Disgusted and Afraid Consumer Choices Under The Threat of Contagious Disease

Disgusted and Afraid: Consumer Choices
under the Threat of Contagious Disease
CHELSEA GALONI
GREGORY S. CARPENTER
Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

HAYAGREEVA RAO
Consumers regularly encounter cues of contagious disease in daily life—a com-

muter sneezes on the train, a colleague blows their nose in a meeting, or they
read recent headlines about the dangerous spread of a disease. Research has
overwhelmingly argued that the dominant response to these cues is disgust—an
emotion that leads to a desire to reject and avoid potential contamination. We ar-
gue, however, that contagious disease cues can also elicit fear. Across four
experiments and two large empirical data analyses of the presence of contagious
disease on actual consumption behavior, we find that cues of contagious disease
increase both fear and disgust, and these emotions together form a unique behav-
ioral tendency with respect to consumer behavior. Relative to either emotion
alone, disgust and fear increase preference for more-familiar products asymmetri-
cally over less-familiar ones. These results contribute theoretically to research on
complex emotional states and the behavioral tendencies of emotions, document a
systematic and consequential impact of contagious disease cues on real con-
sumption behavior, and have significant practical implications for marketers.
Keywords: contagion, emotion, disgust, familiarity, fear, disease
O n March 11, 2020, the World Health Organization de-

clared COVID-19—or Coronavirus Disease 2019—a
pandemic. Even before this announcement, the disease
Reyes, and Ortiz 2020; Ivanova and Craver 2020). Hand
sanitizer, disinfectants, shelf-stable foods, and even toilet
paper were out of stock on major store shelves, and the
caused mass hysteria around the world. Schools, restau- supply of respirator masks and other necessary medical
rants, and places of work closed, and governments man- equipment was so low due to public demand that hospitals
dated that people shelter in place, sometimes requiring that had a hard time getting enough personal protective equip-
people carry documentation when leaving home (Hauck, ment for healthcare workers on the front lines (Campbell
and Murphy 2020; Jacobs, Richtel, and Baker 2020; Togoh
Chelsea Galoni ([email protected]) is an assistant professor 2020). The pandemic vividly illustrates that a contagious
of marketing at the Tippie College of Business, University of Iowa, 21 E disease can provoke a powerful emotional response that
Market Street, Iowa City, IA 52242. Gregory S. Carpenter (g-
[email protected]) is the James Farley/Booz Allen influences consumers’ choices. While COVID-19 is a par-
Hamilton professor of marketing strategy at the Kellogg School of ticularly salient example of this effect, consumers often in-
Management, Northwestern University, 2211 Campus Drive, Evanston, IL cidentally find themselves in the presence of contagious
60208. Hayagreeva Rao ([email protected]) is the Atholl McBean
Professor of Organizational Behavior and Human Resources at the disease. For instance, a commuter sneezes on a train, a
Stanford Graduate School of Business, Stanford University, 655 Knight friend complains of a sore throat, or a partner comes home
Way, Stanford, CA 94305. Please address correspondence to Chelsea with a chesty cough. Unlike COVID-19, these instances
Galoni. Supplementary materials are included in the web appendix accom-
panying the online version of this article. are unlikely to lead to hysteria, but they may nevertheless
have a systematic and consequential impact on consump-
Editor: J. Jeffrey Inman
tion patterns.
Associate Editor: JoAndrea Hoegg But what might this impact look like? Cues of conta-
gious disease can influence consumption behavior system-
Advance Access publication May 28, 2020
atically by eliciting emotions. The existing literature
C The Author(s) 2020. Published by Oxford University Press on behalf of Journal of Consumer Research, Inc.
V
All rights reserved. For permissions, please e-mail: [email protected] Vol. 47 2020
DOI: 10.1093/jcr/ucaa025
373
374 JOURNAL OF CONSUMER RESEARCH
overwhelmingly suggests that, when an individual is con- on the basis of how the situation is judged. For instance,
fronted with the threat of contagious disease, the dominant Smith and Ellsworth (1985) identified six dimensions on
emotional response is disgust (Han, Lerner, and which emotions can be classified and differentiated: atten-
Zeckhauser 2012; Rozin and Haidt 2013; Rozin, Haidt, and tion (whether people are driven to pay attention to, ignore,
McCauley 2008; Tybur et al. 2013). When a consumer or avoid the situation), certainty (whether the situation is
senses the presence of contaminants in the environment predictable or not), control (whether the situation is con-
and feels the emotion of disgust, the dominant behavioral trolled by the person, someone else, or something else),
tendency is to reject the contaminant and to isolate the self pleasantness (whether the situation is pleasant or unpleas-
from potential infection by reducing consumption (Han ant), responsibility (whether the situation was brought
et al. 2012; Lerner and Keltner 2000; Lerner, Small, and about by the self or others), and anticipated effort (how
Loewenstein 2004). much the person has to work in the situation).

However, we propose that cues of a contagious disease While discrete emotions can share appraisals on some of
can evoke fear as well. Fear is an emotional response to un- the underlying dimensions (i.e., both fear and disgust are
certainty and lack of personal control over an outcome unpleasant emotions), one key tenant of the appraisal ten-
(Ellsworth and Smith 1988; Lerner and Keltner 2000). dency framework is that each discrete emotion differs from
When a consumer encounters a contagious disease cue, in others on the basis of its unique profile on each of the ap-
addition to providing a salient cue signaling the presence praisal dimensions. For example, fear and anger are dis-
of contaminants, there is also often a level of uncertainty tinct from each other because fear is defined by uncertainty
and lack of personal control. In other words, when a fellow and a lack of personal control whereas anger is defined by
commuter sneezes on the train, a nearby commuter may a high sense of certainty and other responsibility (Lerner
feel disgusted and also fears the possibility of becoming and Keltner 2001; Lerner et al. 2003; Tsai and Young
sick. 2010).
In this article, we explore how this joint emotional re- Each discrete emotion has a unique pattern of cognitive
sponse to contagious disease cues can alter the systematic appraisals, and research has shown that each emotion also
impact these cues have on consumption behavior. We draw has a unique influence on how people tend to think and be-
on research on appraisal tendencies and mixed emotions to have in subsequent tasks. For example, fear and anger have
argue that the behavioral response to a contagious disease opposite appraisals on the dimension of certainty, so they
cue will asymmetrically increase preference for familiar have opposing influences on assessments of risk. Because
and comforting products, and therefore sometimes increase fear results from an appraisal of low certainty, this leads
consumption overall, contrary to the prediction of prior re- people feeling afraid to believe that outcomes are generally
search on disgust. We assess this proposal across six stud- uncertain and to therefore make more risk averse choices
ies, including two large-scale empirical analyses of in subsequent tasks (Lerner and Keltner 2001;
influenza incidence and panel or retail sales data as well as Raghunathan and Pham 1999). In contrast, because anger
four experiments. Across these studies, we find that cues of results from an appraisal of high certainty, people become
contagious disease can elicit both disgust and fear in con- more risk seeking in subsequent tasks (Lerner and Keltner
sumers. Rather than avoid purchasing, consumers demon- 2001; Lerner and Tiedens 2006). Disgust results from an
strate an asymmetric preference for familiar products. Our assessment of being too close to an offensive object, so
findings highlight the importance of understanding when, people feeling disgusted exhibit a general tendency to
why, and how the joint experience of emotions can impact avoid and expel (Di Muro and Noseworthy 2013; Galoni
behavioral tendencies. We demonstrate a significant influ- and Noseworthy 2015; Han et al. 2012; Lerner et al. 2004).
ence of contagious cues on real consumption decisions, However, the situations people face in daily life can of-
add theoretical insight into the role of joint emotions in ten afford multiple judgments on appraisal dimensions, and
influencing consumption behavior, and conclude with dis- research as well as common wisdom has shown that people
cussing practical implications for marketers. can and do experience multiple, and sometimes conflicting,
emotions simultaneously. Most research has focused on the
simultaneous experience of multiple emotions that differ
CONCEPTUAL DEVELOPMENT on valence. For instance, feeling “bittersweet” (sadness
The Appraisal Tendency Framework and and contentment) or “guilty pleasures” (guilt and happi-
ness) is the example of the simultaneous experience of
Discrete versus Mixed Emotional States emotions that share appraisals on some dimensions but
The appraisal tendency framework argues that emotions conflict on the appraisal of pleasantness. However, people
arise from complex judgments consumers make about their can and do feel emotions of the same valence simulta-
relationship to their environment (Lazarus 1991; Lerner neously, even when the emotions conflict on other ap-
and Keltner 2000; Smith and Ellsworth 1985). Each dis- praisal dimensions (Lee and Ellsworth 2013; Morales, Wu,
crete emotion can be defined and differentiated from others and Fitzsimons 2012). Notably, research shows that people
GALONI, CARPENTER, AND RAO 375
can simultaneously experience fear and disgust, even Regardless of whether this cue signals a real risk of harm
though the two emotions conflict on appraisals of certainty to the observer (e.g., if they are immune), the salience of
and attention (Morales et al. 2012). the cue nevertheless allows for an appraisal that there are
Despite some research documenting the phenomenon of contaminants present. This appraisal elicits a disgust re-
mixed emotions with conflicting appraisals, the appraisal sponse. In addition, the contagious disease cue often
tendency framework provides little insight and guidance affords an appraisal that there is a real threat imminent in
on how or whether conflicting appraisals of mixed emo- the environment that is out of their control (e.g., the lack of
tional states resolve within the individual, or how the expe- control over getting sick). This appraisal elicits a fear
rience of multiple emotions might influence behavioral response.
tendencies. This has led to a call for an increased focus on
the psychology of these mixed emotional states and how Conflicting Appraisals in a Joint Fear and

the interaction of multiple appraisals can predict behavior Disgust Response
(So et al. 2015). Here, we use the appraisal tendency
framework and integrate research on the psychology of While the disgust and fear responses to contagious cues
emotional responses to understand how the simultaneous share appraisals of unpleasantness (both are unpleasant),
experience of fear and disgust might influence subsequent control (both lack personal control), responsibility (both
behavioral tendencies. are other responsibility), and anticipated effort (both are
high), they do not share appraisals of certainty and atten-
Contagious Disease Cues: A Joint Fear and tional activity. Disgust arises from a certainty appraisal
whereas fear arises from an uncertainty appraisal (Smith
Disgust Response
and Ellsworth 1985; Tiedens and Linton 2001). Disgust is
When people encounter a contagious disease cue, it is characterized by low attentional activity whereas fear is
widely assumed that the immediate emotional response is ambivalent with respect to attentional activity (Blanchard
disgust. In fact, the emotion of disgust itself is thought to et al. 2011; Coleman et al. 2017; Krusemark and Li 2011).
have evolved as an adaptive mechanism of avoiding dis- On first glance, it might seem that these conflicting
ease, poison, and food-borne illness (Haidt, McCauley, and appraisals might mean disgust and fear ought not to be ex-
Rozin 1994; Nemeroff and Rozin 1994; Rozin et al. 2008; perienced together, or, at the very least, these conflicting
Rozin, Millman, and Nemeroff 1986; Tybur et al. 2013). appraisals might cancel each other out and thus have no in-
The appraisals that define the experience of disgust include fluence on subsequent consumption behavior. However,
certainty that there are contaminants in the environment we propose and show that fear and disgust can be experi-
(high certainty) and a desire to avoid becoming contami- enced together and that these seemingly conflicting
nated (low attentional activity; Lerner et al. 2004; Smith appraisals hold the key to the unique behavioral tendencies
and Ellsworth 1985). in response to contagious disease.
However, contagious disease cues can also evoke fear.
Like disgust, fear is an adaptive response that helps con- Resolving Certainty: Certain Source, Uncertain
sumers detect and defend against threats in their environ- Outcome
ment (Marks and Nesse 1994; Neuberg, Kenrick, and
Schaller 2011). Consumers feel fear when they appraise First, we propose that people do not experience a cer-
situational uncertainty and a lack of personal control over tainty conflict when simultaneously experiencing fear and
outcomes (Ellsworth and Smith 1988; Lerner and Keltner disgust because, in the case of the response to contagious
2000, 2001; Smith and Ellsworth 1985). Beyond making disease cues, fear and disgust are driven by certainty
the presence of contaminants salient and evoking disgust, appraisals directed toward different aspects of the situation.
contagious disease cues usually afford appraisals of uncer- In particular, appraisals of certainty with respect to the emo-
tainty and lack of control with respect to whether the con- tion of disgust are directed toward the source of the threat,
sumer is likely to get sick. It is therefore plausible that whereas appraisals of uncertainty with respect to the emo-
encountering cues about contagious illness can elicit fear tion of fear are directed toward the outcome of the threat.
in consumers. In fact, the response to COVID-19 was char- When faced with a contagious disease cue, the certainty
acterized as “panic” by the media (Gross 2020) and some consumers’ experience arises from a feeling of knowing that
researchers have called the response to pathogen cues there are in fact pathogens present in the environment. In
“contamination anxiety” (Dutta and Rao 2015). contrast, the appraisal of uncertainty arises from not know-
We thus propose that contagious disease cues elicit a ing whether those pathogens will contaminate the consumer
joint disgust and fear response in consumers. First, a conta- (Dutta and Rao 2015; Marzillier and Davey 2005).
gious disease cue usually makes contaminants salient (e.g., Imagined in a different context, people can still feel certain
the presence of pathogens via coughing, sneezing, vomit- that a tiger is chasing them whilst simultaneously being
ing, or news reports about the spread of disease). afraid of what might happen if it catches them. Thus, in
situations where appraisals of certainty or uncertainty can be Resolving Attention: Avoiding Pathogens and
directed toward both the source and outcome, the conflicting Approaching Control
certainty appraisals of disgust and fear may not cancel each
other out, and people could thus feel both disgusted and Another appraisal dimension on which fear and disgust
afraid simultaneously. More formally: do not align is that of attentional activity. The experience
of disgust is accompanied by an intense desire to avoid and
H1a: When confronted with a contagious disease cue, con- expel the source of contamination (i.e., a strong appraisal
sumers feel disgust from appraising a certain source of of attentional avoidance). Indeed, research on the neuro-
threat and fear from an uncertain outcome of that threat. physiological response to disgust has shown that people en-
But how might these diverging certainty appraisals af- gage in significant and immediate repulsion behaviors
fect subsequent behavior? We propose that the appraisal of when confronted with disgusting stimuli (Marks and Nesse
1994; Neuberg et al. 2011). Attention narrows to the

a certain source inherent in the disgust response to conta-
gious disease cues spurs action that might not occur under disgust-causing stimuli, and bodily systems associated with
conditions of fear alone. Research has shown that there are restricting access to internal functions are engaged (e.g., re-
two possible behavioral tendencies of fear: freezing or ac- duced activity in brain centers associated with attention to
tion (Blanchard et al. 2011; Gray 1987; Izard 1977; the external environment, closing of eyes, and slowing of
Tiedens and Linton 2001). The key decision criteria be- breath; Anderson et al. 2013; Ekman, Levenson, and
tween these two behavioral outcomes is whether people be- Friesen 1983; Marks and Nesse 1994; Neuberg et al.
lieve the source of threat is looming or imminent. When 2011). In contrast, fear is ambivalent with respect to atten-
people first feel that a threat is looming, fear leads to freez- tional activity and whether people are motivated to avoid
ing behavior in service of detecting the threat, assessing or approach is contingent upon which behavior is believed
the real risk, and deciding on an appropriate course of ac- to help resolve the threat in that context (Anderson et al.
tion (Blanchard et al. 2011). We propose that when con- 2013; Coleman et al. 2017; Neuberg et al. 2011; Smith and
fronted with a contagious disease cue, the certainty Ellsworth 1985). For instance, people avoid making risky
appraisal about the source of the threat (i.e., that the con- choices when they feel afraid because that behavior leads
sumer believes pathogens are present) satisfies the purpose to even greater uncertainty in outcomes (Raghunathan and
of freezing behavior, making the threat appear more immi- Pham 1999). Yet people feeling afraid are also more likely
nent, and therefore encouraging consumers to take action. to make (rather than defer) a choice when doing so resolves
One additional feature of pathogens as a source of threat uncertainty and helps the consumer regain a feeling of con-
is that, once exposed, consumers have little control over trol (Coleman et al. 2017).
whether they get sick. Thus, when pathogen presence is We propose that experiencing fear and disgust when
isolated as the source of threat, it might further highlight confronted with a contagious disease cue will thus narrow
and amplify the perceived lack of control over outcomes. avoidance behaviors to the source of the threat, specifi-
Feeling like one lacks control motivates action toward cally, but will spark approach behaviors toward objects
regaining control (Brehm 1966; Rothbaum, Weisz, and that help restore a sense of control. This pattern of behavior
Snyder 1982; Whitson and Galinsky 2008). For example, seems to be present—albeit untested—in other research on
people who feel like they lack control show a preference disgust. In research paradigms where researchers manipu-
for making a choice over deferring (Coleman et al. 2017), late disgust but not fear, the experience of disgust alone
have a greater desire for power (Lammers et al. 2016), and seems to lead to avoidance behaviors in general, irrespec-
are more likely to find illusory patterns in random data tive of the actual benefit of approach behaviors in subse-
(Whitson and Galinsky 2008). Thus, when pathogens are quent tasks. For example, when normatively disgusting
isolated as the source of threat, they might highlight and products that offer no real threat to the self (i.e., new and
amplify the perceived lack of control over outcomes, fur- unopened feminine hygiene products) come into contact
ther motivating action toward regaining a sense of control. with cookies, people feel disgusted and are less willing to
Thus, the differences in disgust and fear on the appraisal try the cookies (Morales and Fitzsimons 2007). In contrast,
dimension of certainty are not in conflict and might even in research paradigms where manipulations of disgust
enhance the behavioral tendency to act against the threat. might also induce fear, participants seem to avoid the path-
More formally: ogen, specifically, but also approach objects that might
help regain a sense of control over the outcome. One such
H1b: Relative to fear alone, experiencing disgust and fear paradigm is studies on physically dirty money (Di Muro
together in response to contagious disease cues increases and Noseworthy 2013; Galoni and Noseworthy 2015).
perceived lack of control over outcomes. When people encounter physically dirty money, research
H1c: Relative to fear alone, experiencing disgust and fear has shown that participants feel disgusted, but participants
together in response to contagious disease cues motivates may have also felt fear from the uncertainty about what
taking action toward eliminating the threat. those pathogens may have transferred to them. In turn,
empirical results show that participants do indeed avoid the tendencies toward regaining control over the outcome. The
source of pathogens (e.g., by spending more of the dirty joint experience of disgust and fear thus leads to greater ac-
currency and de-valuing products that the currency may tion tendencies than fear alone and greater approach behav-
have come into contact with; Di Muro and Noseworthy iors than disgust alone (hypothesis 1c), thus
2013; Galoni and Noseworthy 2015), and also actively ap- asymmetrically increasing preference for the familiar (hy-
proach and increase the value they place on products that pothesis 2). We test these hypotheses across six studies.
help restore a sense of control over the outcome—in this
case, cleansing products (Galoni and Noseworthy 2015). STUDY 1: CONSUMER PANEL DATA
Thus, the joint fear and disgust response to contagious dis-
ANALYSIS
ease cues might narrow avoidance behaviors to objects as-
sociated with the source of threat, specifically, but increase Our initial analysis examines whether cues of contagious

approach behaviors toward objects associated with regain- disease consumers confront in daily life might systemati-
ing a sense of control over the threat. cally influence consumption choices. To provide an initial
test of our hypothesis that disease should asymmetrically
Unique Behavioral Tendencies of the Joint increase preference for the familiar (hypothesis 2), we turn
Emotion of Fear and Disgust to consumer panel data from Nielsen alongside archival
We have thus far proposed that the joint fear and disgust data from the Center for Disease Control (CDC) and
response to contagious disease cues will spark action more Google Flu Trends over a period of 6 years. We use this
than fear alone and will draw attention to and spark ap- data to assess whether the presence of contagious disease
proach behaviors toward objects that help restore a sense in a particular place at a particular time systematically
of control over outcomes more than disgust alone (hypoth- affects what and how much people buy.
esis 1b-c). But what might this mean with respect to the
systematic effect of such cues on consumption behavior, Data Treatment Methods
specifically? We propose that contagious disease cues will Our time frame is the period starting the first week of
asymmetrically increase preference for familiar products. January 2009 and ending the last week of December 2014.
By familiar, we mean brands or products that are well- Our unit of analysis is a weekly estimate of flu levels and
known and commonly used by the consumer. When con- household sales within each state in the contiguous 48
sumers feel they lack control, one way in which they can states of the United States, plus the District of Columbia.
restore a sense of control is to reestablish a feeling of struc- We chose this unit of analysis and time frame because they
ture and predictability (Whitson and Galinsky 2008). provide a level of granularity that was consistently avail-
Familiar products can thus help restore a sense of control able across the Google, CDC, and Nielsen datasets, and
because they are seen as more predictable (Bornstein, therefore did not require that we make additional assump-
Kale, and Cornell 1990), certain (Park and Lessig 1981), tions trying to estimate a more general or granular level of
and comforting (Titchener 1910) to the consumer than analysis. This time frame and unit of analysis yield a data-
less-familiar or novel products. Indeed, research has shown set representing 312 weeks across 49 states. We log trans-
that unfamiliar products can reduce a consumer’s sense of form our disease index and household purchasing data
control, further compounding the feelings of a lack of con- because the data have strong positive skew. In addition, a
trol rather than resolving them (Faraji-Rad, Melumad, and log–log analysis allows us to interpret coefficients as elas-
Johar 2017). Thus, if the joint experience of disgust and ticities so that a 1% change in disease results in a corre-
fear leads to a greater tendency to approach objects that sponding percent change in sales.
can help restore a sense of control, then when faced with a
choice between a familiar product and a less-familiar prod- An Index of Disease Presence. We chose to study the
uct, we expect that consumers will choose the familiar. impact of the influenza virus on consumption behavior be-
cause it is a well-known, highly contagious, and well-
H2: Cues of contagious disease asymmetrically increase publicized disease with publicly available data. We con-
preference for familiar products. ceptualized disease presence as the aggregate presence of
Taken together, our hypotheses suggest that cues of con- flu-like illnesses in a particular state in a particular week.
tagious disease evoke a joint fear and disgust response in We created a weekly state-level disease index by taking
consumers through a unique mix of appraisals. The sa- the average of five standardized variables available from
lience of contaminants in the environment elicits disgust Google Flu Trends and the CDC. The variables from the
by making the source of the threat appear evident and CDC included a variable rating the overall presence of the
evokes behavioral tendencies toward avoiding the offend- flu in the state, a variable indicating the total number of
ing contaminant. Likewise, the lack of control in the likeli- flu-like illness cases presented in hospitals, the number of
hood of getting sick evokes fear and elicits behavioral flu-like specimens sent to the laboratory, and the percent of
FIGURE 1
AVERAGE ILLNESS INDEX BY WEEK OF THE YEAR ACROSS THE UNITED STATES.

laboratory specimens returned positive. The Google Flu infrequently (e.g., furniture, appliances, electronics) or
Trends data included a single state-level weekly count vari- only by a subset of consumers (e.g., baby products, auto-
able that represents the estimated number of cases of the motive supplies) are not conducive to assessing a system-
flu in a particular state based on flu-related queries on the atic shift in preference. It was also important to choose
search engine. As a check to ensure that our disease index relatively homogenous categories where different products
was tracking the seasonal pattern of the flu, we plotted the are likely to be considered substitutes. For example, the
average weekly level of illness across the United States. As category “household paper products” is too broad because
is evident in figure 1, our illness index indeed captures the buying more toilet paper than paper towels does not neces-
regular seasonality of the flu (increased illness in the first sarily mean that toilet paper is more familiar. To provide a
and last weeks of the year, decreased illness over the conservative estimate of our effect, we excluded personal
summer). cleansing products and household cleaners because of their
close association with disgust (Galoni and Noseworthy
Household Spending. To assess the systematic influ-
2015). Finally, we included a variety of food and nonfood
ence of illness on consumption within household, we used
products to assess the generalizability of the effect.
the Nielsen consumer panel data. The panel data provides
With these constraints, we chose four product categories
household demographic information along with date, store,
for analysis: canned soup, cookies, paper towels, and bat-
product, and price information for every shopping trip the
teries. Each row in our data represents a purchase occasion
household makes during their time on the panel. Over
of a product in one of these four categories and includes a
60,000 households participate in the consumer panel pro-
unique product identifier (upc), a variable indicating which
gram each year. Households spend an average of 3.7 years
category the product belongs to, the number of equivalent
in the panel during our time frame and make an average of
units purchased (in ounces for soup and cookies, and in
161.5 shopping trips per year. A total of 112,764 unique
units for batteries and paper towel), the price paid (per
households engaging in over 12 million purchases are rep-
equivalent unit), the time and location where the purchase
resented in our time frame (January 2009–December
was made, the disease index at the time the purchase was
2014).
made, and household demographics. Each unique product
Choosing Product Categories. To have rich enough purchased in a given shopping trip has its own line in the
data to assess our proposal, we chose product categories dataset. If a household bought two different products in the
that most households consume relatively frequently and same product category (e.g., Brawny and Scott paper tow-
consistently throughout the year. Products that are bought els) or two products in different product categories (e.g.,
batteries and paper towels) during the same shopping trip, families enjoy eating them over the holidays, or more soup
each of these products would have its own line in the data, because it is just generally colder in December than in
and we control for any dependency by including a trip con- April. To control for this potentially shared seasonality, we
trol in our model. However, all units purchased of the same included a measure of how sick people were, on average,
product in the same shopping trip (e.g., a purchase of five in any given week of the year (an effect of illness within
cans of private label chicken noodle soup in the same shop- week of the year). Also, the datasets we use were newer to
ping trip) would be included in the same row. states, retailers, and hospitals in 2009 than they were in
Assessing Familiarity. The panel data provides the 2014, and companies often introduce more products over
household’s entire purchase history in each category in our time. We therefore included a variable that controls for in-
time frame. We operationalize familiarity based on the creased compliance in reporting over time as well as
share a given product takes up within the household’s pur- increases in sales volume over time (an effect of time). We

chase history for the category. For example, if a household included a variable that controls for price (an effect of
buys Scott paper towels 90 times and Brawny paper towels price, in equivalent units) and variables that control for
10 times, Scott has a 90% share, and is therefore the more- product category and household-level effects.
familiar product in the category, whereas Brawny has 10% To see whether the presence of disease increased pur-
share and is less familiar. We compute familiarity using chases of products familiar to the household (hypothesis
the entire time frame rather than a training or calibration 2), we predict purchase quantity by our disease index, the
year to account for the fact that new products are intro- household’s familiarity with the product, and their interac-
duced throughout the time frame and that households tion. We log-transformed these variables to account for
might sometimes grow to be more familiar with different strong positive skew. We report the core effects of interest
brands over time. By conceptualizing familiarity as share in the model results section below, but additional details,
of category, we are also able to account for the fact that including separate analyses for each product category, are
some households are more familiar with private label prod- found in the web appendix.
ucts than national brand products.
Operationalizing familiarity this way raises one potential
issue. When a household only buys one product in a cate-
gory across the entire 6-year time period, the single pur-
Results
chase is considered as familiar (1/1 ¼ 100%) as a We found a significant effect of the disease index.
household that loyally purchases the same product 100 Households bought more overall when disease was more
times in the data set (100/100 ¼ 100%). To address this sit- present in the state (b ¼ .003, SE ¼ 0.0002, t ¼ 16.97, p <
uation, we removed observations where a household .001). Supporting our predictions, this effect was qualified
bought only a single product in the category during our by a significant disease familiarity interaction (b ¼ .005,
time frame (n ¼ 46,794). The final dataset consists of SE ¼ 0.0009, t ¼ 5.99, p < .001). The positive sign of the
12,196,440 purchase instances across our categories of in- coefficient means that the positive effect of disease in-
terest from January 2009 to December 2014. creased for products that were more familiar to the house-
Model and Controls. To determine if the disease index hold, supporting hypothesis 2.
predicted asymmetries in the consumer choices between
more-familiar and less-familiar products, we used a
multilevel-modeling framework with restricted maximum Discussion
likelihood estimation (Bates et al. 2015). This framework
Across four product categories, the present analysis indi-
allows us to control for a series of variables that could in-
cates that the presence of disease can systematically alter
fluence our results. In particular, we wanted to control for
consumer behavior in a real way that is practically relevant
the fact that some states might have higher baseline illness
and sales than others as well as border effects—the fact to marketers. Within households in the United States, the
that the level of illness and preference for particular prod- presence of disease increases purchase quantity overall and
ucts in New York at any particular time is likely more cor- leads to an asymmetric preference for familiar products.
related with illness and preference in New Jersey than in We cannot, however, assess whether the changes in con-
California. We used regional assignments given by the sumption patterns are driven by consumers who fear be-
CDC called Health and Human Service regions to create coming sick, as we propose, or by consumers who are
an effect of state within region, allowing us to control for actually sick. We also cannot account for differences in the
these two issues. The flu is a seasonal illness but sales are availability of products that might be influencing purchase
seasonal as well. For example, people are sicker in behavior. Likewise, the empirical analysis lacks a direct
December than they are in April and also might unrelatedly measure of consumer appraisals and emotion. Our experi-
buy more cookies in December than in April because ments help address these limitations.
STUDY 2: CONTAGIOUS DISEASE, emotion words were four items that measured disgust
DISGUST, AND FEAR (“Unclean,” “Disgusted,” “Dirty,” “Revolted;” a ¼ .89).
Likewise, five items measured fear (“Scared,” “Anxious,”
We designed study 2 to test whether cues of contagious “Afraid,” “Nervous,” “Vulnerable;” a ¼ .95).
disease elicit a joint disgust and fear response in consumers Measuring Preference. We measured preference for
(hypothesis 1a) and to assess whether this joint emotional the familiar under the guise of a second study interested in
response causes the asymmetric increase in preference for grocery shopping and food habits. First, all participants
familiar products (hypothesis 2). In addition, we use a non- were shown a list of 14 common grocery store items and
contagious disease as a secondary control condition to rule were asked to choose (by dragging and dropping) the five
out the alternative possibility that the pattern of results items they would be most likely to buy on their next shop-
found in the empirical analyses represents consumer re- ping trip. The 14 items were presented all together in a ran-

sponse to illness in general rather than contagious diseases dom order to participants but actually consisted of seven
specifically. While both contagious and noncontagious dis- pairs of items from the same product category that differed
eases might increase appraisals of uncertainty and feelings in their degree of familiarity to a Western audience. For ex-
of fear over the control, we propose that only contagious ample, one pair represented common starches and included
diseases should increase both appraisals that pathogens are “Potatoes”—an ingredient very common in a Western
present and that participants have little control over out- household—and “Rice”—an ingredient less common in a
comes, therefore evoking both fear and disgust. By making traditionally Western household and more familiar to an
the threat seem more imminent, experiencing both emo-
Asian or Hispanic household. Other examples of pairs in
tions together should lead to an asymmetric preference for the choice set included spices: “Oregano” versus “Curry
familiar products in service of regaining control over out- Powder” and sauces: “Ketchup” versus “Soy Sauce.” A
comes. Thus, we predict that the asymmetric preference for pretest (N ¼ 53, Mage ¼ 34.32, 44% females) confirmed
familiar products should be present when reading about
that the items coded as familiar to a Western household
contagious disease but not noncontagious disease.
were indeed considered more familiar to a Western house-
hold (M ¼ 5.45, SD ¼ 0.63) than the non-familiar items
Method (M ¼ 2.64, SD ¼ 1.00; F(1, 52) ¼ 253.52, p < .001, g2 ¼
Design. Two hundred twenty-six participants from .83). Finally, participants completed survey demographics
Amazon’s Mechanical Turk were randomly assigned in a (including their own cultural heritage and a multicultural
three-cell between-subjects design with cue (contagious ideology scale; Berry and Kalin 1995) and were debriefed.
disease, noncontagious disease, control) as the manipulated
factor. Results
Manipulating Emotions. After consenting to partici- Exclusion. Of the 226 participants, 24 did not follow
pate, all participants read a short news report. In the control instructions in the basket composition task (either put no
condition, participants read about a new pressure cooker items in their basket (n ¼ 2) or put more than five items in
called the “Instant Pot.” In the noncontagious disease cue their basket (n ¼ 22)) and were excluded from all analyses.
condition, participants read a report about the dangers of The remaining participants (N ¼ 202, Mage ¼ 36.05, 59.9%
heart disease: “the Lead Killer” of people in the United females) are included in analyses below.
States. In the contagious disease cue condition, participants
Emotions. A one-way analysis of variance revealed a
read about the spread of the flu in the United States this
significant effect of experimental condition on partici-
season. Thus, participants in both the noncontagious and
pants’ experienced fear (F(2, 199) ¼ 29.44, p < .001) and
contagious disease conditions read about a serious disease,
experienced disgust (F(2, 197) ¼ 17.86, p < .001).
but because heart disease does not make impurities salient,
Planned contrasts revealed that both participants who read
only the flu condition should elicit both disgust and fear.
about the flu (M ¼ 3.36, SD ¼ 1.74; t(199) ¼ 7.35, p <
All news stories were sourced from real information on the
.001, d ¼ 1.29) and who read about heart attacks
product or respective disease and were equal in length.
(M ¼ 2.91, SD ¼ 1.58, t(199) ¼ 5.50, p < .001, d ¼ 1.05)
Complete language of the news stories is available in the
felt more fear than participants who read about the instant
web appendix.
pot (M ¼ 1.50, SD ¼ 1.07). The flu and heart attack condi-
Measuring Emotions. After reading the news report, all tions were marginally different from each other (t(199) ¼
participants responded to an emotions index adapted from 1.72, p ¼ .086, d ¼ .27). As expected, participants who
prior research where they indicated on a scale from 1 ¼ read about the flu also felt more disgust (M ¼ 2.47, SD ¼
“not at all” to 7 ¼ “very much” the extent to which they 1.50) than participants in the control condition (M ¼ 1.32,
felt each of 20 emotion words during the news report task SD ¼ 0.76; t(199) ¼ 5.92, p < .001, d ¼ .97). The heart at-
(Galoni and Noseworthy 2015). Embedded in the 20 tack condition also elicited greater disgust over the control
condition (M ¼ 1.74, SD ¼ 1.03; t(199) ¼ 2.14, p ¼ .034, was not the behavioral response to illness in general, but
d ¼ .46). Importantly, however, participants reading about rather the fear and disgust arising from contagious disease
the flu experienced significantly more disgust than partici- cues driving the effect. However, heart disease is an illness
pants reading about heart attacks (t(199) ¼ 3.66, p < .001, that can be managed through diet and it is possible that par-
d ¼ .57), which supports hypothesis 1a. ticipants in this condition are choosing more non-familiar
products because they are perceived as healthier.
Basket Composition. To assess whether asymmetric
preference for the familiar changes as a function of which
news story participants read, we first formed an index STUDY 3: USING A DIFFERENT
wherein we computed the proportion of culturally familiar CONTROL
items participants put in their baskets during the shopping
task. From the demographics collected, we had 180 self- We designed study 3 to replicate and extend study 2 us-

identifying North American or European participants and ing a different secondary control and a modified version of
22 self-identifying Asian or Hispanic participants in the the basket composition task. By removing constraints on
sample. We coded the traditionally Western basket items the number of items participants could place in their bas-
as familiar for participants who identified as North kets, we can assess the hypothesis that the threat response
American or European and the non-Western items as cul- to cues of contagious disease should asymmetrically in-
turally familiar for participants who identified as Asian or crease preference for familiar products (hypothesis 2)
Hispanic. In this way, a higher number on the index indi- when consumers are free to choose what and how much
cates that the participant’s basket contained a higher pro- they buy. This change to the basket composition task also
portion of their own culture’s familiar ingredients. allows us to test whether experiencing disgust and fear to-
We expect that participants who read about the flu gether increases action toward eliminating the threat (hy-
would be more likely to choose culturally familiar ingre- pothesis 1c). To the extent that consuming products is a
dients than participants in the heart attack or control condi- form of taking action, we might expect that participants
tions (hypothesis 2). A one-way analysis of variance feeling both disgusted and afraid would put more products
supported this prediction with a significant effect of experi- in their basket, overall. This experiment also included a
mental condition on the basket composition index (F(2, second measure of preference for familiarity in which we
199) ¼ 3.26, p ¼ .040). Planned contrasts revealed that asked participants to choose a restaurant that varied on
participants who read about the flu chose a significantly how familiar it was to the average American. We expected
greater proportion of culturally traditional ingredients (M that participants encountering contagious disease cues
¼ 0.62, SD ¼ 0.17) over participants in the control (M ¼ would show a greater preference for more-familiar
0.56, SD ¼ 0.19; t(199) ¼ 2.04, p ¼ .043, d ¼ .33) and restaurants.
heart attack conditions (M ¼ 0.55, SD ¼ 0.20; t(199) ¼
2.35, p ¼ .02, d ¼ .38), as we predicted. The control and Method
heart attack conditions did not differ from each other Design. Three hundred fifty-four participants from
(t(199) ¼ .35, p ¼ .724, d ¼ .05). Participants who read Amazon’s Mechanical Turk were randomly assigned in a
about contagious disease chose significantly more cultur- three-condition between-subjects design with reminder
ally traditional ingredients than participants in the other (contagious illness, unemployment risk, control) as the ma-
conditions, supporting hypothesis 2. These results hold nipulated factor. We also had a second exploratory factor
both when we run the analysis on all participants, as in this experiment manipulating geographic location of the
reported, and when controlling for any differences between reminder; there were no significant differences across loca-
Western and non-Western participants (details are in web tions within reminder condition; however, so for simplicity
appendix). in reporting we have collapsed this factor into a single fac-
tor design.
Discussion Manipulating and Measuring Emotion. Like study 2,
Study 2 provides evidence that cues of contagious dis- participants in the contagious disease condition read a short
ease can increase both fear and disgust. Study 2 also pro- news story about the spread of the flu and participants in
vides support for the hypothesis that the behavioral the control condition read about the “Instant Pot.” To avoid
response to cues of contagious disease is to asymmetrically the effects that heart disease might have had on food
increase preference for familiar products. Indeed, partici- choice, we manipulated a similarly negative but nonconta-
pants reading about the flu chose significantly more cultur- gious reminder in this experiment using a news report
ally familiar ingredients than participants reading about about the current state of unemployment. Following the
noncontagious heart disease or the instant pot. We used manipulation, all participants responded to the same emo-
heart disease as a secondary control to make sure that it tions index included in study 2 (fear a ¼ .93; disgust a ¼
.90). We included an additional five emotion words (M ¼ 1.44, SD ¼ 0.81; t(317) ¼ 2.73, p ¼ .007, d ¼ .51).
(“Relieved,” “Lucky,” “Grateful,” “Fortunate,” “Secure”) There were no significant differences in disgust between
designed to measure how relieved participants might have participants in the control and unemployment fear condi-
felt after reading the news reports. We included this mea- tion (M ¼ 1.74, SD ¼ 1.15; t(317) ¼ 1.51, p ¼ .132, d ¼
sure, as well as questions in the demographics assessing .32). Thus, the unemployment reminder elicited more fear,
whether participants had received a flu shot and whether but not disgust, over the control, whereas the flu reminder
they were currently employed, to control for any variation elicited significantly more disgust and fear over the control
in actual circumstances that could have been influencing (hypothesis 1a).
the results. The measure of relief had no effect on results
Preference. We tested whether the flu news report led
and is thus not included in the analysis.
to taking greater action toward the threat (hypothesis 1c)
Measuring Preference. We measured preference for and an asymmetric preference for familiar products (hy-

the familiar in the same way as in study 2 with minor mod- pothesis 2). We had 294 self-identifying North American
ifications. Participants saw a list of 20 food items (three or European participants and 26 self-identifying Hispanic
new pairs of fresh ingredients) and were asked to simply or Asian participants. Like in study 2, we reverse-coded
indicate what they would be most likely to buy and least the pairs of grocery items for participants who self-
likely to buy on their next shopping trip. We also asked identified as Asian or Hispanic as the alternative item in
participants to explain why they made the choices they did. the pair would be more familiar to these cultures. To assess
As a second measure of preference for familiarity, partici- whether reading the flu news report changed the number of
pants also saw a set of four restaurant choices familiar items in the basket (hypothesis 1c), we used the
(“American,” “Mexican,” “Malaysian,” and “Korean- number of familiar items participants indicated they would
Mexican”) and were asked to choose their likely dining be likely to buy on their next shopping trip as our depen-
choice among the options. Participants then completed a dent variable of interest. As in study 2, we took the propor-
set of demographics and were debriefed to the true nature tion of the basket that was traditional to the participant to
of the study. assess whether reading the flu news report led to an asym-
metric preference for familiar products (hypothesis 2).
Results To test our predictions, we ran a one-way analysis of
variance on these two variables with reminder condition as
Exclusion. Of the 354 participants, 34 gave straight-
the predictor, this time contrasting the flu reminder condi-
line responses to all emotions, did not choose a restaurant,
tion against both the control and the unemployment re-
or did not give any explanation for the choices they made
minder conditions. This analysis revealed an effect on the
in the basket composition task. The remaining participants
number of items in the basket (F(2, 317) ¼ 5.29, p ¼
(N ¼ 320, Mage ¼ 36.35, 46.3% females) are included in all
.005). Participants in the flu reminder condition put signifi-
analyses below. The analyses including employment secu-
cantly more items in their baskets (M ¼ 3.72, SD ¼ 1.48)
rity and flu vaccination as covariates had the same pattern
than both participants in the control (M ¼ 3.07, SD ¼ 1.32;
of results. We report the results not including covariates
t(317) ¼ 2.69, p ¼ .008, d ¼ .46) and unemployment
below.
(M ¼ 3.27, SD ¼ 1.39; t(317) ¼ 2.65, p ¼ .008, d ¼ .31)
Emotions. To assess whether the flu news report eli- conditions. Thus, participants reading about the flu indi-
cited both fear and disgust and whether the unemployment cated that they would be likely to choose more of the items
news report elicited fear alone against the control, we ran a on the list than participants reading about unemployment
one-way analysis of variance with reminder condition as or participants in the control condition.
the predictor and fear and disgust as dependent variables. To explore whether our results replicated study 2, we ex-
This analysis revealed a significant effect of condition on amined the difference between participant choices in the
experienced fear (F(2, 317) ¼ 20.45, p < .001). In line flu reminder, unemployment, and control conditions. The
with study 2, the planned contrasts revealed that both the predicted effect on asymmetric preference for the familiar
flu news report (M ¼ 2.95, SD ¼ 1.66; t(317) ¼ 5.40, p < was significant (F(2, 316) ¼ 3.49, p ¼ .032). Participants
.001, d ¼ 1.07) and the unemployment news report in the flu reminder condition chose a significantly greater
(M ¼ 3.19, SD ¼ 1.53; t(317) ¼ 6.32, p < .001, d ¼ 1.34) proportion of familiar items (M ¼ 0.67, SD ¼ 0.21) than
made people feel more afraid than participants in the con- both participants in the control (M ¼ 0.58, SD ¼ 0.29,
trol condition (M ¼ 1.54, SD ¼ 0.84). Likewise, the analy- t(316) ¼ 2.28, p ¼ .023, d ¼ .36) and unemployment re-
sis revealed a significant effect of condition on minder (M ¼ 0.61, SD ¼ 0.21; t(316) ¼ 2.04, p ¼ .042, d
experienced disgust (F(2, 317) ¼ 4.07, p ¼ .018). Unlike ¼ .29) conditions, as predicted.
experienced fear, the planned contrasts revealed that only Finally, we assess whether reminder condition influ-
the flu news report (M ¼ 2.01, SD ¼ 1.35) made people enced restaurant choice. A pretest (N ¼ 50, Mage ¼ 34.08,
feel significantly more disgusted than the control condition 46% females) confirmed that the restaurant choices
linearly decreased in familiarity to a North American audi- response—and not other features of the flu specifically—
ence (see web appendix), so we treated restaurant choice as that are driving the behavioral response to contagious dis-
a continuous variable where lower values represent a pref- ease cues. To do so, we use a different disease (shingles),
erence for familiarity. A one-way analysis of variance on hold disease constant, and orthogonally manipulate
restaurant choice with reminder condition as the predictor whether news about the disease affords appraisals of salient
revealed a marginally significant effect of condition (F(2, contaminants and uncertainty over outcomes. Second, we
317) ¼ 2.58, p ¼ .077). The planned contrasts revealed assess the generalizability of our effect to nonfood catego-
that participants reading about the flu did indeed choose ries. To do so, we examine preference for a pair of head-
more-familiar restaurants (M ¼ 2.09, SD ¼ 1.01) over par- phones that are either framed as a familiar and trusted
ticipants in the control condition (M ¼ 2.49, SD ¼ 1.14; brand or as a novel and innovative brand. We expect that
t(317) ¼ 2.27, p ¼ .024). However, participants reading when reading about the shingles causes appraisals of out-

about the flu did not prefer the more-familiar restaurant come uncertainty and makes impurities salient (hypothesis
over participants reading about unemployment (M ¼ 2.19, 1a), consumers will show an asymmetric preference for the
SD ¼ 0.98; t(317) ¼ 0.81, p ¼ .418). Given that unemploy- headphones framed as familiar (hypothesis 2).
ment is often a contentious issue with respect to immigra-
tion policy (despite evidence to the contrary; American Method
Immigration Council 2013; Okkerse 2008), it is possible
that participants in the unemployment reminder condition Design. Six hundred participants recruited from
were responding to this contention by choosing what they Amazon’s Mechanical Turk were randomly assigned in a 2
thought was a more local restaurant. While understanding (impurities: salient vs. not) 2 (outcome: certain vs.
this shift is outside of scope of this work, this explanation uncertain) 2 (product framing: familiar vs. novel)
is an interesting possibility worthy of future research. between-subjects design where they read a short scenario
about the shingles and then rated a pair of headphones that
Discussion were framed as familiar or novel.
The results of study 3 replicated study 2’s findings and Manipulating Appraisals. All participants first read a
provided additional evidence for the unique effect of conta- short vignette that asked them to imagine there had been an
gious disease cues on consumption behavior using a differ- outbreak of shingles spreading in their city. To manipulate
ent negative control condition (unemployment) and a whether the vignette made contaminants salient, half of
modified basket composition task. In particular, study 3 participants read about the visible symptom that the shin-
provides support for the hypothesis that contagious disease gles can cause (blisters, scabs, sweating, nausea) whereas
cues can increase action toward eliminating the threat (hy- the other half of participants read about the more internal,
pothesis 1c) and can asymmetrically increase preference non-visible symptoms that the shingles can cause (head-
for the familiar (hypothesis 2). aches, body chills, muscle aches, fever). A pretest (N ¼ 51,
Studies 2 and 3 together provide convergent evidence Mage ¼ 32.94, 47.1% females) established that participants
that cues of contagious disease asymmetrically increase considered both sets of symptoms to be equally severe
preference for the more familiar over the less familiar, (Mvisible ¼ 4.14 vs. Minvisible ¼ 4.18; F(1, 50) ¼ 0.07, p ¼
compared to cues that induce fear alone. One remaining is- .793), but thought the visible symptoms made the presence
sue is that the effect might be driven by disgust alone. We of disease more apparent (M ¼ 5.66, SD ¼ 0.95) than the
propose that the behavioral response to cues of contagious non-visible symptoms (M ¼ 3.10, SD ¼ 1.11; F(1, 50) ¼
illness is a joint disgust and fear response that amplifies the 204.68, p < .001, g2 ¼ .80).
behavioral tendency of disgust to avoid and the behavioral To manipulate appraisals of uncertainty over outcomes,
tendency of fear to approach comforting familiar products. we capitalized on the common (but false) belief that the
To assess that both emotional components are necessary to shingles is potentially contagious. As established in study
explain the pattern of results, subsequent experiments ma- 2, if a disease is not contagious, it is certain that the con-
nipulate whether participants are feeling fear, disgust, nei- sumer will not catch the illness and should thus not evoke
ther, or both. fear. In contrast, if a disease is contagious, there is a level
of uncertainty in whether the consumer will become sick.
STUDY 4: SALIENCE OF Half of participants read that the shingles were contagious
CONTAMINANTS AND OUTCOME and that a close coworker arrived at work complaining of
UNCERTAINTY symptoms. For these participants, the potential risk of
catching the shingles provides a source of uncertainty over
The purpose of study 4 is twofold. First, we assess the outcome that might produce fear. The other half of par-
whether it is indeed the appraisals of salient contaminants ticipants read the truth about shingles: that they are a non-
and outcome uncertainty causing the joint disgust and fear contagious disease caused by harboring a dormant virus.
These participants also read that a close coworker arrived attention check. The remaining participants (N ¼ 516, Mage
at work complaining of symptoms, but that their doctor ¼ 37.69, 49.6% females) are included in the analysis
had recently told them they did not have the dormant below.
disease.
Preference. We expected participants appraising both
We ran a pretest (N ¼ 160, Mage ¼ 36.98, 55% females)
outcome uncertainty and salient contaminants to show an
to establish that the manipulation of uncertainty in these
asymmetric preference for the Beats headphone framed as
scenarios elicited fear whereas the manipulation of symp-
traditional. To assess this hypothesis, we ran a three-way
tom visibility in these scenarios elicited disgust. A two-
analysis of variance on the average rating of the three lik-
way ANOVA revealed a significant main effect of symp-
tom visibility on perceived disgust (F(1, 155) ¼ 123.18, p ing items (a ¼ .91). None of the lower-order interactions
< .001, g2 ¼ .44) such that visible symptoms produced or main effects were significant except for a marginal out-
come uncertainty familiarity interaction (F(1, 506) ¼

more disgust (M ¼ 5.83, SD ¼ 1.23) than non-visible
symptoms (M ¼ 3.18, SD ¼ 1.75). Importantly, there was 3.03, p ¼ .082, g2 ¼ .006). However, this analysis revealed
no main effect of visibility of symptoms on experienced the predicted three-way outcome uncertainty contami-
fear (Mvisible ¼ 4.80 vs. Minvisible ¼ 4.25, ns). Likewise, the nant salience familiarity interaction (F(1, 506) ¼ 4.28, p
analysis revealed a main effect of outcome control on ex- ¼ .039, g2 ¼ .008). This interaction was such that when
perienced fear (F(1, 155) ¼ 38.32, p < .001, g2 ¼ .20) the outcome was known and certain (no possibility of
such that the potential risk of catching the disease led to catching shingles), there were no main effects of impurity
more fear (M ¼ 5.43, SD ¼ 1.63) than when there was no salience or familiarity on preferences (ps > .387).
potential risk (M ¼ 3.63, SD ¼ 2.04). Importantly, there However, when there was a risk of catching the shingles,
was no main effect of outcome control on disgust (Mcertain we found a significant main effect of familiarity (F(1, 506)
¼ 4.38 vs. Muncertain ¼ 4.68, ns). Thus, holding disease ¼ 5.28, p ¼ .022, g2 ¼ .010) such that participants
constant, the salience of impurities made the disease cue experiencing outcome uncertainty preferred the familiar
more disgusting whereas outcome uncertainty made the headphones (M ¼ 4.70, SD ¼ 1.77) to the novel head-
disease cue scarier. phones overall (M ¼ 4.20, SD ¼ 1.80). This main effect
was qualified by the predicted contaminant salien-
Measuring Preference. All participants read about and ce familiarity interaction (F(1, 506) ¼ 4.36, p ¼ .038, g2
rated the newest release of Beats headphones. For half of ¼ .016). As is evident in figure 2, planned contrasts
participants, we framed the headphones as familiar and revealed that when the outcome was uncertain but contami-
trusted whereas for the other half we framed the head- nants were not salient, participants liked the headphones
phones as novel and innovative. Participants read:
equally regardless of whether they were framed as familiar
These new Beats by Dr. Dre headphones are from one of the (M ¼ 4.35, SD ¼ 1.81) or novel (M ¼ 4.31, SD ¼ 1.80;
[most familiar and trusted/ newest and most innovative] F(1, 506) ¼ 0.22, p ¼ .882). It was only when the outcome
brands of headphones in the world. These Beats use a [long- was uncertain and when contaminants were salient that
standing and trusted high-fidelity sound/ innovative and participants demonstrated a significant asymmetric prefer-
untested bone conductance] technology to provide crisp ence for the headphones framed as familiar (M ¼ 5.05, SD
sound quality and deep, clean bass notes to listeners. ¼ 1.65) over the headphones framed as novel (M ¼ 4.10,
[Traditional/Novel], high quality sound – brought to you by SD ¼ 1.81; F(1, 506) ¼ 9.64, p ¼ .002, g2 ¼ .019).
Beats.
After reading the information, participants rated the Discussion
headphones on how much they liked, were interested in Study 4 provides convergent evidence that cues of con-
learning more about, and their likelihood of buying the tagious disease asymmetrically increase preference for the
headphones on 7-point scales (where 1 ¼ not at all to familiar through the appraisals these cues afford. We repli-
7 ¼ very much). As an attention check, we had participants cate our core effect while holding disease constant and or-
indicating which brand of headphones they had rated at the thogonally manipulating the appraisals that cues about the
end of the study.
disease allow. When the contagious disease cue made con-
taminants salient, it elicited disgust, and when the cue
Results made the outcome appear uncertain, it elicited fear. These
Exclusion. Of the 600 participants, 40 cases came from two appraisals together caused the joint fear and disgust re-
the same geo-locations and were removed due to the suspi- sponse in consumers, which lead to the unique behavioral
cion of being fraudulent entries (Kennedy et al. 2018). An tendency to asymmetrically value the familiar. Study 4,
additional 46 participants were excluded for indicating sus- consistent with the empirical analysis of the household
picion or knowledge about the shingles or for failing to data, provides evidence that this effect extends beyond cul-
identify the brand of headphones, therefore failing the turally bound or food-related items. Thus, the joint fear
FIGURE 2 outcome uncertainty as study 4, measure participants’ per-

ceptions of controllability over outcomes, and have partici-
BOTH APPRAISALS OF OUTCOME UNCERTAINTY AND pants engage in the basket composition task used in study
CONTAMINANT SALIENCE LEAD TO ASYMMETRIC
PREFERENCES FOR FAMILIAR HEADPHONES.
3. We expect that perceived control over outcomes will be
lowest when people are feeling both disgusted and afraid
(hypothesis 1b) and that this perceived lack of control will
mediate the relationship between emotion and subsequent
preference for the familiar (hypothesis 2).
Method

Design. We recruited 240 participants from Amazon’s
Mechanical Turk and assigned them randomly in a 2 (con-
taminants: salient vs. not) 2 (outcome: certain vs. uncer-
tain) between-subjects design where they read a short
scenario about the shingles and then completed a basket
composition task.
Manipulating Appraisals. We manipulated appraisals
using the same set of vignettes as study 4 where partici-
pants read about the shingles. To manipulate disgust, we
altered whether the disease was described as producing
very visible contamination symptoms (e.g., vomiting) or
invisible symptoms (e.g., fever) and therefore whether con-
taminants were salient. To manipulate fear, we altered
whether the disease was contagious (and therefore getting
sick was uncertain) or not contagious (thus participants
were certain they would not get sick).
Measuring Perceived Controllability. We measured
perceived control over outcomes using a three-item scale
adapted from Kushner et al. (1993). These items included
“I always control whether or not bad things happen to me,”
“When bad things happen, I can always make things come
out the way I want them to,” and “No matter what happens,
I am always in control of my situation” (all measured on 7-
point scales where 1 ¼ “strongly disagree” to 7 ¼
“strongly agree;” a ¼ .90). The presentation of the per-
ceived control measure was counterbalanced with the pre-
sentation of the basket composition-dependent variable to
account for any order effects.
and disgust response to contagious disease cues increases Measuring Preference for Familiarity. To measure
preference for the familiar irrespective of the xenophobia preference for familiarity, we used the same unconstrained
or cultural preference that past research on disgust has basket composition task used in study 3 wherein partici-
shown. pants were asked to choose which items they would be
most likely to buy from a list of 20 grocery items. The
STUDY 5: THE MEDIATING ROLE OF items were actually 10 pairs of products where one product
PERCEIVED CONTROLLABILITY was more familiar to a Western audience than the other
(e.g., Frank’s Red Hot vs. Sriracha). We expected that par-
We propose that, when consumers feel that a threat is ticipants feeling both disgusted and afraid would put more-
imminent and that the outcome is uncertain and beyond familiar items in their baskets than participants feeling ei-
their control, consumers take actions to restore control (hy- ther emotion alone (hypothesis 2). The survey concluded
potheses 1b and 1c). To assess this possibility, in study 5, with demographics and an attention check asking partici-
we use the same manipulation of contaminant salience and pants to select which disease they had just read about.
Results g2 ¼ .000). In contrast, when the symptoms were disgust-

ing, people felt like they had significantly less control over
Exclusion. Of the 240 participants, the data from 21
outcomes when the outcome was uncertain (M ¼ 3.58, SD
were removed for indicating knowledge about the shingles,
¼ 1.42) than when it was certain (M ¼ 4.30, SD ¼ 1.44;
failing the attention check, or were from duplicate geo-
F(1, 215) ¼ 6.20, p ¼ .014, g2 ¼ .028). Thus, participants
locations tagged as potentially fraudulent. The remaining
felt like they had the least control over outcomes when
participants (N ¼ 219, Mage ¼ 34.69, 32% females) are rep-
they were feeling both disgusted and afraid.
resented in all analyses below.
Mediation. To check whether perceived controllability
Basket Composition. In line with prior studies, six par-
mediated the relationship between emotion and basket
ticipants indicated that they were from East Asian or
composition choices, we ran a moderated mediation model
Hispanic cultures and we reverse-coded familiarity for
where we predicted share of familiar items in the basket by

these participants. We ran a two-way ANOVA predicting
outcome uncertainty and contamination salience, mediated
the number of familiar items in the basket. This analysis
revealed a significant main effect of contaminant salience by perceived controllability (Hayes 2012; model 7; 5,000
(F(1, 215) ¼ 4.45, p ¼ .036, g2 ¼ .020) such that people draws). This analysis revealed a significant index of mod-
feeling disgusted put significantly more-familiar items in erated mediation, 95% CI [0.001; 0.018]. Supporting hy-
their baskets (M ¼ 6.16, SD ¼ 2.19) than people not feeling potheses 1b and 2, feeling both disgusted and afraid
disgusted (M ¼ 5.49, SD ¼ 2.45). However, replicating reduced perceived control over outcomes (hypothesis 1b),
study 3, this main effect was qualified by a marginal con- which subsequently increased the share of familiar prod-
taminant salience outcome uncertainty interaction (F(1, ucts participants indicated they would purchase (hypothesis
215) ¼ 3.21, p ¼ .075, g2 ¼ .015). Given that our contrasts 2; full mediation results are available in the web appendix).
were planned, it is appropriate to conduct them despite the
fact that the omnibus test did not achieve conventional lev- Discussion
els of statistical significance (Hsu 1996; see also Hancock Study 5 examines the mechanism underlying the joint
1997 for discussion). These planned contrasts revealed that disgust and fear response on subsequent behavior. We ar-
when the outcome was certain, people chose the same gued that experiencing disgust and fear together in re-
number of familiar items whether the symptoms were dis- sponse to contagious cues makes the threat seem
gusting (M ¼ 5.82, SD ¼ 2.21) or not (M ¼ 5.72, SD ¼ imminent, increasing perceived lack of control over out-
2.30; F(1, 215) ¼ 0.05, p ¼ .823, g2 ¼ .000). In contrast, comes, which increases preference for the familiar. Our
when the outcome was uncertain, people put significantly results support our prediction.
more-familiar items in their basket when the symptoms
were also disgusting (M ¼ 6.44, SD ¼ 2.14) than when the
STUDY 6: ASSESSING BEHAVIOR
symptoms were not (M ¼ 5.22, SD ¼ 2.60; F(1, 215) ¼
7.66, p ¼ .006, g2 ¼ .034). We ran this same analysis on ACROSS THE UNITED STATES
the proportion of familiar items in the basket and found the So far, we have demonstrated that contagious disease
same pattern of results (details are available in web cues elicit fear and disgust, highlighting a lack of control
appendix). over outcomes, so consumers asymmetrically prefer the fa-
Perceived Controllability. We expected that partici- miliar over the unfamiliar. We have shown this effect
pants would feel like they most lacked control when they across four product categories in the Nielsen consumer
were feeling both disgusted and afraid. To assess this pos- panel data. In study 6, we use the same disease index as
sibility, we ran a two-way ANOVA predicting perceived study 1 alongside the Nielsen retail scanner data to exam-
controllability. This analysis revealed a marginal main ef- ine how preferences shift across the country.
fect of outcome uncertainty on perceived controllability
such that when the outcome was uncertain, people felt like Data Treatment
it was also less controllable (M ¼ 3.69, SD ¼ 1.37) than
An Index of Disease Presence. We use the same time
when the outcome was certain (M ¼ 4.03, SD ¼ 1.61; F(1,
period (6 years: January 2009–December 2014) and dis-
215) ¼ 3.07, p ¼ .081, g2 ¼ .014). Important to our core
ease index used in study 1.
hypothesis, this main effect was qualified by an outcome
uncertainty contaminant salience interaction (F(1, 215) Measuring Sales of Familiar and Non-Familiar
¼ 3.22, p ¼ .074, g2 ¼ .015). This interaction was such Products. The Nielsen retail scanner data is a very large
that when the symptoms were not disgusting, there were no data set. To maintain a feasible amount of computing
significant differences in perceived controllability whether power required for the analysis, we limit our analysis to
the outcome was certain (M ¼ 3.81, SD ¼ 1.72) or uncer- one within-brand comparison and one within-category
tain (M ¼ 3.82, SD ¼ 1.31; F(1, 215) ¼ 0.001, p ¼ .976, comparison. For consistency with study 1, we analyzed
unit sales from one national cookie brand (OREOs) and therefore more popular, there was an effect of familiarity
one entire product category (canned soup) across more such that people bought fewer non-familiar products than
than 35,000 retail outlets in the United States.1 familiar products on average (b ¼ .32, SE ¼ 0.008, t ¼
We do not have individuals’ shopping histories on which 39.31, p < .001). However, both of these effects were
to build an individualized familiarity variable in this data qualified by a significant familiarity illness interaction (b
set, so we operationalize familiarity in this analysis by us- ¼ .08, SE ¼ 0.002, t ¼ 45.20, p < .001). This interac-
ing the market leader in the category as the familiar brand. tion is pictured in figure 3 and indicates that in line with
In the soup category, we used sales of the market-leading the hypothesis, people asymmetrically preferred
national brand (Campbell’s) as familiar and sales of any Campbell’s national brand and asymmetrically punished
private label product as less familiar. We coded sales of private label brands when they feared getting sick. While
the traditional OREO as familiar and sales of any product soup sales generally increased as people became sicker, a

with additional attributes beyond the standard OREO (e.g., 10% increase in illness was associated with a 1.6% in-
different flavors, coatings, or stuffing) as less familiar. crease in soup sales for familiar brands but only a 0.1% in-
For both the within-category and within-brand compari- crease in soup sales for non-familiar brands. The average
son, we calculated the total unit sales in a week across all price for a can of private label soup is $1.39 and the aver-
retailers in a particular state. This aggregation provided us age weekly volume is 77,898 units per state. A back of the
with four sales values for each week in each state: two for envelope calculation would indicate that, for every 10% in-
the within-category comparison [the total units of crease in illness, private labels are selling 28,712 fewer
Campbell’s sold (familiar category) vs. the total units of units, losing $39,910 in revenue per week across the
private label soup sold (non-familiar category)] and two United States than they would have in a world where an
for the within-brand comparison [the total units of standard asymmetric preference for the familiar did not exist.
OREOs sold (familiar brand) vs. the total units of nonstan- Similarly, the average price for a can of Campbell’s soup
dard OREOs sold (non-familiar brand)]. The resulting data is $1.89 and the average weekly volume in this time period
set had 30,576 observations: 312 weeks of data 49 states- is 229,131 units per state. By these same calculations, for
2 types of familiarity. every 10% increase in illness, Campbell’s is benefitting
from 84,453 more units sold and $159,616 in revenue per
Model and Controls. To test whether our disease index
week across the United States than they would have if peo-
revealed asymmetries in consumer choices, we used a
ple did not demonstrate this asymmetric preference.
multilevel-modeling framework with restricted maximum
likelihood estimation (Bates et al. 2015). We ran our analy- OREOs. The pattern of results for OREOs was largely
sis of the within-brand and within-category comparisons the same as that for canned soup except the analysis did
separately because we predicted that soup would generally not reveal the same effect of illness on sales on average (b
increase in sales as people became sicker and did not want ¼ .01, SE ¼ 0.007, t ¼ 1.31, ns). However, there was a sig-
to generalize this effect to the within-brand comparison. nificant familiarity illness interaction (b ¼ .05, SE ¼
The model of interest for each comparison included the ef- 0.001, t ¼ 35.71, p < .001). As with canned soup, people
fect of the illness index, the effect of familiarity, and their showed an asymmetric preference toward the familiar
interaction controlling for state, border, seasonality, and OREOs compared to the non-familiar OREOs when they
time effects (see study 1). We report the core effects of in- feared getting sick. A 10% increase in illness was associ-
terest in the model results section below, but methodologi- ated with a 0.61% increase in unit sales of familiar OREOs
cal details including data access instructions are found in but a 0.44% decrease in unit sales for non-familiar
the web appendix. OREOs. Given that a package of non-familiar OREOs
retails for $3.05 and they sell on average 17,879 units per
Results week per state, for every 10% increase in illness, OREO
loses $11,807 in revenue per week across the United States
Canned Soup. The model revealed an effect of illness than they would have in a world where this asymmetric
such that as illness increased, people bought more soup on preference did not exist. However, OREO has a net benefit
average (b ¼ .08, SE ¼ 0.009, t ¼ 9.41, p < .001). from this asymmetric preference for the familiar because
Likewise, given that Campbell’s is the national brand and they sell more-familiar than non-familiar units in general.
This asymmetric increase in preference means that for ev-
1 All results calculated (or derived) based on data from The Nielsen
Company (US), LLC, and marketing databases provided by the Kilts ery 10% increase in illness, OREO earns an additional
Center for Marketing Data Center at The University of Chicago Booth $27,394 per week on sales of traditional OREOs.
School of Business. Copyright V C 2018 The Nielsen Company (US), Replicating the pattern of results within the OREO brand
LLC. All Rights Reserved. The conclusions drawn from the Nielsen is important because OREOs are not a product commonly
data are those of the researchers and do not reflect the views of
Nielsen. Nielsen is not responsible for, had no role in, and was not in- associated with the flu. It is possible that the asymmetric
volved in analyzing and preparing the results reported herein. preference for Campbell’s was not due to people switching
FIGURE 3
THE EFFECT OF PATHOGEN PRESENCE ON SALES VOLUME OF FAMILIAR AND LESS-FAMILIAR SKUS ACROSS CANNED SOUP
AND OREO CATEGORIES.

Note: Data points represent average weekly unit sales of familiar and non-familiar products. Full analysis was run on raw data, and lines are fit lines from the multilevel
model after accounting for the influence of the controls, and therefore capture the unique effect of illness presence on sales. For this reason, we did not expect the fit
lines to capture the individual data points. Soup sales share greater seasonality with the flu than OREO sales, so it makes sense that the actual model fit is more con-
servative than the raw correlation not including controls.
preference, but rather due to new, less frequent consumers disease can have real consequences on shopping habits and
of soup entering the market when they are sick (e.g., cus- product choice.
tomers who only buy soup when they are sick and buy
Campbell’s simply because they have less expertise in the GENERAL DISCUSSION
category and therefore default to the national brand). It is
less likely (both conceptually and empirically given the Across two large empirical data analyses and four
lack of a main effect of illness on OREO sales in this anal- experiments, we find that cues of contagious disease have
ysis) that there exists a parallel segment of consumers who a meaningful and systematic impact on consumption be-
only buy OREOs when they have the flu and prefer tradi- havior. Our results suggest that cues of contagious disease
tional OREOs over nontraditional OREOs because they are evoke appraisals of outcome uncertainty and appraisals of
less familiar with the category. Instead, we can infer from salient contaminants, leading to a joint fear and disgust re-
the increase in unit sales of familiar and decrease in unit sponse in consumers. The joint disgust and fear response
sales of non-familiar OREOs that there is indeed some amplifies the unique behavioral tendencies of each emo-
amount of consumer switching away from the non-familiar tion, producing an asymmetric preference for familiar
and toward the familiar when people are afraid of becom- products and sometimes increasing purchase quantity over-
ing sick. all. Unlike disgust alone, which motivates rejection of the
contaminant, feeling fear and disgust motivates consumers
to seek familiar and helpful products to reestablish a sense
Discussion of control. Unlike fear alone, which can lead to inaction,
feeling disgust and fear together motivates taking action,
Across two product categories, one within-category amplifying the tendency to regain control, and resulting in
comparison and one within-brand comparison, the present an asymmetric preference for more-familiar alternatives.
analysis indicates that the presence of illness can systemat- Our findings contribute meaningfully to understanding
ically alter consumer behavior in a very real way that is how emotions and mixed emotional states can influence
practically relevant to marketers. While an empirical anal- consumption behavior. Most prior research on contagious
ysis lacks a direct measure of consumer emotion, taken to- cues examines the role of disgust specifically on consump-
gether with the results of the laboratory experiments, the tion behavior (Han et al. 2012; Lerner et al. 2004).
present analysis clearly highlights that cues of contagious Likewise, most prior research on emotions has studied
situations where discrete emotions diverge and how they in- Beyond suggesting that mixed emotions can have unique
dividually impact behavioral tendencies (Lerner and Keltner behavioral tendencies and calling for an increased effort in
2001; Lerner et al. 2004). Little research, however, exam- understanding joint emotional states, our findings suggest
ines the behavioral consequences of situations in which peo- that the specific appraisal dimensions that differentiate be-
ple feel both disgust and sadness or both anger and fear. tween discrete emotions in the laboratory might be thought
While the common notion might be that these emotions con- of differently when consumers name and assess their own
tribute discretely to behavior, here we contribute to the liter- emotional experience in daily life. For example, past re-
ature showing that there can be unique and emergent search has shown that consumers can isolate disgust from
behavioral tendencies when multiple emotions converge. fear on the basis of the appraisal dimension of uncertainty:
Cues of contagious disease—by eliciting both fear and dis- people call how they are feeling “fear” when the situation
gust—amplify the behavioral tendencies of each emotion is uncertain, whereas they call how they are feeling

and result in an emergent asymmetric preference for the fa- “disgust” when the situation is certain (Ellsworth and
miliar. These behavioral tendencies cannot be explained by Smith 1988). Situations often have multiple and conflicting
an understanding of fear or disgust, alone. Instead, our sources of information that can influence these appraisals
results suggest that understanding how consumers feel emo- (e.g., certainty that there are impurities present and uncer-
tions simultaneously and how different emotions interact tainty in whether the disease is infectious), and instead of
can reveal the behavioral tendencies that emerge from feel- making tradeoffs to name a specific discrete emotion, the
ing these emotions. present research demonstrates that people can and do inte-
We explore how fear—as an additional response to the grate the information into a joint emotional state. In other
well-known disgust response—can alter the impact of con- words, people can feel disgust while feeling uncertain and
tagious cues on consumption behavior. Many other moder- can feel fear despite some situational certainty. While it is
ators, however, may influence the relative strength of each useful to understand what differentiates the discrete emo-
emotion. For example, an extremely salient impurity (e.g., tions when studying them in isolation, our findings provide
if a consumer can feel the commuter’s sneeze on their evidence that the real experience of emotion is often much
more complex and nuanced and discrete appraisal dimen-
skin) might increase the strength of the disgust response
sions might not be as diagnostic of emotional experience.
relative to the fear response. Likewise, if a consumer is
Understanding the dynamics of joint emotional states is
high in trait anxiety or has a really important interview in
critical to understanding how the holistic assessment of
the following day and cannot afford to get sick, then per-
emotion might influence behavioral tendencies and con-
haps the uncertainty a contagious cue provokes might out-
sumption behavior.
weigh the disgust response. In cases such as these, whether
One lingering issue of interest concerns how consumers
the disgust response might sometimes dominate the fear re-
perceive comfort foods and products in response to conta-
sponse and potentially reduce consumption remains an in- gious disease cues. In studies 1 and 6, we found that the in-
teresting question. A related question for future research is crease in sales held both for products that help combat the
whether there is an order-of-entry effect of different emo- threat like soup, and also for comforting products like
tions. For example, some research has suggested that dis- cookies. Likewise, during the COVID-19 pandemic, in ad-
gust is an immediate and focalized response to a specific dition to items that help control outcomes directly such as
threat, whereas fear—because of the inherent uncertainty hand sanitizer, hand soap, and respirator masks, comfort
in its experience—is a more generalized response to a po- foods seemed to also be in high demand. For instance, one
tential threat (Krusemark and Li 2011; Neuberg et al. reporter noted that beer and Doritos seemed to be finding
2011; Susskind et al. 2008). These differences might mean more and more of an “essential” place in the grocery carts
that certain emotions happen faster and last for shorter of consumers (Bubacz 2020). One potential explanation
durations than others, and these differences might influ- could be that comfort foods help consumers restore a sense
ence the time window in which we observe the unique be- of control in a different way: by providing solace that ev-
havioral tendencies of joint emotional states. Finally, it is erything will be okay. However, this preference for com-
noteworthy that our results found few main effects of fear fort foods could be driven by an entirely different coping
on preference. This finding highlights past research show- mechanism. A supplemental experiment (available in the
ing that the initial behavioral response to fear is inaction in web appendix) manipulated whether products were framed
service of isolating and identifying the threat (Blanchard as helping to control outcomes and found that our effect
et al. 2011). Here, we show that disgust can play a role in only held when the product was designed to help consum-
providing certainty about the source of the threat, spurring ers regain a sense of control. While this provides support
action. However, it would be interesting to explore what for the mechanism of perceived control, it remains to be
factors in other studies led fear to result in action. seen whether or how other methods of restoring control
Exploring these questions is outside of the scope of this ar- (perhaps comfort foods) might influence the effect. Future
ticle but provides a fruitful avenue for future research. research might wish to explore the mechanisms behind
why comforting products seem to be in high demand when the data from the CDC FluView Interactive service in July
consumers are confronted with contagious disease cues. 2016. The first author aggregated and analyzed these data
Our results have important implications for marketing along with the Nielsen retail scanner data under the guid-
practitioners. Many contagious diseases share the same ance of the second and third authors between September
seasonal patterns as the flu, wherein people are usually 2016 and September 2017. The first author received autho-
sicker in the winter than they are in the summer months. rization for access to the Nielsen consumer panel data
The present research suggests that understanding this sea- along with the retail scanner data in July 2016. The con-
sonality might be important for promotion decisions and sumer panel data were analyzed by the first author under
message framing. In particular, if a product is novel, mar- the guidance of the second and third authors between
keters might be better served promoting the product in the September and December 2018. The first author designed,
summer months when people are less likely to encounter collected, and analyzed studies 2 and 3 under the guidance

contagious cues and are more receptive to novelty. If a and feedback of the second and third authors between May
product is familiar, marketers might be better served pro- and December 2017. Study 4 was collected and analyzed
moting in the winter months when people are likely to con- between August and November 2018. Study 5 was col-
sume more, overall, and asymmetrically favor these lected and analyzed between November and December
familiar products. In addition, the present research suggests 2019. These experimental data were shared and discussed
that marketers frame messaging in a way that highlights on multiple occasions by all authors.
the familiar and trusted aspects of a new product in the
winter months and save highlighting the novel features of REFERENCES
the product for the summer months when consumers are
less likely to encounter a contagious disease cue. American Immigration Council (2013), “The Economic Blame
Game: Immigration and Unemployment,” Immigration and
The present research also provides a foundation for fu- Unemployment Fact Sheet. https://www.americanimmigra-
ture research on the influence of contagious cues on social tioncouncil.org/research/economic-blame-game-immigra-
behavior in general. For instance, studies 2, 3, and 5 dem- tion-and-unemployment. Accessed June 2, 2020.
onstrate that people prefer culturally familiar products after Anderson, Adam K., Kalina Christoff, David Panitz, Eve De Rosa,
reading about contagious illness. Might it also be true that and John D. E. Gabrieli (2013), “Neural Correlates of the
consumers show an asymmetric preference toward cultur- Automatic Processing of Threat Facial Signals,” Social
Neuroscience: Key Readings, 23 (13), 185–98.
ally familiar people when making hiring decisions, or show Bates, Douglas, Martin M€achler, Ben Bolker, and Steve Walker
stricter immigration policy preferences when asked about (2015), “Fitting Linear Mixed-Effects Models Using Lme4,”
their political views after encountering a contagious dis- Journal of Statistical Software, 67 (1), 1–48.
ease cue? This is one potential way in which consumer Berry, J. W. and Rudolf Kalin (1995), “Multicultural and Ethnic
prejudice in general, beyond country and origin issues, Attitudes in Canada: An Overview of the 1991 National
Survey,” Canadian Journal of Behavioural Science/Revue
may be influenced or amplified by the presence of conta-
Canadienne Des Sciences du Comportement, 27 (3), 301–20.
gious cues. Research supports this notion. Regions with Bertrand, Philippe (2020), “Les Supermarches Basculent Vers 100
historically higher pathogen prevalence promote societies % de Fruits et Legumes Français,” Les Echos, March 24,
with increased collectivist behaviors such as ethnocentrism https://www.lesechos.fr/industrie-services/conso-distribu-
and conformity in service of reducing the spread of disease tion/les-supermarches-basculent-vers-100-de-fruits-et-
(Fincher et al. 2008). Likewise, during the recent COVID- legumes-francais-1188359. Accessed June 2, 2020.
Blanchard, D. Caroline, Guy Griebel, Roger Pobbe, and Robert J.
19 pandemic, some French grocers switched to selling ex-
Blanchard (2011), “Risk Assessment as an Evolved Threat
clusively French produce (Bertrand 2020). Perhaps cues of Detection and Analysis Process,” Neuroscience &
contagious disease—because of the emotions they elicit Biobehavioral Reviews, 35 (4), 991–8.
beyond the historic presence of disease—might also sys- Bornstein, Robert F., Amy R. Kale, and Karen R. Cornell (1990),
tematically amplify collectivist motives and increase eth- “Boredom as a Limiting Condition on the Mere Exposure
nocentrism and conformity behaviors. Indeed, cues of Effect,” Journal of Personality and Social Psychology, 58
(5), 791–800.
contagious disease might have a systematic influence in Brehm, J. W. (1966), A Theory of Psychological Reactance, New
many domains of social, political, and economic life. York, NY: Academic Press.
Understanding these influences is a fruitful avenue for fu- Bubacz, Kate. (2020), “Take a Look at What People Have Bought
ture research. to Prepare for the Coronavirus Outbreak,” BuzzFeed News,
March 16. https://www.buzzfeednews.com/article/katebu-
bacz/beer-doritos-coronavirus-essentials-groceries. Accessed
DATA COLLECTION INFORMATION June 2, 2020.
Campbell, Courtney and Shayna Murphy. (2020), “These 10
The first author received authorization for access to Household Essentials Are Selling Out—Here’s Where You
Google Flu Trends in May 2016 under the supervision and Can Still Get Them,” USA Today. https://www.usatoday.
permissions of the second author. The first author scraped com/story/tech/reviewedcom/2020/03/30/where-buy-sold-
out-household-essentials-toilet-paper-hand-sanitizer-disin- Hsu, Jason C. (1996), Multiple Comparisons: Theory and

fectant-wipes-and-more/5087376002/. Accessed June 2, Methods, Boca Raton, FL: Chapman & Hall/CRC.
2020. Ivanova, Irina and Thom Craver. (2020), “Closed Due to
Coleman, Nicole Verrochi, Patti Williams, Andrea C. Morales, Coronavirus: List of Activities and State Shutdowns over
and Andrew Edward White (2017), “Attention, Attitudes, COVID-19 Outbreak Concerns,” CBS News, March 18,
and Action: When and Why Incidental Fear Increases https://www.cbsnews.com/news/closed-due-to-coronavirus-
Consumer Choice,” Journal of Consumer Research, 44 (2), list-of-activities-and-state-shutdowns-over-covid-19-out-
283–312. break-concerns/. Accessed June 2, 2020.
Di Muro, Fabrizio, and Theodore J. Noseworthy (2013), “Money Izard, C. E. (1977), Human Emotions, New York, NY: Plenum.
Isn’t Everything, but it Helps if it Doesn’t Look Used: How Jacobs, Andrew, Matt Richtel, and Mike Baker. (2020), “‘At War
the Physical Appearance of Money Influences Spending,” With No Ammo’: Doctors Say Shortage of Protective Gear Is
Journal of Consumer Research, 39 (6), 1330–42. Dire,” New York Times, March 19. https://www.nytimes.
Dutta, Sunasir, and Hayagreeva Rao (2015), “Infectious Diseases, com/2020/03/19/health/coronavirus-masks-shortage.html.

Contamination Rumors and Ethnic Violence: Regimental Accessed June 2, 2020.
Mutinies in the Bengal Native Army in 1857 India,” Kennedy, Ryan, Scott Clifford, Tyler Burleigh, Philip D.
Organizational Behavior and Human Decision Processes, Waggoner, Ryan Jewell, and Nicholas Winter. (2018), “The
129, 36–47. Shape of and Solutions to the MTurk Quality Crisis,”
Ekman, P, R. Levenson, and W. Friesen (1983), “Autonomic Available at SSRN 3272468.
Nervous System Activity Distinguishes among Emotions,” Krusemark, Elizabeth A. and Wen Li (2011), “Do All Threats
Science, 221 (4616), 1208–10. Work the Same Way? Divergent Effects of Fear and Disgust
Ellsworth, Phoebe C. and Craig A. Smith (1988), “From Appraisal on Sensory Perception and Attention,” Journal of
to Emotion: Differences among Unpleasant Feelings,” Neuroscience, 31 (9), 3429–34.
Motivation and Emotion, 12 (3), 271–302. Kushner, Matt G., David S. Riggs, Edna B. Foa, and Suzanne
Faraji-Rad, Ali, Shiri Melumad, and Gita Venkataramani Johar M. Miller (1993), “Perceived Controllability and the
(2017), “Consumer Desire for Control as a Barrier to New Development of Posttraumatic Stress Disorder (PTSD) in
Product Adoption,” Journal of Consumer Psychology, 27 (3), Crime Victims,” Behaviour Research and Therapy, 31 (1),
347–54.
105–10.
Fincher, Corey L., Randy Thornhill, Damian R. Murray, and Mark
Lammers, Joris, Janka I. Stoker, Floor Rink, and Adam D. Galinsky
Schaller (2008), “Pathogen Prevalence Predicts Human
(2016), “To Have Control over or to Be Free from Others? The
Cross-Cultural Variability in Individualism/Collectivism,”
Desire for Power Reflects a Need for Autonomy,” Personality
Proceedings of the Royal Society B: Biological Sciences, 275
and Social Psychology Bulletin, 42 (4), 498–512.
(1640), 1279–85.
Lazarus, Richard S (1991), “Cognition and Motivation in
Galoni, Chelsea and Theodore J. Noseworthy (2015), “Does Dirty
Emotion,” American Psychologist, 46 (4), 352–67.
Money Influence Product Valuations?,” Journal of Consumer
Lee, Spike and Phoebe C. Ellsworth (2013), “Maggots and
Psychology, 25 (2), 304–10.
Gray, Jeffrey A. (1987), The Psychology of Fear and Stress, 5th Morals: Physical Disgust Is to Fear as Moral Disgust is to
ed., New York, NY: CUP Archive. Anger,” in Components of Emotional Meaning: A
Gross, Terry. (2020), “‘All of this Panic Could Have Been Sourcebook, ed. Johnny J. R. Fontaine, Klaus R. Scherer, and
Prevented’: Author Max Brooks On COVID-19,” National Cristina Soriano, Oxford, UK: Oxford University Press,
Public Radio, March 24. https://www.npr.org/2020/03/24/ 271–80.
820601571/all-of-this-panic-could-have-been-prevented-au- Lerner, Jennifer S. and Dacher Keltner (2000), “Beyond Valence:
thor-max-brooks-on-covid-19. Accessed June 2, 2020. Toward a Model of Emotion-Specific Influences on Judgement
Haidt, Jonathan, Clark McCauley, and Paul Rozin (1994), and Choice,” Cognition & Emotion, 14 (4), 473–93.
“Individual Differences in Sensitivity to Disgust: A Scale ——— (2001), “Fear, Anger, and Risk,” Journal of Personality and
Sampling Seven Domains of Disgust Elicitors,” Personality Social Psychology, 81 (1), 146–59.
and Individual Differences, 16 (5), 701–13. Lerner, Jennifer S., Deborah A. Small, and George Loewenstein
Han, Seunghee, Jennifer S. Lerner, and Richard Zeckhauser (2004), “Heart Strings and Purse Strings: Carryover Effects
(2012), “The Disgust-Promotes-Disposal Effect,” Journal of of Emotions and Economic Decisions,” Psychological
Risk and Uncertainty, 44 (2), 101–13. Science, 15 (5), 337–41.
Hancock, Greg. (1997), “Re: Omnibus F Test,” http://www.math. Lerner, Jennifer S. and Larissa Z. Tiedens (2006), “Portrait of the
yorku.ca/Who/Faculty/Monette/Ed-stat/0525.html. Accessed Angry Decision Maker: How Appraisal Tendencies Shape
June 2, 2020. Anger’s Influence on Cognition,” Journal of Behavioral
Hauck, Grace, Lorenzo Reyes, and Jorge L. Ortiz. (2020), “‘Stay Decision Making, 19 (2), 115–37.
Home, Stay Healthy’: These States Have Ordered Residents Lerner, Jennifer S, Roxana M. Gonzalez, Deborah A. Small, and
to Avoid Nonessential Travel amid Coronavirus,” USA Baruch Fischhoff (2003), “Effects of Fear and Anger on
Today, March 29, https://www.usatoday.com/story/news/na- Perceived Risks of Terrorism: A National Field Experiment,”
tion/2020/03/21/coronavirus-lockdown-orders-shelter-place- Psychological Science, 14 (2), 144–50.
stay-home-state-list/2891193001/. Accessed June 2, 2020. Marks, Isaac M. and Randolph M. Nesse (1994), “Fear and
Hayes, Andrew F. (2012), “PROCESS: A Versatile Fitness: An Evolutionary Analysis of Anxiety Disorders,”
Computational Tool for Observed Variable Mediation, Ethology and Sociobiology, 15 (5–6), 247–61.
Moderation, and Conditional Process Modeling [White Marzillier, Sarah L. and Graham C. L. Davey (2005), “Anxiety
Paper],” 1–39. http://claudiaflowers.net/rsch8140/ and Disgust: Evidence for a Unidirectional Relationship,”
Hayesprocess.pdf. Cognition & Emotion, 19 (5), 729–50.
Morales, Andrea C., Eugenia C. Wu, and Gavan J. Fitzsimons Rozin, Paul, Linda Millman, and Carol Nemeroff (1986),
(2012), “How Disgust Enhances the Effectiveness of Fear “Operation of the Laws of Sympathetic Magic in Disgust and
Appeals,” Journal of Marketing Research, 49 (3), 383–93. Other Domains,” Journal of Personality and Social
Morales, Andrea C. and Gavan J. Fitzsimons (2007), “Product Psychology, 50 (4), 703–12.
Contagion: Changing Consumer Evaluations through Smith, Craig A. and Phoebe C. Ellsworth (1985), “Patterns of
Physical Contact with ‘Disgusting’ Products,” Journal of Cognitive Appraisal in Emotion,” Journal of Personality and
Marketing Research, 44 (2), 272–83. Social Psychology, 48 (4), 813–38.
Nemeroff, Carol, and Paul Rozin (1994), “The Contagion Concept So, Jane, Chethana Achar, Da Hee Han, Nidhi Agrawal, Adam
Adult Thinking in the United States: Transmission of Germs Duhachek, and Durairaj Maheswaran (2015), “The
and of Interpersonal Influence,” Ethos, 22 (2), 158–86. Psychology of Appraisal: Specific Emotions and Decision-
Neuberg, Steven L., Douglas T. Kenrick, and Mark Schaller Making,” Journal of Consumer Psychology, 25 (3), 359–71.
(2011), “Human Threat Management Systems: Self- Susskind, Joshua M., Daniel H. Lee, Andree Cusi, Roman Feiman,
Protection and Disease Avoidance,” Neuroscience & Wojtek Grabski, and Adam K. Anderson (2008), “Expressing

Biobehavioral Reviews, 35 (4), 1042–51. Fear Enhances Sensory Acquisition,” Nature Neuroscience,
Okkerse, Liesbet (2008), “How to Measure Labour Market Effects 11 (7), 843–50.
of Immigration: A Review,” Journal of Economic Surveys, Tiedens, Larissa Z. and Susan Linton (2001), “Judgment under
Emotional Certainty and Uncertainty: The Effects of Specific
22 (1), 1–30.
Emotions on Information Processing,” Journal of Personality
Park, C. Whan and V. Parker Lessig (1981), “Familiarity and Its
and Social Psychology, 81 (6), 973–88.
Impact on Consumer Decision Biases and Heuristics,”
Titchener, Edward B. (1910), A Text-Book of Psychology, New
Journal of Consumer Research, 8 (2), 223.
York, NY: Macmillan.
Raghunathan, Rajagopal and Michel Tuan Pham (1999), “All
Togoh, Isabel. (2020), “Here’s How Some of the Countries Worst
Negative Moods Are Not Equal: Motivational Influences of Hit by Coronavirus Are Dealing with Shortages of Protective
Anxiety and Sadness on Decision Making,” Organizational Equipment for Healthcare Workers,” Forbes, March 31.
Behavior and Human Decision Processes, 79 (1), 56–77. https://www.forbes.com/sites/isabeltogoh/2020/03/31/heres-
Rothbaum, Fred, John R. Weisz, and Samuel S. Snyder (1982), how-some-of-the-countries-worst-hit-by-coronavirus-are-
“Changing the World and Changing the Self: A Two-Process dealing-with-shortages-of-protective-equipment-for-health-
Model of Perceived Control,” Journal of Personality and care-workers/#376015e2c13b. Accessed June 2, 2020.
Social Psychology, 42 (1), 5–37. Tsai, Ming-Hong and Maia J. Young (2010), “Anger, Fear, and
Rozin, Paul and Jonathan Haidt (2013), “The Domains of Disgust Escalation of Commitment,” Cognition & Emotion, 24 (6),
and Their Origins: Contrasting Biological and Cultural 962–73.
Evolutionary Accounts,” Trends in Cognitive Sciences, 17 Tybur, Joshua M., Debra Lieberman, Robert Kurzban, and Peter
(8), 367–8. DeScioli (2013), “Disgust: Evolved Function and Structure,”
Rozin, Paul, Jonathan Haidt, and Clark McCauley. (2008), Psychological Review, 120 (1), 65–84.
“Disgust,” in Handbook of Emotions, ed. M. Lewis, J. M. Whitson, Jennifer A. and Adam D. Galinsky (2008), “Lacking
Haviland-Jones, and Lisa Feldman Barrett, New York, NY: Control Increases Illusory Pattern Perception,” Science, 322
Guilford Press, 757–76. (5898), 115–7.

Disgusted and Afraid Consumer Choices Under The Threat of Contagious Disease

Uploaded by

Copyright:

Available Formats

Disgusted and Afraid Consumer Choices Under The Threat of Contagious Disease

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Disgusted and Afraid Consumer Choices Under The Threat of Contagious Disease

Uploaded by

Copyright:

Available Formats

Disgusted and Afraid: Consumer Choices

under the Threat of Contagious Disease

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Consumers regularly encounter cues of contagious disease in daily life—a com-

Keywords: contagion, emotion, disgust, familiarity, fear, disease

O n March 11, 2020, the World Health Organization de-

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

FIGURE 2 outcome uncertainty as study 4, measure participants’ per-

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Results g2 ¼ .000). In contrast, when the symptoms were disgust-

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

out-household-essentials-toilet-paper-hand-sanitizer-disin- Hsu, Jason C. (1996), Multiple Comparisons: Theory and

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

Downloaded from https://academic.oup.com/jcr/article/47/3/373/5840732 by guest on 25 September 2020

You might also like