, W., & Wehby, G. L. (2020). Community use of face masks and COVID-19: Evidence from
a natural experiment of state mandates in the US. Health Affairs, 39(8), 1419–1425.
https://doi.org/10.1377/hlthaff.2020.00818

Maeda, K., & Hashimoto, H. (2020). Time Pressure and In-group Favoritism in a Minimal Group
Paradigm. Frontiers in Psychology, 11(November), 1–6.
https://doi.org/10.3389/fpsyg.2020.603117

Mangum, M. (2013). The racial underpinnings of party identification and political ideology.
Social Science Quarterly, 94(5), 1222–1244. https://doi.org/10.1111/ssqu.12029

Marder, E., & Goaillard, J. M. (2006). Variability, compensation and homeostasis in neuron and
network function. Nature Reviews Neuroscience, 7(7), 563–574.
https://doi.org/10.1038/nrn1949

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

113
Mason, L. (2015). “I disrespectfully agree”: The differential effects of partisan sorting on social
and issue polarization. American Journal of Political Science, 59(1), 128–145.
https://doi.org/10.1111/ajps.12089

Mason, L. (2016). A cross-cutting calm: How social sorting drives affective polarization. Public
Opinion Quarterly, 80(Specialissue1), 351–377. https://doi.org/10.1093/poq/nfw001

Mason, L. (2018a). Ideologues without issues: The polarizing consequences of ideological
identities. Public Opinion Quarterly, 82(S1), 280–301. https://doi.org/10.1093/poq/nfy005

Mason, L. (2018b). Uncivil agreement: How poltics became our identity. University of Chicago
Press.

Mason, L., & Wronski, J. (2018). One Tribe to Bind Them All: How Our Social Group
Attachments Strengthen Partisanship. Political Psychology, 39, 257–277.
https://doi.org/10.1111/pops.12485

Mason, L., Wronski, J., & Kane, J. v. (2021). Activating Animus: The Uniquely Social Roots of
Trump Support. American Political Science Review, 1–9.
https://doi.org/10.1017/S0003055421000563

McConnell, C., Margalit, Y., Malhotra, N., & Levendusky, M. (2018). The Economic
Consequences of Partisanship in a Polarized Era. American Journal of Political Science,
62(1), 5–18. https://doi.org/10.1111/ajps.12330

Monbiot, G. (2010). The Tea Party movement: deluded and inspired by billionaires. The
Guardian.

Moons, W. G., & MacKie, D. M. (2007). Thinking straight while seeing red: The influence of
anger on information processing. Personality and Social Psychology Bulletin, 33(5), 706–
720. https://doi.org/10.1177/0146167206298566

Niemeyer, R. E. (2013). What Are the Neurological Foundations of Identities and Identity-
Related Processes? In D. D. Franks & J. H. Turner (Eds.), Handbook of Neurosociology
(pp. 149–166). Springer. www.springer.com/series/6055

Nyhan, B., & Reifler, J. (2010). When corrections fail: The persistence of political
misperceptions. Political Behavior, 32(2), 303–330. https://doi.org/10.1007/s11109-010-
9112-2

Oyserman, D. (2015a). Culture as Situated Cognition. Emerging Trends in the Social and
Behavioral Sciences, 1–20. https://doi.org/10.1002/9781118900772.etrds0067

Oyserman, D. (2015b). Identity-Based Motivation. In R. Scott & S. Kosslyn (Eds.), Emerging
Trends in the Social and Behavioral Sciences (pp. 1–11). John Wiley & Sons, Inc.

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

114
Oyserman, D., & Yan, V. X. (n.d.). Making meaning: A culture-as-situated cognition approach
to the consequences of cultural fluency and disfluency. In S. Kitayama & D. Cohen (Eds.),
Handbook of Cultural Psychology. Guilford Press.

Peer, E., Rothschild, D., Gordon, A., Evernden, Z., & Damer, E. (2021). Data quality of
platforms and panels for online behavioral research. Behavior Research Methods.
https://doi.org/10.3758/s13428-021-01694-3

Pennycook, G., & Rand, D. G. (2019). Lazy, not biased: Susceptibility to partisan fake news is
better explained by lack of reasoning than by motivated reasoning. Cognition, 188, 39–50.
https://doi.org/10.1016/j.cognition.2018.06.011

Petty, R. E., & Briñol, P. (2014). The Elaboration Likelihood and Metacognitive Models of
Attitudes. In Dual-Process Theories of the Social Mind (pp. 172–187).

Petty, R. E., & Briñol, P. (2015). Emotion and persuasion: Cognitive and meta-cognitive
processes impact attitudes. Cognition and Emotion, 29(1), 1–26.
https://doi.org/10.1080/02699931.2014.967183

Politico. (2016). Full text: Donald Trump’s speech on fighting terrorism. Politico.
https://www.politico.com/story/2016/08/donald-trump-terrorism-speech-227025

Rangel, A., & Clithero, J. A. (2013). The Computation of Stimulus Values in Simple Choice. In
Neuroeconomics: Decision Making and the Brain: Second Edition. Elsevier Inc.
https://doi.org/10.1016/B978-0-12-416008-8.00008-5

Roccas, S., & Brewer, M. B. (2002). Social Identity Complexity. In Personality and Social
Psychology Review (Vol. 6, Issue 2).

Ruff, C. C., & Fehr, E. (2014). The neurobiology of rewards and values in social decision
making. Nature Reviews Neuroscience, 15(8), 549–562. https://doi.org/10.1038/nrn3776

Santoro, E., & Broockman, D. (2021). The short-term, circumscribed, and conditional effects of
cross-partisan conversation. https://osf.io/km3q2/

Scherr, S., & Müller, P. (2017). How Perceived Persuasive Intent and Reactance Contribute to
Third-Person Perceptions: Evidence from Two Experiments. Mass Communication and
Society, 20(3), 315–335. https://doi.org/10.1080/15205436.2016.1250911

Schneider, W., & Shiffrin, R. M. (1977). Controlled and automatic human information
processing: I. Detection, search, and attention. Psychological Review, 84(1), 1–66.
https://doi.org/10.1037/0033-295X.84.1.1

Shiffrin, R. M., & Schneider, W. (1977). Perceptual Learning and Automatic Attending.
Psychological Review, 84(2), 127–190.

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

115
Stolberg, S. G. (2022, June 27). For Gun Violence Researchers, Bipartisan Bill Is a ‘Glass Half
Full’ - The New York Times. The New York Time.
https://www.nytimes.com/2022/06/26/us/politics/gun-control-research-bipartisan-
bill.html?searchResultPosition=1

Stone, J., Harrison, C. K., & Mottley, J. V. (2012). “Don’t call me a student-athlete”: The effect
of identity priming on stereotype threat for academically engaged African American college
athletes. Basic and Applied Social Psychology, 34(2), 99–106.
https://doi.org/10.1080/01973533.2012.655624

Stroud, N. J. (2010). Polarization and partisan selective exposure. Journal of Communication,
60(3), 556–576. https://doi.org/10.1111/j.1460-2466.2010.01497.x

Stroud, N. J., & Muddiman, A. (2013). Selective exposure, tolerance, and satirical news.
International Journal of Public Opinion Research, 25(3), 271–290.
https://doi.org/10.1093/ijpor/edt013

Taber, C. S., & Lodge, M. (2006). Motivated Skepticism in the Evaluation of Political Beliefs.
American Journal of Political Science, 50(3), 755–769.

Tajfel, H. (1959). Quantitative Judgement in Social Perception. British Journal of Psychology,
50(1), 16–29. https://doi.org/10.1111/j.2044-8295.1959.tb00677.x

Tajfel, H. (1974). Social Identity and intergroup Behavior. Social Science Information, 13(2),
65–93.

Tajfel, H., Billig, M. G., Bundy, R. P., & Flament, C. (1971). Social categorization and
intergroup behaviour. In European Journal of Social Psychology (Vol. 1, Issue 2, pp. 149–
178). https://doi.org/10.1002/ejsp.2420010202

Tajfel, H., & Turner, J. (1979). An integrative theory of inter-group conflict. The Social
Psychology of Intergroup Relations.

Tversky, A., & Kahneman, D. (1973). Availability: A heuristic for judging frequency and
probability. Cognitive Psychology, 5(2), 207–232. https://doi.org/10.1016/0010-
0285(73)90033-9

Tzieropoulos, H. (2013). The Trust Game in neuroscience: A short review. Social Neuroscience,
8(5), 407–416. https://doi.org/10.1080/17470919.2013.832375

Unsworth, K. L., & Fielding, K. S. (2014). It’s political: How the salience of one’s political
identity changes climate change beliefs and policy support. Global Environmental Change,
27(1), 131–137. https://doi.org/10.1016/j.gloenvcha.2014.05.002

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

116
van Bavel, J. J., & Pereira, A. (2018). The Partisan Brain: An Identity-Based Model of Political
Belief. Trends in Cognitive Sciences, 22(3), 213–224.
https://doi.org/10.1016/j.tics.2018.01.004

Voelkel, J. G., Chu, J., Stagnaro, M. N., Mernyk, J. S., Redekopp, C., Pink, S. L., Druchman, J.
N., Rand, D. G., & Willer, R. (2021). Depolarization Interventions and Anti-Democratic
Attitudes. OSF Preprints, April, 5–24. https://osf.io/7evmp

Voelkel, J. G., Stagnaro, M. N., Chu, J., Pink, S., Mernyk, J. S., Redekopp, C., Cashman, M.,
Druckman, J. N., Rand, D. G., & Willer, R. (2022). Megastudy identifying successful
interventions to strengthen Americans’ democratic attitudes.

Vosoughi, S., Roy, D., & Aral, S. (2018). The spread of true and false news online. Science, 359,
1146–1151. http://science.sciencemag.org/

Weber, C., & Klar, S. (2019). Exploring the Psychological Foundations of Ideological and Social
Sorting. Political Psychology, 40(S1), 215–243. https://doi.org/10.1111/pops.12574

Webster, S. W., & Abramowitz, A. I. (2017). The Ideological Foundations of Affective
Polarization in the U.S. Electorate. American Politics Research, 45(4), 621–647.
https://doi.org/10.1177/1532673X17703132

Weiss, J. N., Karma, A., MacLellan, W. R., Deng, M., Rau, C. D., Rees, C. M., Wang, J.,
Wisniewski, N., Eskin, E., Horvath, S., Qu, Z., Wang, Y., & Lusis, A. J. (2012). “Good
Enough Solutions” and the Genetics of Complex Diseases. Circulation Research, 111(4),
493–504. https://doi.org/10.1161/CIRCRESAHA.112.269084

Westen, D., Blagov, P. S., Harenski, K., Kilts, C., & Hamann, S. (2006). Neural Bases of
Motivated Reasoning: An fMRI Study of Emotional Constraints on Partisan Political
Judgment in the 2004 U.S. Presidential Election. Journal of Cognitive Neuroscience,
18(11), 1947–1958. http://mitprc.silverchair.com/jocn/article-
pdf/18/11/1947/1755991/jocn.2006.18.11.1947.pdf

Westwood, S. J., & Peterson, E. (2020). The Inseparability of Race and Partisanship in the
United States. Political Behavior, 0123456789. https://doi.org/10.1007/s11109-020-09648-
9

Wojcieszak, M., de Leeuw, S., Menchen-Trevino, E., Lee, S., Huang-Isherwood, K. M., &
Weeks, B. (2021). No Polarization From Partisan News: Over-Time Evidence From Trace
Data. International Journal of Press/Politics, 1–26.
https://doi.org/10.1177/19401612211047194

Wojcieszak, M., Winter, S., & Yu, X. (2020). Social Norms and Selectivity: Effects of Norms of
Open-Mindedness on Content Selection and Affective Polarization. Mass Communication
and Society, 23(4), 455–483. https://doi.org/10.1080/15205436.2020.1714663

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

117

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

118
Appendix: Tables

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

119
Table 1: Survey 2 Sample Demographics

Control Treatment (Trust Game) Sig.
% or Mean Count % or Mean Count p
Party

Democrat 51% 166 50% 162 .967
Republican 49% 168 50% 165

Race

White 78% 261 83% 270 .153
Latino 10% 35 10% 34 .973
Black 5% 16 5% 16 .951

Religion

Christian 49% 162 46% 150 .498
Atheist or Agnostic 36% 119 31% 103 .261

Locality

Urban 28% 94 26% 86 .594
Rural 11% 38 17% 56 .035*
Suburban 60% 132 50% 185 .308

Age 34.38 334 36.52 327 .027*

Income

$25,000 23% 78 24% 78 .742
$25,000 - $50,000 26% 88 24% 79 .858
$50,001 - $74,999 19% 65 23% 74 .366
$75,000 of Greater 31% 103 29% 96 .678

Education

High School / GED 30% 101 32% 104 .515
Two Year Degree 14% 46 16% 51 .412
Bachelor's Degree 41% 136 38% 124 .382
Graduate Degree 15% 51 15% 48 .893
Note. Significance tests based on Chi-squared test (Party), linear regression (Age), and logistic
and multinomial logistic regression (all else). *<.05, **<.01. ***<.001.

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

120
Table 2: Survey 1 Sample Demographics

Negative Positive Sig.
% or Mean Count % or Mean Count p
Party

Democrat 44% 87 44% 88 .940
Republican 56% 111 56% 114

Race

White 86% 171 85% 172 .728
Latino 8% 15 5% 11 .390
Black 3% 6 5% 10 .332

Religion

Christian 53% 105 52% 105 .833
Atheist or Agnostic 35% 70 35% 71 .966

Locality

Urban 27% 54 25% 51 .617
Rural 16% 32 16% 32 .830
Suburban 57% 112 59% 119 .635

Age 40.47 198 41.64 202 .399

Income

$25,000 18% 35 20% 40 .148
$25,000 - $50,000 23% 46 27% 54 .094
$50,001 - $74,999 23% 46 27% 55 .081
$75,000 of Greater 36% 71 26% 53 .038*

Education

High School / GED 26% 51 27% 54 .705
Two Year Degree 13% 25 16% 33 .307
Bachelor's Degree 40% 80 38% 77 0.341
Graduate Degree 21% 42 19% 38 .822
Note. Significance tests based on Chi-squared test (Party), linear regression (Age), and
logistic and multinomial logistic regression (all else). *<.05, **<.01. ***<.001.

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

121

Table 3: Partisan Demographics (Survey 2)

Republican Democrat Sig.

% or Mean Count % or Mean Count p
Race

White 84% 277 76% 254 .009**
Latino 10% 34 10% 35 .952
Black 2% 6 8% 26 .001**

Religion

Christian 67% 219 28% 93 <.001***
Atheist or Agnostic 16% 51 51% 171 <.001***

Locality

Urban 20% 56 35% 115 <.001***
Rural 18% 205 11% 182 .128
Suburban 63% 58 55% 36 base

Age 34.93 328 35.93 333 .302

Income

$25,000 19% 62 28% 94 .004
$25,000 - $50,000 25% 83 25% 84 .287
$50,001 - $74,999 22% 73 20% 66 .617
$75,000 of Greater 34% 110 27% 89 .057

Education

High School / GED 34% 110 29% 95 .108
Two Year Degree 17% 57 12% 40 .035*
Bachelor's Degree 37% 120 42% 140 .008**
Graduate Degree 13% 41 17% 58 .420
Note. Significance tests based on Chi-squared test (Party), linear regression (Age), and
logistic and multinomial logistic regression (all else). *<.05, **<.01. ***<.001

RUNNING HEAD: How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

Table 4: Great American Outdoors Act (Experiment 1)

Unadjusted Tukey
Party & Comparison Contrast Std. Err. t p 95% CI p 95% CI
Republicans

Control

Out-party vs Own Party Frame -.036 .031 -1.16 .248 -.098 .025 .655 -.117 .045

Treatment

Out-party vs Own Party Frame -.081 .032 -2.55 .011 -.144 -.019 .055 -.164 .001

Democrats

Control

Out-party vs Own Party Frame -.040 .022 -1.78 .075 -.084 .004 .283 -.098 .018

Treatment

Out-party vs Own Party Frame -.061 .023 -2.69 .008 -.106 -.016 .038 -.120 -.002
Note. Pairwise contrasts reporting p-values and confidence intervals using Tukey's correction as well as unadjusted test
statistics based on linear regression analysis controlling for age for hypothesized between group differences. Significant p-
values at <.1 in bold.

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

123
Table 5: Gun Violence Prevention Act (Experiment 1)

Unadjusted Tukey
Party & Comparison Contrast Std. Err. t p 95% CI p 95% CI
Republicans

Control

Out-party vs Own Party Frame -.078 .046 -1.69 .092 -.169 .013 .332 -.197 .041

Treatment

Out-party vs Own Party Frame -.105 .047 -2.25 .025 -.198 -.013 .113 -.226 .016

Democrats

Control

Out-party vs Own Party Frame -.053 .024 -2.24 .025 -.099 -.007 .114 -.114 .008

Treatment

Out-party vs Own Party Frame -.013 .024 -0.53 .596 -.060 .034 .951 -.074 .049
Note. Pairwise contrasts reporting p-values and confidence intervals using Tukey's correction as well as unadjusted test
statistics based on linear regression analysis controlling for age. Only hypothesized between group differences are displayed,
not all possible comparisons. Significant p -values at <.1 in bold.

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

124
Table 6: Support for Undemocratic Practices (Experiment 2)

Unadjusted Tukey
Party & Comparison Contrast Std. Err. t p 95% CI p 95% CI
Republicans

Negative Experience

Justified vs Unjustified -.028 .055 -0.51 .613 -.137 .081 .958 -.172 .116

Positive Experience

Justified vs Unjustified .129 .054 2.37 .019 .021 .236 .087 -.012 .270

Democrats

Negative Experience

Justified vs Unjustified -.021 .046 -0.46 .643 -.111 .069 .967 -.139 .097

Positive Experience

Justified vs Unjustified .033 .045 0.74 .461 -.055 .122 .882 -.083 .150
Note. Data from Survey 1. Pairwise contrasts reporting p-values and confidence intervals using Tukey's correction as well
as unadjusted test statistics based on linear regression analysis controlling for age. Only hypothesized between group
differences are displayed, not all possible comparisons. Significant p-values at <.1 in bold.

How “Reducing” Affective Polarization Increases Politically Motivated Reasoning

125
Table 7: Proportion of Correct Responses and Contrasts for Democrats (Experiment 3)

Condition: Without Refugees With Refugees
SD Sorting Prop. Correct SE Prop. Correct SE
Contrast χ
2
(1) p (Sidak)
-3 .12 .08 .90 .09 -.78 40.41 <.001
-2 .22 .08 .75 .12 -.53 14.32 <.001
-1 .37 .06 .51 .07 -.14 2.23 0.135
Mean 0 .55 .08 .27 .07 .29 7.39 .007
1 .72 .12 .11 .07 .61 20.12 <.001
2 .84 .11 .04 .05 .80 42.59 <.001
3 .92 .09 .02 .02 .90 99.86 <.001
Note. Proportion of correct answers based on estimated marginal means from logistic regression controlling
for age and college education.

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Three text-based approaches to the evolution of political values and attitudes

PDF

The effects of a COVID-19 narrative message on beliefs and intentions among vaccine hesitant parents

PDF

Identity salience and political violence: centering social identity in the study of intrastate conflict

PDF

Navigating political polarization: a group case study of community engagement in the adoption of intersectional ethnic studies in a Southern California K-12 school district

PDF

Navigating political polarization: a group case study of community engagement in the adoption of intersectional ethnic studies in a Southern California K-12 school district

PDF

Democrats vs. Republicans: how Twitter users share and react to political fact-checking messages

PDF

Navigating political polarization: a group case study of community engagement in the adoption of intersectional ethnic studies in a Southern California K–12 school district

PDF

Navigating political polarization: a group case study of community engagement in the adoption of intersectional ethnic studies in a southern California K-12 school district

PDF

The practice of everyday politics: lifestyle and identity as radical activism

PDF

Navigating political polarization: a group case study of community engagement in the adoption of intersectional ethnic studies in a Southern California K–12 school district

PDF

The Sparks Project Dance Studio: strategic integrated planning model

PDF

Feeling polarized: emotional partisan polarization and its reinforcement through campaign advertising

PDF

Signaling identity: how race and gender shape what representatives say online

PDF

Developing and testing a heuristic-systematic model of health decision making: the role of affect, trust, confidence and media influence

PDF

Difficulty-as-importance and difficulty-as-impossibility: unpacking the context-sensitivity and consequences of identity-based inferences from difficulty

PDF

How socioeconomic status influences career planning for college business majors in Hong Kong

PDF

Common ground reasoning for communicative agents

PDF

Can I make the time or is time running out? That depends in part on how I think about difficulty

PDF

The virtue of reasonableness: on the normative epistemic implications of public reason liberalism

PDF

Identity and regional integration: case studies on central Asia and southeast Asia

Asset Metadata

Creator Sparks, Paul L. (author)

Core Title How "reducing" affective polarization increases politically motivated reasoning

School Annenberg School for Communication

Degree Doctor of Philosophy

Degree Program Communication

Degree Conferral Date 2023-05

Publication Date 04/03/2023

Defense Date 01/10/2023

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag affective polarization,experimental methodology,identity-based motivated reasoning,OAI-PMH Harvest,political identity,politically motivated reasoning,priming

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Murphy, Sheila (committee chair), Stevens, Robin (committee member), Young, Lindsay (committee member)

Creator Email paullsparks@gmail.com,psparks@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC112922130

Unique identifier UC112922130

Identifier etd-SparksPaul-11558.pdf (filename)

Legacy Identifier etd-SparksPaul-11558

Document Type Dissertation

Format theses (aat)

Rights Sparks, Paul L.

Internet Media Type application/pdf

Type texts

Source 20230404-usctheses-batch-1015 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu