psychology project

Article

Can Threat Increase Support for Liberalism?
New Insights Into the Relationship Between
Threat and Political Attitudes

Fade R. Eadeh
1

and Katharine K. Chang
2,3

Abstract

The extant literature demonstrates that exposure to threat almost always increases support for political conservatism. But can
threat increase the support for political liberalism? The current article provides evidence that threat can increase the aspects of
political liberalism. Across three experiments, we find that experimentally manipulated threats to health-care access (Experiment
1, N ¼ 558), pollution (Experiment 2, N ¼ 184), and corporate misconduct (Experiment 3, N ¼ 225) produced increased support
for components of liberalism. These findings fill a notable gap in the literature, broadening larger theoretical discussions of threat
as a psychological construct and current understandings of experimentally manipulated attitudinal change.

Keywords

threat, emotion, attitudes, political ideology

Threat: A person or thing likely to cause damage or danger.

Oxford English Dictionary (2018)

Threats come in all shapes and sizes. Some are acts of nature

(e.g., tornados), some are caused by human error (e.g., fire

caused by improper wiring), and others reflect malevolence

(e.g., burglars breaking into your home). However, nearly all

threats share one element: it motivates people to seek solutions

to the problem at hand. In some cases, one can achieve this

solution on one’s own, but other threats require collective help.

In the latter case, we often call on folks who are perceived as

effective in dealing with that particular threat. When there is

a fire in the basement, for example, one calls the fire depart-

ment (see Lambert, Eadeh, & Hanson, in press).

“Issue Ownership” in Politics

These considerations are relevant to the concept of issue own-

ership in political science (Egan, 2013; Petrocik, 1996). The

idea is that different political groups are perceived as being

effective at handling particular problems. In the United States,

conservative parties (i.e., Republicans) are perceived as more

effective at dealing with terrorism (Newport, 2014), whereas

liberal parties (Democrats) are perceived as better handling

health care and environmental issues (Saad, 2007). Across cul-

tures, liberal parties have advantages in dealing with the envi-

ronment, social welfare policies, and health care. In contrast,

conservatives have perceptual advantages in issues pertaining

to law and order (Seeberg, 2017).

An “Ideology–Affordance” Framework of
Threat and Political Attitudes

These considerations provide a foundation for a heuristically

useful framework for understanding why certain threats

enhance the attractiveness of conservativism or liberalism. Our

framework is rooted in two assumptions. First, we assume

political parties are (stereotypically) perceived as better han-

dling certain problems. The second assumption is that priming

a particular threat enhances the appeal of whichever political

entity (or beliefs) best seen as “fixing” that threat. This latter

assumption reflects an extension of the “affordance” concept

(Gibson, 1977), which is defined as something that facilitates

goal attainment. Affordances in our environment “are what it

offers the animal, what it provides or furnishes” (Gibson,

1977, p. 67, emphases in original). Many things serve as affor-

dances. Umbrellas afford staying dry; blankets afford warmth.

It is not difficult to see how this construct explains how context

can shape the evaluation of a given stimulus.

Political ideologies, too, represent affordances. Being faced

with terrorist threat may enhance aspects of conservatism. In

1
Goizueta School, Emory University, Atlanta, GA, USA

2
National Institute of Mental Health, MD, USA

3 Washington University, St. Louis, MO, USA

Corresponding Author:

Fade R. Eadeh, Goizueta School, Emory University, 1300 Clifton

Boulevard, Atlanta, GA 30322, USA.

Email: [email protected]

Social Psychological and
Personality Science
2020, Vol. 11(1) 88-96
ª The Author(s) 2019
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/1948550618815919
journals.sagepub.com/home/spp

mailto:[email protected]

https://sagepub.com/journals-permissions

https://doi.org/10.1177/1948550618815919

http://journals.sagepub.com/home/spp

particular, terrorist threat is likely to enhance the appeal of

hawkish policies, which represents one aspect of conservatism

within and outside of the United States. Our framework also

provides for the possibility that threat could enhance the appeal

of liberalism. For this to occur, the threat must correspond to

issues for which liberals are seen as effective affordances.

Relevance to Research on Threat and
Political Attitudes

At this point, it is useful to consider the kinds of threat research-

ers have studied. Because this literature is vast, we draw on a

comprehensive review by Jost, Stern, Rule, and Sterling

(2017), which analyzed 55 samples examining the impact of

threat on political attitudes. Of these, a vast majority (n ¼
45, 82%) focused on terrorism. The remaining samples mostly
focused on xenophobic threats or the threat of physical harm

from muggings or burglary. The extent of these and other find-

ings allowed Jost and colleagues (2017) to conclude that “In

light of these forces, it seems to be a remarkable fact of social

and political psychology that subjective feelings and objective

exposure to fearful and threatening stimuli contribute to obser-

vable ‘conservative shifts’ more often than not” (pp. 344–345).

Notably, these primes represent threats for which conserva-

tives are perceived as more effective agents (Seeberg, 2017).

According to our framework, these threats should enhance the

appeal of conservative values. However, our model makes a

more specific prediction: When these threats occur, they

should most reliably pertain to aspects of conservatism that

represent a solution to the threat. This is in fact the case

(Eadeh, 2017; Lambert et al., 2010), as participants primed

with terrorist threat find hawkish, “get tough” policies more

appealing.

More important, our framework also explains the absence

of studies showing threat can produce more favorable

appraisals of liberalism. As noted above, many of the previous

studies have focused on the kinds of threats for which conser-

vatives are perceived as effective affordances. However, there

is a wide diversity of threats in the universe. In short, we do

not believe there is something “intrinsic” to threat predispos-

ing people to enhance the value of conservatism. Rather, the

critical issue is the type of threat examined. We are not aware

of any research that studies threats for which liberals are

seen as effective affordances. Our framework predicts

priming these threats enhance the appeal of liberal—but not

conservative—ideologies. The goal of our research was to fill

this gap in the literature.

Considerations of Affective Mediation

The goal of the current article is to demonstrate threat-driven

shifts to the left using three different types of threat. Of

course, the activation of threat evokes a diversity of negative

emotions including anger, fear, and anxiety (Huddy, Feldman,

& Cassese, 2007; Lambert et al., 2010; 2014). Although both

anger and fear-related processes may be activated during such

events, we predicted anger to be the causal mechanism for

attitude change. This is because anger, unlike fear, is part

of the approach system (Carver & Harmon-Jones, 2009).

More important for our purposes, anger, unlike fear, is the

emotion that makes us take action and resolve the perceived

threat (Cottrell & Neuberg, 2005; Litvak, Lerner, Tiedens,

& Shonk, 2010).

In order to maximize the comparison of our results across

studies, we performed three separate principal components

analyses (PCAs) for each study with varimax rotation on the

entire mood inventory. Analyses across studies were virtually

identical. A primary component emerged reflecting an “anger”

dimension, containing extremely high loadings on several

anger-related emotions (e.g., angry, mad) along with mood

items corresponding to moral outrage (e.g., disgusted,

revulsed). This finding is in line with literature suggesting that,

in cases involving moral violations, anger and disgust reflect

similar action tendencies (Hutcherson & Gross, 2011; Nabi,

2002; see also Giner-Sorolla, Kupfer, & Sabo, 2018). The var-

iance accounted for this primary component was relatively high

(36.31% in Experiment 1, 26.50% in Experiment 2, and
33.50% in Experiment 3). Beyond anger, a component repre-
senting fear/anxiety also emerged (17.14% in Experiment 1,
14.57% in Experiment 2, and 15.15% in Experiment 3).

In light of this convergence, a generalized anger composite

was formed based on an average of 6 items that consistently

loaded most highly across all three studies (i.e., mad, angry,

furious, disgusted, revulsed, sickened; as ¼ .97, .97, and .96,
respectively). In addition, we formed a fear/anxiety composite

based on the average of 6 items that highly loaded on this com-

ponent (nervous, scared, afraid, fearful, anxious, worried; as ¼
.93 across all three experiments).

1

Preliminary Study

Before turning our attention to the main experiments, it was

important to determine whether the primes presented here were

equally threatening to other threats studied in the literature

(e.g., terrorism). To this end, a separate group of participants

took part in a preregistered study (https://osf.io/7waf5/). Parti-

cipants were randomly assigned to one of the five priming con-

ditions: the health-care prime (Experiment 1), the water

pollution prime (Experiment 2), the corporate misconduct

prime (Experiment 3), the threat of terrorism (see Online Sup-

plement [Appendix A]), or a neutral control (Experiments 1–3).

After condition assignment, participants completed two coun-

terbalanced blocks of questions. The first block included a stan-

dard mood inventory, and a second block included a 14-item

perceived threat questionnaire. For both dependent measures,

we predicted that participants in the experimental primes would

report higher levels of negative affect and greater levels of per-

ceived threat, compared to control. Moreover, we predicted

there would be no differences between the threat primes, in either

affective experience or perceived threat. As seen in Figure 1A and

1B, these are precisely what the data show. Indeed, there were no

significant differences in perceived threat, regardless of how it

Eadeh and Chang 89

https://osf.io/7waf5/

was operationalized (all ps > .25) nor were there differences in

affective experience (all ps > .09). The design, materials, and

summary of the findings are available in Online Supplement

(Appendix E).

Experiment 1

Participants in this experiment read about a case where a

young child was denied health insurance or a neutral control

condition. Participants next completed a mood inventory, fol-

lowed by an evaluation of political attitudes including atti-

tudes toward health care. We also examined attitudes

toward social conservatism, hawkish military attitudes, and

general liberal attitudes.

Method

Participants and Design

Five hundred fifty-eight participants (171 male, 381 female, 2

other, 4 did not answer) were recruited from Amazon

Mechanical Turk (MTurk). The design included a between-

subjects priming manipulation (health-care threat vs. control),

followed by a standard mood inventory and a series of polit-

ical statements.

In all experiments, (a) sample size was determined in

advance, (b) aiming for at least n ¼ 100 per each experimental
cell (before any data exclusions) to achieve suitable statistical

power. Since no previous studies examined the link between

threat and liberalism, we estimated a modest effect of Cohen’s

f ¼ .20. This effect size indicated that 200 participants were
needed to achieve power of 0.80. Moreover, data from Experi-

ment 1 represented all experimental manipulations of

health-care threat (vs. control) in the first author’s doctoral

dissertation, with data collection taking place in fall 2016

and winter 2017, with no evidence the effects differed by wave.

All participants were MTurk workers who completed between

100 and 500 HITs and had at least an 80% approval rating; each
participant completed only one of our studies.

2

We had three exclusion criteria for removing participants.

First, we excluded noncitizens of the United States, as they may

not possess knowledge about some political questions (e.g.,

Medicare). Noncitizens (about 2% in each sample) were paid
but not included in the final sample. Second, participants com-

pleted an open-ended writing task, which provided an opportu-

nity to screen participants who displayed disinterest by

submitting nonsense characters or unrelated answers (e.g.,

x#$&*M). In this case, three participants from Experiment 1,

two from Experiment 2, and two from Experiment 3 were

1

2

3

4

5

Control Terrorism Healthcare Pollution Corporate Misconduct

A

B

Anger Disgust Fear Anxiety

1

2

3

4

5

6

7

Control Terrorism Healthcare Pollution Corporate Misconduct

Overall Index Personal Threat Group Threat

Figure 1. Effects of priming condition on affective experience and perceived threat. (A) Effects of priming condition on affect. (B) Effects of
priming condition on perceived threat indices.

90 Social Psychological and Personality Science 11(1)

excluded from analyses. Finally, an attention check task was

utilized (e.g., for the next question, please select “slightly dis-

agree”). Participants failing this check (22 participants in

Experiment 1, 14 participants in Experiment 2, and 9 partici-

pants in Experiment 3) were excluded from analyses.

Health-Care Threat

Participants read about the true story of Kyler Van Nocker, a

5-year-old stricken with cancer who was denied health-care

coverage by his parents’ insurer. To assess spontaneous reac-

tions to the article, participants were presented with the follow-

ing prompt: “In the space provided, we would like you to list

whatever thoughts and feelings you have about the article you

just read. Five or six sentences are sufficient.” Participants

were free to write anything; the only constraint was answers

needed to be between 200 and 5,000 characters. All newspaper

articles are presented in the Online Supplement (Appendix A).
3

Control Prime

Participants assigned to the control condition read an article

about food allergies and then completed the same writing

prompt.

Assessment of Mood

Immediately after the threat manipulation, participants com-

pleted a mood inventory in which they were presented with a

randomized set of 35 mood adjectives (happy, proud, irritated,

satisfied, mad, upset, sad, pleased, relaxed, unhappy, angry,

irate, dejected, anxious, nervous, worried, confident, calm,

content, relieved, dissatisfied, furious, scared, guilty, dis-

gusted, revulsion, repulsed, afraid, fearful, interested, deter-

mined, excited, offended, nauseated, and sickened). For each

adjective, participants evaluated their mood on a scale ranging

from 1 (not at all) to 5 (very much so). We were most directly

interested in the effects of anger (mad, angry, furious, dis-

gusted, revulsed, sickened; a ¼ .97) but also created a fear/
anxiety index (nervous, scared, afraid, fearful, anxious, wor-

ried; a ¼ .93). In the analyses to follow, references to “anger”
and “fear/anxiety” refer to composite indices.

Political Attitudes Rating Task

Following the mood inventory task, participants evaluated a

series of 50 political statements, presented in a different rando-

mized order for each participant (see Online Supplement

[Appendix B] for a complete listing). All items were completed

on a 7-point Likert-type scale, ranging from 1 (strongly dis-

agree) to 7 (strongly agree).
4

To test our main prediction, we formed a health-care atti-

tudes index, based on 18 items. Some items were written in a

proliberal direction (I would support regulation of privatized

health insurance companies), whereas others were written in

a pro-conservative direction (The government should have a

minimal role in the health-care industry). After reverse coding

the pro-conservative items, a composite of all 18 items dis-

played high levels of reliability (a ¼ .91).
To examine whether our priming condition altered other

political attitudes, we formed social liberalism, social conser-

vatism, and hawkish military indices. Social liberalism was

constructed from a battery of 9 items (e.g., I am happy that

same-sex marriage is now legalized), whereas social conserva-

tism was based off an 11-item composite (e.g., Abortion should

be outlawed in all circumstances). Both composites displayed

high levels of reliability (as ¼ .83 and .88, respectively).
Finally, hawkish military attitudes were constructed from a bat-

tery of 12 items (The best way to ensure peace is through mil-

itary strength) after reverse scoring as necessary (a ¼ .94).
Because we did not have strong predictions of between-

subjects differences on these indices, we present these findings

in supplemental analyses.

Results

Effects of Health-Care Threat on Mood

We expected and found higher levels of anger in the threat ver-

sus control condition, M ¼ 3.32 (95% confidence interval [CI]
¼ [3.22, 3.43]) versus M ¼ 1.13 (95% CI [1.02, 1.24]), F(1,
556) ¼ 782.77, p < .001, Z2 ¼ 0.59, and (b) higher levels of fear/anxiety in the threat versus control condition, M ¼ 2.23 (95% CI [2.13, 2.33]) versus M ¼ 1.30 (95% CI [1.20, 1.40]), F(1, 556) ¼ 171.43, p < .001, Z2 ¼ 0.24. Effects of Health-Care Threat on Political Attitudes We expected and found being primed with the health-care threat (vs. control) increased support for liberal health-care atti- tudes, Ms ¼ 5.22 (95% CI [5.11, 5.33]) versus 5.02 (95% CI [4.91, 5.13]), F(1, 556) ¼ 6.06, p ¼ .01, Z2 ¼ 0.01. Mediation Analyses To test for mediation, we used Model 4 of Hayes’s (2018) Process macro. Experimental condition served as the indepen- dent variable (X), anger (M1) and fear/anxiety (M2) served as the mediators (M), and health-care attitudes served as the out- come measure (Y). These findings, presented in Figure 2, show evidence for mediation, confirming that anger, but not fear/anxiety, contributed to increased support for liberal health-care attitudes. Supplemental Analyses Analyses revealed no effects for social liberalism, Ms ¼ 5.07 versus 4.19, F(1, 556) ¼ 1.53, p ¼ .22, Z2 ¼ 0.00, nor did we observe between-subject differences on social conserva- tism, Ms ¼ 3.35 versus 3.25, F(1, 556) ¼ 0.82, p ¼ .37, or hawkish attitudes, Ms ¼ 3.84 versus 3.88, F(1, 556) ¼ 0.11, p ¼ .74. Eadeh and Chang 91 Experiment 2 Experiment 2 represents the second test of our theoretical model, which predicts pollution threats (vs. control) will increase the support for related aspects of political liberalism. To offset the possibility that our findings were due to how threat was operationalized, participants read about either a water or air pollution threat. After condition assignment, parti- cipants rated their mood and a series of political attitudes. Method Participants and Design One hundred eight-four (62 male, 120 female, 2 did not answer) U.S. participants were recruited from MTurk. The design included a between-subjects manipulation of priming type (air pollution threat vs. water pollution threat vs. control). After completing a standard mood inventory, all participants evaluated a series of political statements, presented in counter- balanced order for each participant. Because we anticipated a similar pattern across the two pollution threats, half of the par- ticipants were assigned to the control condition, with the remaining participants split evenly across the threat primes. As expected, there were no differences in affect, nor were there differences in political attitudes across the two priming manip- ulations. For these reasons, we merged the two priming condi- tions together. Air Pollution Prime Participants in this condition read about Kavon Cooper, a 12-year-old child who died from Asthma in Evansville, IN. The article detailed Kavon’s bout with asthma and the proximity of extreme pollution near his home. After reading this article, participants completed the same writing task from Experiment 1. Participants were free to write anything they wished; the only constraint was that answers needed to be between 200 and 5,000 characters. Water Pollution Prime Participants in this condition read about the story of Josephine Reed, a 2-year-old who suffered heavy metal poisoning as a result of drinking polluted water. After reading this article, par- ticipants completed the same writing task from Experiment 1. Participants were free to write anything they wished; the only constraint was that answers needed to be between 200 and 5,000 characters. Control Prime The control prime was identical to Experiment 1. Participants were able to write anything they wished about the article, with similar length constraints to the pollution prime conditions. Mood and Political Attitudes The method and materials in this study were similar to our ear- lier study. Our anger and fear/anxiety mood composites were identical (as ¼ .97 and .93, respectively). Because the threat in this study was related to the environment, we generated a set of 16 items to assess participants’ views on environmental mat- ters in the United States. Sample items on this scale included support for clean energy solutions, funding the Environmental Protective Agency, and the role of government in regulating private industries. After reverse scoring as necessary, a compo- site of all 16 items indicated high levels of reliability (a ¼ .92). We hypothesized that pollution threats would alter environ- mental attitudes in a pro-liberal direction. Finally, we measured Index of Indirect Effects Mediating Variable esb 95% LLCI 95% ULCI 48.073.021.016.0regnA Fear/Anxiety 0.01 0.05 -0.09 0.10 Note: numbers in parentheses refer to standard error estimates. * p < .05, *** p < .001. Total Effect: b = 0.20 (0.08)* Direct Effect: b = -0.42 (0.12)*** Anger Fear/Anxiety Experimental Condition (1 = Healthcare Threat, 0 = Control) Liberal Healthcare Attitudes b = 0.93 (0.07) *** b = 2.19 (0.08)*** b =0.28 (0.05) *** b = 0.01 (0.05) Figure 2. Mediation analyses on liberal health-care attitudes, Experiment 1, anger (N ¼ 558). Index of indirect effects. 92 Social Psychological and Personality Science 11(1) the other political attitudes (social liberalism, social conserva- tism, hawkish attitudes, and health-care attitudes) similarly to Experiment 1 but used fewer items for each construct (See Online Supplement [Appendix C]). Results Effects of Priming Condition on Mood Compared to participants in the control condition, we expected and found higher levels of anger among participants assigned to the threat condition M ¼ 2.99 (95% CI [2.79, 3.19]) versus 1.17 (95% CI [0.97, 1.37]), F(1, 182) ¼ 163.41, p < .001, Z2 ¼ 0.47, and this was also true with respect to fear/anxiety M ¼ 2.33 (95% CI [2.14, 2.52]) versus M ¼ 1.35 (95% CI [1.16, 1.53]), F(1, 182) ¼ 54.61, p < .001, Z2 ¼ 0.23. Effects of Priming Condition on Political Attitudes As predicted, participants assigned to the pollution threat con- dition (vs. control) expressed greater support for liberal envi- ronmental attitudes, M ¼ 5.65 (95% CI [5.44, 5.85]) versus M ¼ 5.16 (95% CI [4.96, 5.37]), F(1, 182) ¼ 10.72, p ¼ .001, Z2 ¼ 0.06. Mediation Analyses We performed mediation analyses using Model 4 of Hayes’s PROCESS Macro (2018), with priming condition serving as the independent variable (X), anger (M1) and fear/anxiety (M2) serving as mediators, and environmental attitudes serving as the outcome measure (Y). As shown in Figure 3, analyses provided strong evidence of affective mediation, as higher levels of anger, but not fear/anxiety, correlated with liberal environmental attitudes. Supplemental Analyses With Other Attitude Indices We also examined whether there were between-subject differ- ences on social conservatism, general social liberalism, liberal health-care attitudes, and hawkish military indices. Of these indices, we found differences (threat vs. control) on liberal health-care attitudes, Ms ¼ 4.91 (1.18) versus 4.48 (1.32), F(1, 182) ¼ 5.56, p ¼ .02, Z2 ¼ 0.03, and social conservatism, Ms ¼ 2.98 (1.29) versus 3.39 (1.26), F(1, 182) ¼ 4.93, p ¼ .03, Z2 ¼ 0.03, indicating participants in the threat condition were more likely to support liberal health-care attitudes and were less likely to support social conservatism. In contrast, there were no differences for our general measure of social liberal- ism, Ms ¼ 4.90 (1.22) versus 4.62 (1.31), F(1, 182) ¼ 2.26, p ¼ .14, Z2 ¼ 0.01, nor were there differences on hawkish atti- tudes, Ms ¼ 4.18 (1.58) versus 4.46 (1.61), F(1, 182) ¼ 1.42, p ¼ .24, Z2 ¼ 0.01. Experiment 3 Experiments 1 and 2 provide consistent evidence that if the activated threat is in domains in which liberals “own,” partici- pants would be more supportive of those aspects of liberalism. Nevertheless, one drawback from these studies is that both threats involved children, raising an important alternative hypothesis: threats involving children increased support for lib- eral attitudes. To offset this possibility, our third experiment did not include a threatening context involving children. More- over, Experiment 3 examined a different domain in which lib- erals carry perceptual “ownership”: corporate misconduct. Indeed, Democrats (vs. Republicans) generally claim owner- ship in controlling the power of large corporations and dealing with corporate financial fraud (ABC-Washington Post, 2002). Index of Indirect Effects Mediating Variable esb 95% LLCI 95% ULCI 19.041.091.015.0regnA Fear/Anxiety -0.14 0.11 -0.38 0.06 Note: numbers in parentheses refer to standard error estimates. * p < .05, ** p < .01, *** p < .001. Total Effect: b = 0.48 (0.15)** Direct Effect: b = 0.11 (0.20) Anger Fear/Anxiety Experimental Condition (1 = Pollution Threat, 0 = Control) Liberal Environmental Attitudes b = 1.82 (0.14) *** b = 0.98 (0.13)*** b = 0.28 (0.11) * b = -0.14 (0.11) Figure 3. Mediation analyses on liberal environmental attitudes, Experiment 2, anger (N ¼ 184). Index of indirect effects. Eadeh and Chang 93 Participants in this experiment read about corporate conduct from the 2008 financial crisis (vs. control). We predicted parti- cipants in this threat (vs. control) would increase the support for financial regulation. Method Participants and Design A total of 225 participants (82 male, 141 female, 1 other, 1 did not answer) from the United States were recruited from MTurk. The design included a between-subjects manipulation of prim- ing threat (financial threat vs. control). After completing a stan- dard mood inventory, all participants evaluated a series of political statements. 2008 Financial Crisis Prime Participants in this condition read a summary of the 2008 finan- cial crisis, which detailed predatory lending tactics taken by many large banks, the bundling of mortgage-backed securities, and chronicles a wrongly foreclosed homeowner. After reading this article, participants completed the same writing task from previous experiments, with similar length constraints. Control Prime The control prime (food allergies) was identical to previous experiments and was accompanied by the same length con- straints as the financial threat condition. Mood and Political Attitudes Anger (a ¼ .96) and fear/anxiety (a ¼ .93) were collected in a similar fashion to earlier experiments. Since the threat acti- vated in Experiment 3 involved corporate misconduct by the banking industry, we examined financial regulatory attitudes. These 14 items revolved around the regulation of financial institutions and increasing consumer financial protections. A composite based on the average of these items was formed (a¼ .93). We also measured all other political attitudes indices seen in our previous studies. However, none of these indices, presented in Online Supplement (Appendix D), approached statistical significance, all Fs < 1.00, all ps > .25.

Results

Effects of Priming Condition on Mood

We expected and found participants expressed greater levels of

anger if they were assigned to the financial threat versus

control condition, M ¼ 2.76 (95% CI [2.59, 2.93]) versus
1.16 (95% CI [0.99, 1.33]), F(1, 223) ¼ 229.64, p < .001, Z2 ¼ 0.51. Parallel effects were found for fear/anxiety, M ¼ 1.96 (95% CI [1.83, 2.10]) versus M ¼ 1.37 (95% CI [1.23, 1.51]), F(1, 223) ¼35.36, p < .001, Z2 …