Physiological Arousal and Political Beliefs
Jonathan Renshon
University of Wisconsin-Madison
Jooa Julia Lee
Harvard University
Dustin Tingley
Harvard University
It is by now well known that political attitudes can be affected by emotions. Most earlier studies have focused
on emotions generated by some political event (e.g., terrorism or increased immigration). However, the methods
used in previous efforts have made it difficult to untangle the various causal pathways that might link emotions
to political beliefs. In contrast, we focus on emotions incidental (i.e., irrelevant) to the decision process,
allowing us to cleanly trace and estimate the effect of experimentally induced anxiety on political beliefs.
Further, we build upon innovative new work that links physiological reactivity (Oxley et al., 2008a; Hatemi,
McDermott, Eaves, Kendler, & Neale, 2013) to attitudes by using skin conductance reactivity as a measure of
emotional arousal. We found that anxiety—generated by a video stimulus—significantly affected physiological
arousal as measured by tonic skin-conductance levels, and that higher physiological reactivity predicted more
anti-immigration attitudes. We show that physiological reactivity mediated the relationship between anxiety and
political attitudes.
KEY WORDS: Experimental methods, emotions, anxiety, skin conductance, immigration, causal mediation, beliefs
A wealth of new research on the formation and change of political beliefs has begun to establish
what pollsters, campaign managers and candidates have long known: emotions lie at the heart of
political campaigns (Glaser & Salovey, 1998; Marmor-Lavie & Weimann, 2006). Campaigns design
television (Brader, 2005) and print (Chang, 2001) ad campaigns to play upon our emotions (Ridout
& Searles, 2011) and candidates’ own emotions “leak out” and affect how they are perceived (Glaser
& Salovey, 1998; Sullivan & Masters, 1988). As the centrality of emotions (and emotional appeals)
in politics has become more generally accepted, the focus has naturally turned to a new question:
How do emotions affect political judgment and behavior?
We follow the path of Brader, Valentino, and Suhay (2008b), Ladd and Lenz (2008), Brader
(2011), Civettini and Redlawsk (2009), Gadarian (2010), and others in focusing on anxiety and its
effects on political beliefs and preferences. Anxiety is attractive to researchers both because of its
strong, documented effects on the judgment and choices of the voting public and because
the manipulation of this emotion figures so prominently in modern political campaigns (Jerit,
2004). However, research on anxiety and political beliefs—whether using large panel datasets,
Political Psychology, Vol. xx, No. xx, 2014
doi: 10.1111/pops.12173
bs_bs_banner
1
0162-895X © 2014 International Society of Political Psychology
Published by Wiley Periodicals, Inc., 350 Main Street, Malden, MA 02148, USA, 9600 Garsington Road, Oxford, OX4 2DQ,
and PO Box 378 Carlton South, 3053 Victoria, Australia
exogenous events, or experimentally varying the type of out-group discussed in mock political
coverage (respectively, Huddy, Feldman, Taber, & Lahav, 2005; Marcus & MacKuen, 1993;
Brader et al., 2008b)—has typically suffered from a common methodological problem: the emotions
in question are generated by political information or events. This makes isolating the effect
of emotions in these studies difficult, since the political information that acts as the stimulus is
likely to affect political attitudes in a variety of ways besides through emotion (such as through
learning or ideology; see Hatemi, Eaves, & McDermott, 2012). Even if the self-reported emotions
used are statistically significant predictors of attitudes, it is still difficult to estimate how much
work is being done by emotions rather than via alternative (and perhaps unobserved) causal pathways.
We trigger anxiety (and measure it using physiological reactivity) using a video stimulus
unrelated to politics in order to cleanly estimate the “pure” effect of anxiety on political beliefs
about immigration.
We contribute in two critical ways to the literature on emotions and politics. Our first contribution,
in the area of measurement, is built upon two complementary methodological innovations.
First, we focus on experimentally induced anxiety that is incidental to the political judgment
being made. This allows us to cleanly identify the effects of anxiety on political beliefs without
any confounding informational effects (such as might occur if the emotion induction contained
politically relevant information). This helps to provide important confirmation of the importance
of anxiety for political beliefs. It also provides important verification for the notion that incidental
emotions may “carry over” to affect even political beliefs (this was established in Small & Lerner
[2008], but it has not been a major focus of the literature on emotions in politics).
This is complemented by our use of skin-conductance reactivity as a physiological measure of
emotional arousal and/or anxiety. Skin-conductance reactivity is best conceived as a generalized
measure of emotional arousal (i.e., emotions other than anxiety might also trigger high levels of
arousal), so we used a video stimulus commonly used in the past to trigger anxiety and conducted a
separate manipulation-check study to verify that other emotions were not inadvertently triggered as
well. This allowed us to use physiological arousal as a measure of anxiety and bypass the more
commonly used self-reported measures of emotion, which are subject to both impression management
on the part of subjects, as well as inaccuracy due to subjects basing their responses on implicit
causal theories rather than “true introspection” (Nisbett & Wilson, 1977).
Our experimental design follows naturally from—and builds upon—recent work, which has
tackled the links between emotions (incidental or otherwise), physiological arousal, and political
beliefs, but often separately from one another. Oxley et al. (2008a) showed that trait differences in
physiological reactivity to environmental threats unrelated to politics can predict political beliefs, but
they did so without implicating emotions as a factor in influencing state physiological arousal.
Halperin, Porat, Tamir, & Gross (2013) and Hatemi et al. (2013) show that emotions (and our ability
to regulate and manage them) may predict political beliefs but without using incidental emotions to
further isolate their impact on attitudes. Small and Lerner (2008) did incorporate the use of incidental
emotions to study their effect on political beliefs, but without the advantages in measurement that
come from using physiological arousal as a proxy for anxiety.
Thus, our second contribution is to synthesize these recent developments through the use of an
experimental design that incorporates and builds upon these important findings. In doing so, we
further develop some of the ideas found in these earlier works and provide additional clarity on the
mechanisms that may underpin those patterns. We also answer the call of Hatemi and McDermott
(2012) to broaden political psychology by using novel methods and research strategies to address
traditional questions (“what accounts for our political beliefs?”), keeping in mind that “not everyone
reacts to the same stimulus the same way.” This last observation suggests the importance of
accounting for heterogeneous treatment effects through causal mediation. We find that our experimental
manipulation induces physiological changes which in turn influence attitudes towards
Renshon et al.2
immigration policy. More specifically, anxiety—the effect of which was transmitted through
increased physiological arousal—increased hostility toward immigrant out-groups.
Emotions and Politics
Incidental Emotions and Carry-Over Effects
We use the appraisal tendency framework (ATF) as an overarching theoretical scaffolding for
our research because it focuses our attention on both the way that emotions enter into the decision
process and the differential effects of specific, discrete emotions (Lerner & Keltner, 2000). The latter
observation is now commonplace, as a focus on “mood” or “valence” can be misleading given that
similarly valenced emotions can have diverging behavioral consequences (Lerner, Gonzalez, Small,
& Fischhoff, 2003; Lerner & Keltner, 2001; Litvak, Lerner, Tiedens, & Shonk, 2010).
On the former issue, current research has suggested a division between emotions that are
integral (i.e., normatively related) to the decision process and emotions that are incidental, or
unrelated, to the decision at hand (Loewenstein & Lerner, 2003). Most research on emotions and
politics has focused on integral emotions, which include both predicted emotional reactions to
certain events (expected emotions) or the actual emotional reactions that are experienced during a
decision process (immediate emotions).
In stark contrast, incidental emotions are those that are unrelated to the decision at hand, but
often “carry-over” to unrelated domains and affect the judgments and decision making of individuals
in critical and often unappreciated ways (Lerner & Keltner, 2000; Lerner, Small, & Loewenstein,
2004). We use these “carry-over” effects as leverage to cleanly measure the pure effects of a
particular emotion (anxiety) on political beliefs. Carry-over effects describe what occurs when
emotions influence subsequent judgments by giving rise to an instinctual predisposition to appraise
future events in line with the key themes associated with the original emotion. As an illustration,
selling one’s stock in a company because the CEO’s reckless spending makes you angry illustrates
an integral emotion; anger is in this case relevant to the decision at hand (buying or selling that
company’s stock). However, if you sell your stock in that same company because a driver cut you off
on your morning commute, then the emotion that is affecting your financial decision making is
irrelevant and is probably best thought of as a form of “mental contamination” (Lerner, Goldberg, &
Tetlock, 1998).
Anxiety, Politics, and Prejudice
We focus in this article on anxiety, which has occupied a prominent place in the literature on
both emotions and decision making as well as its relation to political judgment and behavior (e.g.,
Brader et al., 2008a, 2008b; Gadarian, 2010; Gadarian & Albertson, 2012; Huddy et al., 2005).
Anxiety, under the Appraisal Tendency Framework, is characterized by high uncertainty and lack of
control (Smith & Ellsworth, 1985). In response to environmental signals of a potential threat,
individuals act to reduce feelings of uncertainty and lack of control that were generated by anxiety
(Frijda, Kuipers, & Ter Schure, 1989; Raghunathan et al., 1999). However, because the ATF doesn’t
necessarily suggest useful hypotheses in the domain of political preferences regarding immigration
and feelings toward out-groups, we turn to the empirical literature to help generate hypotheses.
Broadly, there is strong support for the impact of emotions on policy preferences related to
out-groups (Halperin, 2008; Halperin, Canetti-Nisim, & Hirsch-Hoeer, 2009; Reifen Tagar,
Federico, & Halperin, 2011). When there is anxiety that is directed at particular out-groups, that
emotion is a strong predictor of prejudicial beliefs (Stephan, Ybarra, & Bachman, 2006; Voci &
Hewstone, 2003; though situational factors may reverse this relationship, see Spanovic, Lickel,
Physiological Arousal and Political Beliefs 3
Denson, & Petrovic, 2010). There is also some preliminary evidence that specific emotional states
may lead to automatic prejudice toward out-group members even when the emotional state is
incidental to the judgments being made. Along these lines, DeSteno, Dasgupta, Bartlett, & Cajdric
(2004) found that incidental anger induced in the lab created automatic prejudice toward an outgroup,
while sadness and neutrality resulted in no automatic bias. Similarly, Dasgupta, DeSteno,
Williams, & Hunsinger (2009) and Inbar, Pizarro, & Bloom (2012) found that incidental disgust
(e.g., a disgusting smell) increased bias against disgust-relevant groups (such as homosexuals) while
incidental anger increased bias against anger-relevant groups (such as Arabs). Finally and perhaps
most relevantly, Butz and Yogeeswaran (2011) found that economic threats led to more prejudice
against Asian-Americans, mediated by heightened anxiety.
Based on the appraisal-tendency framework and accumulating empirical evidence on the effects
of anxiety and intergroup relations, we hypothesize that incidental anxiety should intervene in the
appraisal process, triggering prejudice toward outgroup members (i.e., immigrants). More specifically,
we predict that physiological (electrodermal) reactivity, as a biological proxy for anxiety,
should mediate the relationship between induced anxiety and attitudes toward immigrants. Such an
outcome would also be consistent with work on negative selective-retrieval bias and mood-dependent
memory, which suggest that subjects who experience heightened anxiety will be likely to recall more
negative experiences with immigrants and interpret ambiguous information in a more negative and
threatening manner (Bower, 1981; Broadbent & Broadbent, 1988; Burke & Mathews, 1992).
Methods
Participants and Procedure
We recruited 138 male1
adults (M age = 22.8, SD = 4.6) from the Cambridge, Massachusetts,
area for a study that took place at the Harvard Decision Science Laboratory. After completing the
consent process, research assistants applied sensors to subjects individually to measure physiological
reactivity. To measure electrodermal reactivity, two electrodes were attached to the palm of the
nondominant hand (Blascovich, Mendes, Vanman, & Dickerson, 2011) and tonic skin-conductance
level (SCL) was recorded continuously throughout the study.
Following the application of sensors and verbal instructions, subjects completed the study at
individual computer terminals. Participants were told that they would answer two unrelated surveys.
Each survey appeared first approximately 50% of the time, and there were no order effects on any of
our outcome variables. At the beginning of each survey, subjects watched a video (2m49s) featuring
relaxing images of beaches and palm trees with calm music in the background to measure physiological
responses at baseline.2
The entire survey was conducted using the Qualtrics online survey
platform.
One common issue in psycho-physiological studies is knowing exactly what a subject is seeing
or doing at any given moment in order to integrate physiological recordings with survey responses
and experimental treatments. In other words, researchers want to use convenient online survey
platforms, such as Qualtrics, but there was no (existing) efficient way to link these to the proprietary
software (sometimes located on a different computer) that typically captures physiological signals.
1
There are important sex differences in how females respond on physiological dimensions to emotionally arousing stimuli
(Kring & Gordon, 1998), and so we focus on males only to reduce the influence of such differences on our statistical
analysis. Focusing on only males eliminates the potential confounding effect of gender that—even with a control variable—
would reduce the efficiency of our estimates. The strategy of focusing on one gender in psychophysiology studies is
common. See, for example, McDermott, Tingley, Cowden, Frazzetto, & Johnson (2009); Dindo and Fowles (2011), which
use male-only samples, and Giuliani, McRae, & Gross (2008), which uses a female-only sample.
2
All videos used in the project may be viewed online at http://jonathanrenshon.net.
Renshon et al.4
To address this issue, we designed a program which allowed the survey software on the subject’s
computer to send event-markers via the computer’s parallel port to the computers recording the
physiological signals. Each event-marker was unique and contained information about what was
happening in the study, allowing us to integrate physiological analysis with a more traditional survey
platform (Qualtrics).
The first part of the survey recorded several demographic variables and some pretreatment
ideology markers (e.g., party identification). Each subject was randomly assigned to read one of two
stories about immigration. This does not figure centrally in our analysis, but it represents an effort to
expose subjects to at least two different immigration frames. The frames are reproduced in online
Appendix B. In our analysis, we show that these frames did not differentially impact the physiological
or preference measures.
Following this, subjects were randomly assigned into one of three potential treatment conditions,
each of which watched a different video.3
Thus, the overall design was 2 (story: Jose/Niko) X
3 (video: Relax/Neutral/Anxiety) between-subjects. In the Relax condition, subjects watched a video
(3m46s) entitled “Crystal Chakra Meditation” in which soothing music is played over visuals of
abstract shapes and colors. In the Neutral condition, subjects watched a video (2m41s) of a screensaver
of abstract shapes (no sound).4
Finally, in the Anxiety condition, subjects watched a clip
(2m32s) from the film “Cliffhanger.” The film depicts a character played by Sylvester Stallone
attempting to rescue a female mountain climber who is dangling over a precipice attached only by
a metal carabiner to a single rope (though in the film the woman does fall to her death, the clip shown
to subjects ends before that occurs). This specific film clip has been demonstrated to be a reliable way
of inducing anxiety (Fredrickson & Branigan, 2003; for more on the general use of film clips to
induce emotional states, see Hubert & de Jong-Meyer, 1990 and for the use of other, related emotion
inductions, such as music, see Zentner, Grandjean, & Scherer, 2008). We also conducted a separate
manipulation check for the film clip, the results of which are described in the next section.
It is worthwhile to reiterate briefly here the advantage attendant in using incidental emotions to
study the impact of emotions on political beliefs. One approach we might take would be to induce
anxiety by asking participants to think about the terrorist attacks of 9/11. The problem with such an
approach is that—while it is likely to generate anxiety—the emotions it triggered would be related
to political beliefs and perceptions about out-groups. A similar problem would occur if we induced
anxiety by asking participants to read a document about globalization and the economy: any anxiety
we measured would be impossible to disentangle from the informational effects of the stimulus. We
bypass this problem by using a stimulus (the “Cliffhanger” video) that is unrelated in any way to the
outcome of interest (immigration preferences/political beliefs), but which nevertheless has been
demonstrated to be a reliable way of inducing anxiety in subjects.
Following the video manipulation, we asked two sets of questions.5
The first set of questions
related to emotions and asked: “Now, moving on, we would like to know how you feel about
increased immigration. The following questions will ask how you feel when you think about the high
levels of immigration to this country.” Subjects were then asked to rate their feelings on a 4-point
scale (“Not at all”; “A little”; “Somewhat”; “Very”) for the following emotions: anxious, proud,
3
This design was used in order to also pilot the application of the parallel encouragement design discussed in Imai, Tingley,
& Yamamoto (2012). As stated above, however, the story frames did not have a measurable impact on our outcome or
mediator measures, perhaps because of changing attitudes towards the immigration groups in question in the time elapsed
since they were used successfully in Brader et al. (2008b).
4
In theory, the additional length of the Relax video might have had some unintended impact on the subjects. In practice, this
condition was dropped in order to more cleanly compare Anxiety to Neutral, which are of roughly equal length.
5
The separate task paradigm is extremely common in the experimental sciences, and subjects rarely guess the true relationship
between the manipulation (video) and the surveys that generally follow (see, e.g., Bodenhausen, Kramer, & Süsser,
1994; Tiedens & Linton, 2001). This is especially true in cases such as this one when the manipulation has no substantive
relevance for the surveys that follow.
Physiological Arousal and Political Beliefs 5
angry, hopeful, worried, and excited. We also asked several questions in which subjects recorded
their level of agreement/disagreement (on a 5-point scale from “Strongly Agree” to “Strongly
Disagree”) with statements relating to immigration (sample statement: “Immigrants should only be
allowed to take jobs that cannot be filled by American workers”). All immigration questions are
borrowed from Brader et al. (2008b, α = .67, see online Appendix A for all items) to maximize
comparability. The experimental procedure is depicted in Figure 1.
While SCL was recorded continuously, we use physiological reactivity during the immigration
questionnaire as our mediator. More detail on our rationale for this is contained in our discussion of
the results.
Physiological Data Collection and Analysis
Emotions, once triggered, are accompanied by physiological changes that occur without our
control and sometimes without our awareness (Safta, Grigore, & Caruntu, 2011). In particular,
emotional arousal can trigger both the sympathetic (SNS) and parasympathetic (PNS) nervous
systems (Blascovich et al., 2011). This property of emotions allows us to measure them directly (as
opposed to self-report measures) using physiological signals. The principal advantage of using
physiological signals is that they are free from demand characteristics or impression management
and more easily interpreted than facial expressions or speech analysis (van den Broek, Janssen,
Westerink, & Healey, 2009).
One common measure, employed here, is tonic skin-conductance level (SCL; distinct from
another common measure, skin-conductance responses, SCR, which measure changes within a short
period of time in response to a particular stimulus).6
In particular, we focus on reactivity (relative to
baseline). SCL variations are determined by the quantity of sweat secreted by the eccrine sweat
glands (Sequeira, Hot, Silvert, & Delplanque, 2009) and have been used as measures of attention and
arousal for over a century (for reviews, see Christie, 1981; Sequeira & Roy, 1997).
In linking the anxiety generated by the film clip to patterns of physiological arousal, there are
two questions to disentangle: (1) what specific emotions, if any, are triggered by the stimulus, and (2)
how fine a measure of emotions is skin conductance? On the former question, we rely on previous
research which has used film clips to induce specific emotions (Gross & Levenson, 1995) and has
6
While common throughout psychology, this is a somewhat novel method for political science, but recent examples can be
found. See, for example, Mutz and Reeves (2005) and Mutz (2007), which use it as a general measure of arousal and do not
link it to a specific emotion.
treatment
(video
stimulus)
outcome
Figure 1. Experimental design.
Renshon et al.6
used this particular film clip to stimulate anxiety (Fredrickson & Branigan, 2003). We conceptualize
anxiety as undirected aversive arousal which is distinguished from similar emotions (such as fear) by
lacking a clear object (Ohman, 2000).7
On the latter question, we recognize that skin conductance and broader patterns of physiological
reactivity are best conceptualized as generalized8
measures of emotional arousal, rather than as
indicators for specific emotions (Hubert & de Jong-Meyer, 1991).9
Thus, while our measure of
physiological arousal is unlikely to allow us to distinguish between different discrete emotions, the
targeted nature of the (previously validated) stimulus gives us confidence that we are truly measuring
the physiological arousal associated with elevated anxiety.
In addition, we conducted a short online experiment in order to serve as an additional manipulation
check. The results (pictured in online Figure 5, Appendix E) further buttress our claim that (1)
the “Cliffhanger” video stimulus induced anxiety (and not other emotions), and (2) the physiological
reactivity that we measured in the main study can be attributed to anxiety in particular and not other
emotions that (in theory, at least) would also increase physiological arousal. Addressing these
potential pitfalls allows us to use a measure of arousal that has significant advantages over traditional
self-report measures, most notably by rendering ineffective any impression management strategies
used by subjects in answering questions about how the video made them feel.
All physiological data was collected using two disposable Biopac (Santa Barbara, CA) electrodes
(Model: EL507) filled with Biopac Skin Conductance Electrode Paste placed on the palms of
the participant’s nondominant hand (the thenar and hypothenar eminences). Data were amplified
using a gain of 25 μΩ and a low-pass filter of 5 Hz. The tonic skin conductance data were then
calculated in microsiemens values using Mindware software EDA module 3.0 (Mindware Technologies,
Gahanna, OH) by research assistants who were blind to both the study hypothesis and
conditions.10
Once output into a time series, SCLs were standardized within each participant to
address potential differences in variability across participants (see Ben-Shakhar, 1985; Bush, Hess,
& Wolford, 1993).11
This was done by calculating the following for each subject: (1) average SCL
during the first baseline video that a subject watched (“baseline SCL”); (2) a measure (“diff SCL”)
that recorded the difference between SCL at any given moment in the study and the average baseline
SCL then transformed into z-scores; (3) after excluding outlier observations that were >±3 SD
(following generally accepted practice of detecting outliers, see Hein, Lamm, Brodbeck, & Singer,
2011), we recalculated the mean, SD and z-scores based only on the nonoutlier observations. This
process ensures that SCL levels are not pulled toward extreme values caused by anomalous spikes in
SCL, though our results hold even if we include these rare instances. This process was repeated for
each individual. It’s important to note that this process merely smoothes outlying SCL readings for
a given time interval but does not drop subjects from the analysis. All analysis thus uses physiological
data in which an observation at ti is equal to the standardized difference, excluding outlier signals,
of SCL at tbaseline and ti. In essence, this allows us to ask: how much did a treatment affect SCLs across
different conditions, controlling for heterogeneous baseline and reactivity levels across individuals?
7
For an excellent review of the differences between fear and anxiety—and the typical conflation of the two in both clinical
work and research—see Sylvers, Lilienfeld, & LaPrairie (2011).
8
One potential issue is that emotional arousal is one of several physiological processes that can affect SCL. In addition to
its association with heightened arousal and anxiety, increased skin conductance may also signal increased cognitive effort
and attention (Frith & Allen, 1983; Pecchinenda & Smith, 1996). It is for this reason that we use a previously validated
video stimulus.
9
There is an ongoing debate as to whether certain physiological patterns distinguish discrete emotions. For more, see
Cacioppo, Berntson, Larsen, Poehlmann, & Ito (2000).
10
We failed to collect physiological data for seven participants due to a computer-software malfunction. Several participants
showed zero reactivity (i.e., no variation at all throughout the study) in their skin conductance levels and were thus excluded
from analysis. About 10% of the population have been found to be nonresponders (Dawson, Schell, & Filion, 2007).
11
Sometimes SCL observations are log-transformed if the distribution is significantly skewed. Baseline (raw SCL) observations
from this study had skewness values <0.5, so this step was unnecessary.
Physiological Arousal and Political Beliefs 7
Results
Effects of Video Manipulations
We first tested whether our video manipulation impacted subjects’ level of physiological
arousal. Figure 2 shows skin conductance reactivity during the video playback, by video condition.
The Anxiety (“Cliffhanger”) video manipulation clearly increased physiological arousal relative to
both the Relax (t(86) = −6.95, p < .00001) and Neutral (t(79) = −4.40, p < .00001) conditions. The
Relax manipulation appeared to depress physiological arousal relative to the Neutral condition,
though this difference was only marginally significant (t(87) = −2.65, p < .10). This same pattern
(Anxiety > Neutral > Relax) persists if we measure only reactivity during the time frame when
subjects answered questions, but the differences are no longer statistically significant.
Though the “Cliffhanger” video has been reliably shown to induce anxiety, one potential
concern is that—at least for some participants—it may be also prime identification with the ingroup
(all the characters in the clip are White). However, the magnitude and direction of the video → SCL
effect was very similar for the three ethnicities from which most of the subjects were drawn
(Caucasian, Asian-American, and African American). An additional issue concerns what type of
anxiety is being primed and whether it is a specific and directed anxiety or a more general, abstract
anxiety (such as that caused by mortality salience, à la the “terror management” research program;
Schimel et al., 1999). Either of these interpretations is consistent with the argument advanced in this
article, though this observation (there is more than one type of anxiety) should impel researchers to
be cautious even when focusing on a single emotion.
−0.50.00.51.01.5
Video Condition
SkinConductanceReactivity
yteixnAlartueNxaleR
04=n14=n84=n
Figure 2. Means of skin-conductance reactivity by video condition. 95% confidence intervals are in blue (online version).
Renshon et al.8
The results in Figure 2 show levels of physiological arousal/reactivity by video condition, thus
pooling results across story conditions. The story manipulation (in effect, differing primes about
immigration) had no obvious effect on physiological arousal, and the patterns depicted in Figure 2
for the video manipulation are consistent across story conditions.12
Following the video induction, we asked participants their feelings with respect to immigration.
Neither the video manipulation nor the differing information primes (Jose/Niko stories) had any
effect on self-reported feeling about immigration.13
This is interesting in its own right, but it is also
suggestive of some potential problems that accompany reliance only on self-reported feelings:
self-reported feelings with respect to immigration did not seem to be impacted by experimentally
induced anxiety, and yet immigration preferences (as shown below) were. This is consistent with
earlier research on the difficulties of relying upon self-reports if in fact subjects do not actually rely
upon true introspection in answering questions, but rather on the basis of implicit causal theories,
such as whether their immigration beliefs should be affected by reading the Jose/Niko stories (see
Nisbett & Wilson, 1977).
Effects of Physiological Arousal on Immigration Preferences
Table 1 displays two linear regression models that comprise the two stages of a causal mediation
model. In Model 1, our measure of skin conductivity is regressed on an indicator variable for
the treatment status (“Anxiety Manipulation”). This model conveys the same information displayed
in the difference in means analysis but will be used in the next subsection when we
conduct a causal mediation analysis. In the second model, immigration preferences are regressed
on skin-conductance reactivity and the experimental manipulations.14
Skin-conductance readings
12
See online Appendix C for physiological arousal across all six conditions.
13
The video manipulation did increase self-reported anxiety, but the effect did not come close to achieving statistical
significance. Full text of these questions is contained in online Appendix D.
14
The questions related to immigration were averaged to create a composite variable which served as our outcome measure.
Higher values indicate greater hostility toward, and suspicion of, immigrants. Principal component analysis revealed a
one-factor solution that explained a substantial amount of the variance (51%; the central factor had an eigenvalue of 2.52
while a possible second factor had a value of 0.82), but results are unchanged using the factor as an outcome measure
instead of the mean.
Table 1. Main Results
DV: (1) SC
Reactivity
(2) Immigration
Preferences
Anxiety Manipulation 0.339+ −0.277
(0.195) (0.178)
Story Condition 0.386*
(0.176)
SC Reactivity while answering questions 0.232*
(0.100)
Constant 0.115 1.983**
(0.137) (0.138)
N 81 81
r2 0.0368 0.120
Note. Model (1) shows the effect of the treatment (anxiety) on physiological reactivity while Model (2) shows the effects
of physiological reactivity on immigration preferences, controlling for the story condition. Both models includes only
Neutral & Anxiety conditions (Relax condition is excluded).
Standard errors in brackets.
+p < 0.10, *p < 0.05, **p < 0.01.
Physiological Arousal and Political Beliefs 9
for this analysis were taken during the time frame when subjects answered the questions relating
to their immigration preferences. Although the analyses described here smooth outlier SCL readings
in the manner described earlier, including outlier SCL readings does not alter any of the
results described here. In this table, the Anxiety condition is compared exclusively to the Neutral
(control) condition, and the Relax condition is dropped.15
As shown, elevated skin-conductance
reactivity is a significant predictor of immigration preferences. As the coefficient is positive and
statistically significant, the implication is that elevated skin-conductance levels predict more antiimmigrant
preferences. Unsurprisingly, self-reported beliefs about immigration (focusing only on
negatively valenced emotions like anxiety) predicted preferences related to immigration in every
condition (Table 4 in online Appendix H). This, by itself is not novel. Yet, in the Anxiety condition,
physiological arousal remained significant even when controlling for self-reported feelings
about immigration. As shown in Model 2 (Table 1), these results hold controlling for the story
condition. Here, the story coefficient is positive and significant, but only after controlling for the
post-story emotion manipulation and SCL.16
Causal Mediation
Next, we must turn to causal mediation analysis. This allows for estimation of an average causal
mediation effect (ACME), which can be interpreted as the expected difference in the outcome when
the mediator took the value it would realize under the treatment condition as opposed to the control
condition, while the treatment status itself is held constant. Of course, we never observe the outcome
in both states, but as long as the assumption of sequential ignorability is satisfied, we are still able
to estimate this effect, which is termed the indirect effect of the treatment upon the outcome (through
the mediator).17
The direct effect of the treatment upon the outcome (average direct effect, or ADE)
is the expected difference in the potential outcome when the treatment is changed but the mediator
is held constant. The average treatment effect (ATE) is the sum of the direct and indirect effects.
Causal mediation analysis was implemented via the mediation package in R (Imai, Keele, Tingley,
& Yamamoto, 2010).18
The results of this analysis are depicted in Figure 3.19
The ACME, or indirect effect of physiological
reactivity, is positive and marginally significant (p = .08), implying that to the extent that the
video induction increased anxiety, it led to more anti-immigrant preferences. The causal pathway
through physiological reactivity is estimated to account for 41% of the total effect. The direct and
15
An alternate way of estimating this is to include all observations and estimate effects of dummy variables for Anxiety and
Relax conditions. These results, as well as models that control for demographic differences (e.g., age, ideology, income,
etc.) are shown in Figure 2 in online Appendix F. The results we discuss here remain unchanged by different estimation
strategies.
16
Within each video condition, the story condition did not significantly influence SCL. The story condition only influenced
immigration preferences in the anxiety video condition.This may be seen more clearly inTable 3 in onlineAppendix G, which
demonstrates clearly that the story frames do not impact either the results themselves or the substantive interpretation.
17
Sequential ignorability has two parts. First, we assume that the treatment (here, a video induction) is ignorable or
statistically independent of potential outcomes or mediators. As in most experimental studies where treatments are
randomly assigned, this is not a problem. Second, we assume that the mediator (here, physiological reactivity) is also
ignorable and can be treated as if it were randomly assigned. This is a much stronger assumption and not guaranteed to hold
even in experimental studies. As suggested by Imai, Keele, & Tingley (2010), we ran sensitivity analyses, and our results
are relatively robust to potentially confounding pretreatment covariates. Results for the sensitivity analyses are depicted in
online Appendix J.
18
The literature on mediation analysis is voluminous. For a background on causal mediation analysis using the framework of
potential outcomes, see Imai, Keele, & Tingley (2010), Imai, Keele, Tingley, & Yamamoto (2011), and Imai, Keele, &
Yamamoto (2010). Because the model is linear regression, causal mediation analysis as implemented by the MEDIATION
package in R gives identical results to the product of coefficients methods used throughout linear structural equation
modeling (LSEM) (MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002).
19
Results are even stronger when controls (ideology, age, income, race, education) are included. These results are depicted
in online Figure 6, Appendix I.
Renshon et al.10
total effects are both negative, but the confidence intervals overlap with 0.20
It is worth noting here
that while we are confident that anxiety was generated by the video stimulus and led to physiological
arousal (which in turn impacted immigration preferences), we cannot say for sure whether other
discrete emotions (such as disgust or excitement) or even physical activity (such as running) which
generate physiological arousal might not have similar effects.
Here, we use physiological reactivity measured during the time when participants answered the
questions related to immigration. We do this to illustrate how emotions generated by the video outlast
the stimulus and bleed over on to—what should be—unrelated judgments about immigration. In fact,
classic research on the related subject of emotion and risk has shown that even mundane newspaper
accounts of events may trigger significant changes in perceptions of unrelated risks (Johnson &
Tversky, 1983). However, one concern might be that it is in fact the immigration questions themselves
that are the cause of the physiological arousal (i.e., our mediator is endogenous to the
treatment). In fact, using physiological reactivity during the video stimulus as our mediator, the
20
Models that allow for an interaction between the video condition and mediator indicate that the ACME was slightly larger
under the “Cliffhanger” condition.
−0.6 −0.4 −0.2 0.0 0.2
Total Effect
Direct Effect
ACME
(Physiological
Reactivity)
Figure 3. Causal mediation plot. Treatment is anxiety video manipulation (compared to neutral condition), Mediator is skin
conductance reactivity when answering immigration questions, Outcome is composite variable of immigration preferences.
Horizontal lines represent 90% confidence intervals for estimates.
Physiological Arousal and Political Beliefs 11
results are strengthened even further. In that analysis, the mediator is statistically significant and
positive (p = .07) and is estimated to account for 80% of the total effect. This is wholly consistent
with earlier works on carry-over effects, which show a slight decay over time (DeSteno et al., 2004).
However, we do not emphasize these results because we believe the strength of the research design
relies to a great extent on showing the carry over of emotions on to unrelated judgements, which
requires empirical demonstration that the physiological effects outlast the original stimulus.
An additional potential concern with our study is that an omitted variable might account for both
physiological arousal and immigration preferences. This would violate the “sequential ignorability”
assumption necessary for mediation analysis (Imai, Keele, & Yamamoto, 2010). It could be that
people who are anti-immigration just tend to have higher average skin-conductance levels. If this
were true, this would be unlikely to bias our results, since our SC measure relies on changes from
baseline and thus accounts for differences in baseline SC levels across participants. If, however, the
omitted confounder causes both physiological reactivity (changes from baseline) and immigration
preferences, then the mediation analysis will be suspect, since it relies on controlling for confounding
variables that cause both the mediator and outcome variables.
We address this concern in three ways. First, online Figure 6 (Appendix I) presents our
mediation analysis controlling for several pretreatment covariates, including ideology, age,
income, race, and education. The results do not change. Second, we included baseline SC as a
control in the mediation analysis. If the omitted variable is also related to baseline differences then
this helps us control for confounding. Including the baseline SC control variable did not change
the mediation results presented above. This corroborates similar patterns of findings by others
(e.g., Oxley et al., 2008b) in suggesting that subjects’ beliefs about politics (and here, immigration)
don’t translate to stable physiological differences that we would observe when taking baseline
measurements.
Finally, we conducted a sensitivity analysis. While the mathematical details are dealt with
elsewhere (Imai, Keele, & Yamamoto, 2010), the basic intuition is that the mediation effect can be
calculated for different magnitudes of a potentially omitted variable’s impact on the mediator and
outcome variables. We present the results in online Figure 7 in Appendix J, which uses a model with
pretreatment covariates. The sensitivity analysis is presented in two ways: (a) one where the omitted
variable (for example, genetically influenced variation in baseline levels of anxiety, see Hatemi et al.,
2013) influences SCL and anti-immigration preferences in the same direction, and (b) one where the
omitted variable influences these variables but in opposite directions. Under the assumption that the
confounder influences SCL and anti-immigration preferences in the same direction, our analysis
suggests that for the positive mediation effect to instead be negative, then the confounder must
explain 40% of the remaining variation in SCL and 20% of remaining variation in anti-immigration
preferences.21
Discussion
In support of our hypotheses, the results of the present study indicate that heightened physiological
reactivity, measured by SCL, mediated the relationship between anxiety and anti-immigration
attitudes. This provides critical evidence that emotions incidental to the decision process can have
important effects on our political beliefs. Our evidence for the role of emotions in political beliefs is
consistent with previous research on how anxiety/fear could lead to negative selective-retrieval bias,
21
If, however, we assume that the unobserved confounder is influencing SCL and anti-immigration preferences in opposite
directions, as in online Figure 7(b), then the sensitivity analysis suggests that the confounder only strengthens our results
as we increase the magnitude of variance in SCL and preferences that it explains. This is indicated clearly by mediation
effects attached to each contour line, which are all positive and grow more positive in magnitude as more variance is
explained.
Renshon et al.12
which makes anxious individuals more likely to interpret ambiguous information in a negative,
threatening manner (Broadbent & Broadbent, 1988; Burke & Mathews, 1992; MacLeod & Cohen,
1993). This is also consistent with earlier work on mood-dependent memory, such Bower (1981), who
found that people recalled experiences and information that was “affectively congruent” with the
mood they were in during recall (see also Lord, Ross, & Lepper [1979] on “biased assimilation”). This
might have led participants who felt anxious and aroused due to the video induction to have
subconsciously interpreted the information on increased immigration as threatening, or recalled more
threatening facts about immigration, thus triggering anti-immigration attitudes.
We did not find evidence for a total or direct effect of anxiety on immigration attitudes. The
former implies a role for other causal processes operating in opposite directions. The presence of
alternative mechanisms that run in different directions to the mechanism of primary interest, which
get folded into the ADE, means that it is always possible to get a total effect equal to zero but nonzero
mediation effects. MacKinnon, Fairchild, & Fritz (2007), MacKinnon, Lockwood, Hoffman, West,
& Sheets (2002), Zhao et al. (2010), Rucker, Preacher, Tormala, & Petty (2011) and others point this
out, and it is commonly known as competitive mediation. Future research should explore the role of
alternative mechanisms.
One explanation for the nonsignificant direct effect is that within our sample there existed
pretreatment covariates that moderated the effect of the treatment (i.e., moderated mediation). We
tested for, but did not find any evidence of, moderation effects due to age or ideology. It’s possible
that in a more expansive sample (one not composed only of younger adult males or one that was not
limited geographically) evidence of a direct effect would be observed, but the technologically
intensive nature of physiological studies precluded us from pursuing this line of inquiry. Such an
explanation would be broadly in line with other recent research suggesting the importance of the
heterogeneity in how individuals respond to powerful emotional stimuli (see, e.g., Mehta, Jones, &
Josephs, 2008). There is also the possibility that differences in the ability to regulate emotions
explain some of the variability in responses to the stimulus (Goldin, McRae, Ramel, & Gross, 2008).
This is a promising area for future research to address.
And while our study focused on a very specific question at the intersection of beliefs, emotions,
and psychophysiology, our results have some relevance for larger questions relating to
political psychology. For example, we demonstrate the practicality of isolating the effects of specific
emotions by studying carry-over effects in order to minimize other confounding factors.
While increasingly common in social psychology, this approach is rarely used in political science,
despite its methodological advantages. We also demonstrate the utility of an integrative approach
that takes into account not just how subjects say they feel, but measures physiological reactivity
in order to better triangulate subjects’ true emotional response to stimuli. Finally, in recent years,
scholars of international political economy have increasingly posited a role for noneconomic
explanations of immigration preferences (e.g., Hainmueller & Hiscox, 2010). To the extent that
these arguments operate through emotional pathways, our research highlights the role of anxiety
in those explanations.
ACKNOWLEDGMENTS
We are especially grateful to the team at the Harvard Decision Science Laboratory for facilitating
this research, including Phil Esterman, Judith Ezike, Dan Zangri, Nicole Ludmir, Gabe
Mansur, Jeanie Nguyen, Jakob Schneider, Jack Schultz, and Amee Xu. We would also like to thank
those who have read and commented on earlier versions of this article, principally Pete Hatemi,
Cindy Kam, Gary Sherman, and SoYon Rim. Correspondence concerning this article should be
addressed to Jonathan Renshon, Department of Political Science, University of Wisconsin-Madison,
110 North Hall, 1050 Bascom Mall, Madison, WI 53706. E-mail: renshon@wisc.edu
Physiological Arousal and Political Beliefs 13
REFERENCES
Ben-Shakhar, G. (1985). Standardization within individuals: A simple method to neutralize individual differences in skin
conductance. Psychophysiology, 22(3), 292–299.
Blascovich, J., Mendes, W., Vanman, E., & Dickerson, S. (2011). Social psychophysiology for social and personality
psychology. Beverly Hills, CA: Sage.
Bodenhausen, G., Kramer, G., & Süsser, K. (1994). Happiness and stereotypic thinking in social judgment. Journal of
personality and social psychology, 66(4), 621–632.
Bower, G. H. (1981). Mood and memory. American psychologist, 36(2), 129–148.
Brader, T. (2005). Striking a responsive chord: How political ads motivate and persuade voters by appealing to emotions.
American Journal of Political Science, 49(2), 388–405.
Brader, T. (2011). The political relevance of emotions: “reassessing” revisited. Political Psychology, 32(2), 337–346.
Brader, T., Valentino, N., & Suhay, E. (2008a). Is it immigration or the immigrants? The emotional influence of groups on
public opinion and political action. American Journal of Political Science, 52(4), 959–978.
Brader, T., Valentino, N., & Suhay, E. (2008b). What triggers public opposition to immigration? Anxiety, group cues, and
immigration threat. American Journal of Political Science, 52(4), 959–978.
Broadbent, D., & Broadbent, M. (1988). Anxiety and attentional bias: State and trait. Cognition & Emotion, 2(3), 165–183.
Burke, M., & Mathews, A. (1992). Autobiographical memory and clinical anxiety. Cognition & Emotion, 6(1), 23–35.
Bush, L., Hess, U., & Wolford, G. (1993). Transformations for within-subject designs: A Monte Carlo investigation.
Psychological Bulletin, 113(3), 566–579.
Butz, D., & Yogeeswaran, K. (2011). A new threat in the air: Macroeconomic threat increases prejudice against AsianAmericans.
Journal of Experimental Social Psychology, 47(1), 22–27.
Cacioppo, J., Berntson, G., Larsen, J., Poehlmann, K., & Ito, T. (2000). The psychophysiology of emotion. In R. Lewis & J.
Haviland-Jones (Eds.), Handbook of Emotions (Vol. 2; pp. 173–191). New York, NY: Guilford.
Chang, C. (2001). The impacts of emotion elicited by print political advertising on candidate evaluation. Media psychology,
3(2), 91–118.
Christie, M. (1981). Electrodermal activity in the 1980s: A review. Journal of the Royal Society of Medicine, 74(8), 616–622.
Civettini, A., & Redlawsk, D. (2009). Voters, emotions, and memory. Political Psychology, 30(1), 125–151.
Dasgupta, N., DeSteno, D., Williams, L., & Hunsinger, M. (2009). Fanning the flames of prejudice: The influence of specific
incidental emotions on implicit prejudice. Emotion, 9(4), 585–591.
Dawson, M. E., Schell, A. M., & Filion, D. L. (2007). The electrodermal system. In J. T. Cacioppo, L. G. Tassinary & G. G.
Berntson (eds.) Handbook of psychophysiology (pp. 200–223). New York, NY: Cambridge University Press.
DeSteno, D., Dasgupta, N., Bartlett, M., & Cajdric, A. (2004). Prejudice from thin air. Psychological Science, 15(5), 319–324.
DeSteno, D., Petty, R., Rucker, D., Wegener, D., & Braverman, J. (2004). Discrete emotions and persuasion: The role of
emotion-induced expectancies. Journal of Personality and Social Psychology; Journal of Personality and Social
Psychology, 86(1), 43–56.
Dindo, L., & Fowles, D. (2011). Dual temperamental risk factors for psychopathic personality: Evidence from self-report and
skin conductance. Journal of Personality and Social Psychology, 100(3), 557–566.
Fredrickson, B., & Branigan, C. (2003). Positive emotions broaden the scope of attention and thought-action repertoires.
Cognition & Emotion, 19(3), 313–332.
Frijda, N. H., Kuipers, P., & Ter Schure, E. (1989). Relations among emotion, appraisal, and emotional action readiness.
Journal of Personality and Social Psychology, 57(2), 212–228.
Frith, C., & Allen, H. (1983). The skin conductance orienting response as an index of attention. Biological Psychology, 17(1),
27–39.
Gadarian, S. (2010). The politics of threat: How terrorism news shapes foreign policy attitudes. Journal of Politics, 72(2),
469–483.
Gadarian, S., & Albertson, B. (2012). Anxious politics: The effect of anxiety of political trust.
Giuliani, N. R., McRae, K., & Gross, J. J. (2008). The up-and down-regulation of amusement: experiential, behavioral, and
autonomic consequences. Emotion, 8(5), 714–719.
Glaser, J., & Salovey, P. (1998). Affect in electoral politics. Personality and Social Psychology Review, 2(3), 156–172.
Goldin, P., McRae, K., Ramel, W., & Gross, J. (2008). The neural bases of emotion regulation: Reappraisal and suppression
of negative emotion. Biological Psychiatry, 63(6), 577–586.
Gross, J., & Levenson, R. (1995). Emotion elicitation using films. Cognition & Emotion, 9(1), 87–108.
Renshon et al.14
Hainmueller, J., & Hiscox, M. (2010). Attitudes toward highly skilled and low skilled immigration: Evidence from a survey
experiment. American Political Science Review, 104(1), 61–84.
Halperin, E. (2008). Group-based hatred in intractable conflict in Israel. Journal of Conflict Resolution, 52(5), 713–736.
Halperin, E., Canetti-Nisim, D., & Hirsch-Hoeer, S. (2009). The central role of group-based hatred as an emotional antecedent
of political intolerance: Evidence from Israel. Political Psychology, 30(1), 93–123.
Halperin, E., Porat, R., Tamir, M., & Gross, J. J. (2013). Can emotion regulation change political attitudes in intractable
conflicts? From the laboratory to the field. Psychological Science, 24(1), 106–111.
Hatemi, P., McDermott, R., Eaves, L., Kendler, K., & Neale, M. (2013). Fear dispositions, attachment and out-group political
preferences. American Journal of Political Science. Forthcoming.
Hatemi, P. K., Eaves, L., & McDermott, R. (2012). It’s the end of ideology as we know it. Journal of Theoretical Politics,
24(3), 345–369.
Hatemi, P. K., & McDermott, R. (2012). Broadening political psychology. Political Psychology, 33(1), 11–25.
Hein, G., Lamm, C., Brodbeck, C., & Singer, T. (2011). Skin conductance response to the pain of others predicts later costly
helping. PloS One, 6(8), e22759.
Hubert, W., & de Jong-Meyer, R. (1990). Psychophysiological response patterns to positive and negative film stimuli.
Biological Psychology, 31(1), 73–93.
Hubert, W., & de Jong-Meyer, R. (1991). Autonomic, neuroendocrine, and subjective responses to emotion-inducing film
stimuli. International Journal of Psychophysiology, 11(2), 131–140.
Huddy, L., Feldman, S., Taber, C., & Lahav, G. (2005). Threat, anxiety, and support of antiterrorism policies. American
Journal of Political Science, 49(3), 593–608.
Imai, K., Keele, L., & Tingley, D. (2010). A general approach to causal mediation analysis. Psychological Methods, 15(4),
309–334.
Imai, K., Keele, L., Tingley, D., & Yamamoto, T. (2010). Causal mediation analysis using R. In H.D. Vinod (ed.) Advances
in Social Science Research Using R (pp. 129–154). New York: Springer
Imai, K., Keele, L., Tingley, D., & Yamamoto, T. (2011). Unpacking the black box of causality: Learning about causal
mechanisms from experimental and observational studies. American Political Science Review, 105(4), 765–789.
Imai, K., Keele, L., & Yamamoto, T. (2010). Identification, inference and sensitivity analysis for causal mediation effects.
Statistical Science, 25(1), 51–71.
Imai, K., Tingley, D., & Yamamoto, T. (2012). Experimental designs for identifying causal mechanisms. Journal of the Royal
Statistical Society-Series A, 176(1), 1–27.
Inbar, Y., Pizarro, D., & Bloom, P. (2012). Disgusting smells cause decreased liking of gay men. Emotion, 12(1), 23–27.
Jerit, J. (2004). Survival of the fittest: Rhetoric during the course of an election campaign. Political Psychology, 25(4),
563–575.
Johnson, E. J., & Tversky, A. (1983). Affect, generalization, and the perception of risk. Journal of personality and social
psychology, 45(1), 20–31.
Kring, A., & Gordon, A. (1998). Sex differences in emotion: Expression, experience, and physiology. Journal of personality
and social psychology, 74(3), 686–703.
Ladd, J., & Lenz, G. (2008). Reassessing the role of anxiety in vote choice. Political Psychology, 29(2), 275–296.
Lerner, J., Goldberg, J., & Tetlock, P. (1998). Sober second thought: The effects of accountability, anger, and authoritarianism
on attributions of responsibility. Personality and Social Psychology Bulletin, 24(6), 563–574.
Lerner, J., Gonzalez, R., Small, D., & Fischhoff, B. (2003). Effects of fear and anger on perceived risks of terrorism.
Psychological Science, 14(2), 144–150.
Lerner, J., & Keltner, D. (2000). Beyond valence: Toward a model of emotion-specific influences on judgment and choice.
Cognition & Emotion, 14(4), 473–493.
Lerner, J., & Keltner, D. (2001). Fear, anger, and risk. Journal of Personality and Social Psychology, 81(1), 146–159.
Lerner, J., Small, D., & Loewenstein, G. (2004). Heart strings and purse strings. Psychological Science, 15(5), 337–341.
Litvak, P., Lerner, J., Tiedens, L., & Shonk, K. (2010). Fuel in the fire: How anger impacts judgment and decision-making.
In M. Potegal, G. Stemmler, & C. Spielberger (Eds.), International handbook of anger: Constituent and concomitant
biological, psychological and social processes (pp. 287–310). New York, NY: Springer.
Loewenstein, G., & Lerner, J. (2003). The role of affect in decision making. In R. Davidson, K. R. Scherer, & H. Goldsmith
(Eds.), Handbook of affective sciences (pp. 619–642). Oxford, UK: Oxford University Press.
Lord, C. G., Ross, L., & Lepper, M. R. (1979). Biased assimilation and attitude polarization: The effects of prior theories on
subsequently considered evidence. Journal of Personality and Social Psychology, 37(11), 2098–2109.
MacKinnon, D., Fairchild, A., & Fritz, M. (2007). Mediation analysis. Annual Review of Psychology, 58, 593–614.
Physiological Arousal and Political Beliefs 15
MacKinnon, D., Lockwood, C., Hoffman, J., West, S., & Sheets, V. (2002). A comparison of methods to test mediation and
other intervening variable effects. Psychological Methods, 7(1), 83–104.
MacLeod, C., & Cohen, I. (1993). Anxiety and the interpretation of ambiguity: A text comprehension study. Journal of
Abnormal Psychology, 102(2), 238.
Marcus, G., & MacKuen, M. (1993). Anxiety, enthusiasm, and the vote: The emotional underpinnings of learning and
involvement during presidential campaigns. American Political Science Review, 87(3), 672–685.
Marmor-Lavie, G., & Weimann, G. (2006). Measuring emotional appeals in Israeli election campaigns. International Journal
of Public Opinion Research, 18(3), 318–339.
McDermott, R., Tingley, D., Cowden, J., Frazzetto, G., & Johnson, D. D. (2009). Monoamine oxidase a gene (maoa) predicts
behavioral aggression following provocation. Proceedings of the National Academy of Sciences, 106(7), 2118–2123.
Mehta, P., Jones, A., & Josephs, R. (2008). The social endocrinology of dominance: Basal testosterone predicts cortisol
changes and behavior following victory and defeat. Journal of Personality and Social Psychology, 94(6), 1078–1093.
Mutz, D. C. (2007). Effects of “in-your-face” television discourse on perceptions of a legitimate opposition. American
Political Science Review, 101(4), 621–635.
Mutz, D. C., & Reeves, B. (2005). The new videomalaise: Effects of televised incivility on political trust. American Political
Science Review, 99(1), 1–15.
Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know: Verbal reports on mental processes. Psychological
review, 84(3), 231–259.
Ohman, A. (2000). Fear and anxiety: Evolutionary, cognitive, and clinical perspectives, volume 2 of Handbook of Emotions,
pages 573–593. New York, NY: Guilford Press.
Oxley, D., Smith, K., Alford, J., Hibbing, M., Miller, J., Scalora, M., Hatemi, P., & Hibbing, J. (2008a). Political attitudes vary
with physiological traits. Science, 321(5896), 1667–1670.
Oxley, D., Smith, K., Alford, J., Hibbing, M., Miller, J., Scalora, M., Hatemi, P., & Hibbing, J. (2008b). Response to J.C.
Gewirtz and B.N. Cuthbert’s e-letter. Published online at www.sciencemag.org on 03 September.
Pecchinenda, A., & Smith, C. (1996). The affective significance of skin conductance activity during a difficult problemsolving
task. Cognition & Emotion, 10(5), 481–504.
Raghunathan, R., Pham, M., et al. (1999). All negative moods are not equal: Motivational influences of anxiety and sadness
on decision making. Organizational Behavior and Human Decision Processes, 79, 56–77.
Reifen Tagar, M., Federico, C. M., & Halperin, E. (2011). The positive effect of negative emotions in protracted conflict: The
case of anger. Journal of Experimental Social Psychology, 47(1), 157–164.
Ridout, T., & Searles, K. (2011). It’s my campaign I’ll cry if I want to: How and when campaigns use emotional appeals.
Political Psychology, 32(3), 439–458.
Rucker, D., Preacher, K., Tormala, Z., & Petty, R. (2011). Mediation analysis in social psychology: Current practices and new
recommendations. Social and Personality Psychology Compass, 5(6), 359–371.
Safta, I., Grigore, O., & Caruntu, C. (2011). Emotion detection using psycho-physiological signal processing. In 7th
International Symposium on Advanced Topics in Electrical Engineering (ATEE) (pp. 1–4). IEEE.
Schimel, J., Simon, L., Greenberg, J., Pyszczynski, T., Solomon, S., Waxmonsky, J., & Arndt, J. (1999). Stereotypes and terror
management: Evidence that mortality salience enhances stereotypic thinking and preferences. Journal of Personality
and Social Psychology, 77(5), 905–926.
Sequeira, H., Hot, P., Silvert, L., & Delplanque, S. (2009). Electrical autonomic correlates of emotion. International Journal
of Psychophysiology, 71(1), 50–56.
Sequeira, H., & Roy, J. (1997). Neural control of electrodermal activity. In D. Jordan (Ed.), Central nervous control of
autonomic function (Vol. 11, pp. 259–293). Amsterdam, Netherlands: Harwood Academic.
Small, D. A., & Lerner, J. S. (2008). Emotional policy: Personal sadness and anger shape judgments about a welfare case.
Political Psychology, 29(2), 149–168.
Smith, C., & Ellsworth, P. (1985). Patterns of cognitive appraisal in emotion. Journal of Personality and Social Psychology,
48(4), 813–838.
Spanovic, M., Lickel, B., Denson, T. F., & Petrovic, N. (2010). Fear and anger as predictors of motivation for intergroup
aggression: Evidence from Serbia and Republika Srpska. Group Processes & Intergroup Relations, 13(6), 725–739.
Stephan, W. G., Ybarra, O., & Bachman, G. (2006). Prejudice toward immigrants. Journal of Applied Social Psychology,
29(11), 2221–2237.
Sullivan, D., & Masters, R. (1988). “Happy warriors”: Leaders’ facial displays, viewers’ emotions, and political support.
American Journal of Political Science, 32(2), 345–368.
Sylvers, P., Lilienfeld, S., & LaPrairie, J. (2011). Differences between trait fear and trait anxiety: Implications for psychopathology.
Clinical Psychology Review, 31(1), 122–137.
Renshon et al.16
Tiedens, L., & Linton, S. (2001). Judgment under emotional certainty and uncertainty: The effects of specific emotions on
information processing. Journal of Personality and Social Psychology, 81(6), 973–988.
van den Broek, E. L., Janssen, J. H., Westerink, J. H. D. M., & Healey, J. A. (2009). Prerequisites for affective signal
processing (asp). In P. Encarnacao & A. Veloso (Eds.), International Conference on Bio-Inspired Systems and Signal
Processing, Biosignals 2009 (pp. 426–433). Lisbon, Portugal: INSTICC Press.
Voci, A., & Hewstone, M. (2003). Intergroup contact and prejudice toward immigrants in Italy: The mediational role of
anxiety and the moderational role of group salience. Group Processes & Intergroup Relations, 6(1), 37–54.
Zentner, M., Grandjean, D., & Scherer, K. R. (2008). Emotions evoked by the sound of music: characterization, classification,
and measurement. Emotion, 8(4), 494–521.
Zhao, X., Jr., Lynch, J. G., & Chen, Q. (2010). Reconsidering Baron and Kenny: Myths and truths about mediation analysis.
Journal of Consumer Research, 37(2), 197–206.
Supporting Information
Additional Supporting Information may be found in the online version of this article at the publisher’s
web-site:
A Immigration Questions
B Immigration Story Frames
C EDA by Treatment Condition
D Self-Reported Feelings About Immigration
E Manipulation Check
F Alternate Main Results Table 1
G Alternate Main Results Table 2
H Effects of Physiological Reactivity on Immigration Preferences, Controlling for SelfReported
Immigration Beliefs
I Mediation Results with Controls
J Sensitivity Results for Mediation with Controls
Physiological Arousal and Political Beliefs 17