Академический Документы
Профессиональный Документы
Культура Документы
KIRK KRISTOFFERSON
BRENT MCFERRAN
ANDREA C. MORALES
DARREN W. DAHL
He was bum-rushed by 200 people. They took the doors off They pushed him down and walked all over him. How could
the hinges. He was trampled and killed in front of me. They these people do that?
took me down, too. . . . I didnt know if I was going to live Danielle Damour, sister of victim
through it. I literally had to fight people off my back.
Jimmy Overby, Walmart employee These quotes describe the actions of consumers that
trampled and killed Jdimytai Damour while shopping on
Kirk Kristofferson is Assistant Professor of Marketing at the W.P. Black Friday at a Long Island Walmart location (Gould,
Carey School of Business, Arizona State University, Tempe, AZ 85287 Trapasso, and Schapiro 2008). In addition to causing this
(kirk.kristofferson@asu.edu). Brent McFerran is W.J. VanDusen
death, these same consumers sent three other shoppers,
Associate Professor of Marketing at the Beedie School of Business,
Simon Fraser University, Vancouver, BC, Canada V6C 1W6 (brent. including a 28-year-old pregnant woman, to the hospital
mcferran@sfu.ca). Andrea C. Morales is Lonnie L. Ostrom Chair in with injuries. Each year, the number of violent incidents
Business and Professor of Marketing at the W.P. Carey School of occurring during shopping-crazed holidays, such as Black
Business, Arizona State University, Tempe, AZ 85287 (acmorales@a- Friday (largely in the United States) and Boxing Day (in
su.edu). Darren W. Dahl is BC Innovation Council Professor at the Sauder several countries), increases. In fact, a website called
School of Business, University of British Columbia, Vancouver, BC,
Canada V6T 1Z2 (darren.dahl@sauder.ubc.ca). The authors thank the edi-
Black Friday Death Count keeps track of and details the
tor, associate editor, and reviewers for their insightful and constructive injuries and fatalities that occur directly from Black Friday
comments throughout the review process. Financial support from the promotional sales (http://blackfridaydeathcount.com/).
Social Sciences and the Humanities Research Council of Canada These incidents are not limited only to mob-oriented tram-
(SSHRC) is gratefully acknowledged. pling behavior, but also consist of individuals physically
Mary Frances Luce served as editor and Stacy Wood served as associate and verbally assaulting, robbing, and even shooting fellow
editor for this article. consumers (Morrow 2011). The current research examines
when and why scarcity promotions may lead to such ag-
Advance Access publication September 28, 2016
gressive outcomes.
Published by Oxford University Press on behalf of the Journal of Consumer Research 2016.
This work is written by US Government employees and is in the public domain in the US. Vol. 43 2017
DOI: 10.1093/jcr/ucw056
683
684 JOURNAL OF CONSUMER RESEARCH
Evolutionary psychologists have documented that individ- the underlying process. First, we show that exposure to a
uals will resort to aggression and violence when survival re- scarcity promotion that highlights competition between
sources (e.g., food or water) are in short supply (Brownfield shoppers can lead consumers to perceive others as com-
1986; Cohen and Machalek 1988; Griskevicius et al. 2009). petitive threats to obtaining the desirable good (study 4)
Although not empirically tested in the consumer domain, the and physiologically prepares the consumer to aggress by
multitude of violent incidents reported during marketer- increasing testosterone levels (study 3). To our knowledge,
induced scarcity promotions suggests such behavior can also this is the first consumer behavior work to empirically
occur in resource-rich environments, or environments in demonstrate this type of physiological reaction to a promo-
which consumers survival is not threatened. But what factors tional ad. We provide process evidence for the role of per-
might activate such aggression? Although mob aggression to- ceived competitive threat through direct measurement and
ward other consumers during deep-discounted retail sales is mediation (study 4), as well as through direct manipulation
likely multiply determined, very little is known about the by increasing social affiliation (study 5), reducing aggres-
drivers of such acts. The present research investigates one sive brand image associations (study 6), or changing the
such driver and focuses on the impact that scarcity promo- type of scarcity promotion (quantity vs. time, study 7). We
tions may have on this aggressive outcome. Here, we test the next turn to a review of the literature and outline our con-
idea that simply encountering a scarcity promotion, such as a ceptual framework.
newspaper or television advertisement or online pop-up ad,
may cultivate seeds of aggressive behavior in consumers and
predispose them to act in a violent manner. CONCEPTUAL FRAMEWORK
Strikingly, we find that merely exposing consumers to a At a basic level, scarcity originates from an imbalance
scarcity (vs. control) promotional ad can lead to increased between demand and supply, leading to shortages and com-
aggressive behavior. We find that this outcome results after petition for resources. Foundational scarcity research has
exposure to scarcity promotions that limit product quantity focused primarily on events such as economic recessions
because consumers perceive a potential competitive threat (Griskevicius et al. 2012) or periods of famine or drought
of other people trying to obtain the desired product and ex- (Chakravarthy and Booth 2004). Scarcity as a phenom-
perience a physiological change that prepares the body to
enon, however, is also found in resource-rich environ-
aggress. In service of eliminating this perceived threat,
ments, such as (nonessential) consumer good shortages
consumers respond with aggression.
(Lynn 1992, 1993). These shortages can be the result of
This research makes several important contributions to
many factors, including demand shocks, production delays,
the literature. From a theoretical perspective, we add to the
capacity constraints, and limited production runs
scarcity literature by showing that aggressive reactions to
(Verhallen and Robben 1994). Firms may even attempt to
scarcity can occur not only for survival resources such as
food and water, but also for luxury goods in resource-rich create scarcity by intentionally holding supplies artificially
consumer environmentsa proposition previously dis- low, or creating the perception of scarcity by means of pro-
cussed (Cialdini 2009) but never empirically tested. Most motions or sales (Cialdini 2009; Gitlin 2007). A scarcity
importantly, however, this research is the first to show that promotion is defined as a marketing tactic that empha-
exposure to scarcity promotionsa common marketing sizes limited availability (either in quantity or time) of a
tactic used by firmscan lead to increased aggression specific product or event (Ku, Kuo, and Kuo 2012). Firms
among consumers (studies 1 and 2, 47). We show that utilize scarcity promotional tactics throughout the year, but
marketplace aggression is not merely the outcome of their most salient usage is for high-profile shopping-ori-
crowds during shopping holidays, but can actually be acti- ented events (e.g., Black Friday, Boxing Day) in which
vated beforehand, at ad exposure. Further, we show that large discounts are offered on highly desirable items, but
scarcity promotion exposure increases the human bodys available quantity is often limited, as is the time to access
physiological responses associated with aggression (study the promotion (only that day or week).
3), and facilitates aggression when an opportunity is avail- Prior research has consistently shown that product scar-
able. However, although aggression and competition are city may influence perceptions of value. Specifically, prod-
related constructs, we find that while competition triggered ucts and services seem more valuable when they are in
by limited-quantity scarcity promotions heightens the like- short supply (Brock 1968; Cialdini 1993; Sharma and Alter
lihood that consumers will engage in aggressive competi- 2012). More recent work has examined the degree to which
tive actions like shooting, hitting, and kicking, it does not product familiarity may influence the relationship between
increase nonaggressive competitive actions like working/ scarcity and purchase intentions, demonstrating that con-
thinking harder (study 7), highlighting the specific associ- sumers who are uninformed or unfamiliar with the avail-
ation between scarcity and aggression in particular. able options may rely more on the behavior of others when
In addition to demonstrating the negative, nonnormative making choices, as compared to consumers who already
behavioral outcome of aggression, we provide evidence for possess information about the available options (Castro,
KRISTOFFERSON ET AL. 685
Morales, and Nowlis 2013). In addition, this work shows of the moment. Instead, we show evidence for another po-
that scarce products are evaluated more positively when tential driver: consumers may in fact come to the stores
the scarcity is due to market circumstances (i.e., a product predisposed to aggress as a result of marketing actions, and
is in high demand due to popularity), as compared to when this effect may occur anytime the quantities of desired
it is due to accidental or nonmarket circumstances (i.e., a goods are scarce. Put another way, we propose that a scar-
product has inadvertently not yet been restocked). As a re- city cue outside of the immediate consumption context that
sult, the preference for scarce products is due at least in highlights potential competition between consumers may
part to what the consumer believes is the underlying cause elicit physiological aggressive impulses, which may subse-
of the shortage (Lynn 1992; Verhallen 1982; Verhallen and quently release consumers to act on these impulses when
Robben 1994). given the opportunity to aggress. Further, we propose that
While items that are scarce are frequently perceived as these tendencies can result in more generalized aggressive
valuable, the converse is also true: entities that are valuable actions, meaning that the target of the aggression need not
are often scarce (King, Hicks, and Abdelkhalik 2009). be a person who is actually competing for the scarce items,
When necessary resources, such as food, water, or shelter, nor does the target need be a person at all. We show that as
are in short supply, competition for survival increases and long as there is an opportunity to aggress, consumers for
individuals can resort to aggression. We define aggres- whom aggressive impulses have already been activated are
sion broadly as behaviors intended to harm or injure an- more likely to behave accordingly.
other person or object (Lorber 2004). Previous research But why would consumers resort to aggression versus
supports the link between extreme resource scarcity, com- other, less dangerous competitive behaviors (e.g., product
petition, and aggression. In some cases, it has been shown hoarding; Byun and Sternquist 2011) upon exposure to a
that fatal violence has been used to obtain such scarce re- limited-quantity scarcity promotion? This deleterious be-
sources (Hagmann and Mulugeta 2008; Harvey 2009; havior may be less surprising given psychological research
Reuveny 2007). However, this research is limited to nonex- on aggression. We adopt the perspective of Berkowitz and
perimental contexts (e.g., ethnographies, case studies), LePage (1990), who argues that aggression is multiply
very impoverished environments, and/or nonbehavioral ag-
determined, and that there exist several factors that can
gression measures (e.g., aggression scales, intentions to
make anger and aggression more or less likely to occur.
harm others).
Specifically, he suggests that aggressive behavior can re-
Although aggressing toward fellow consumers within a
sult not only from negative affect (e.g., anger) or perceived
retail context to obtain one of the few 72-inch LED televi-
safety threat (Stein and Levine 1989), but also from situ-
sions or Kleinfeld wedding gowns on sale at a bargain
ational cues that highlight linkages associated with violent
price can hardly be explained by the need to obtain life-
essential resources, aggression in such contexts is perhaps behavior. A review of more recent work suggests that
less surprising given the many documented instances of scarce environmentsspecifically, we would argue, ones
mob-oriented behavior (e.g., Bandura 1977; Bandura et al. with cues of competitive threat within themmay in fact
1996; Zimbardo 1969), even in consumption contexts foster an aggressive association and facilitate violent
(Roberts and Benjamin 2000; Simpson et al. 2011). action.
However, aggressive behavior in a consumption context, In particular, findings suggest that scarce environments
such as a retail store, may not only be the outcome of the can impair consumer cognitive functioning and lead to
scarcity and value of the good itself, but could also be a re- poorer decision making (Mani et al. 2013). Shah,
action to social norms for what is appropriate behavior. Mullainathan, and Shafir (2012) found that a scarcity
One might argue that many of the consumers shopping on mind-set induces an intense present focus and a willingness
Black Friday, for example, would likely not aggress if to sacrifice ones future well-being in order to meet present
other people who were already pushing and shoving did goals. This present focus, in turn, has been shown to cause
not surround them, indicating by their behavior that aggres- consumers to adopt a more agentic and competitive mind-
sion in this context is acceptable (Asch 1954; Sherif 1936). set (Roux, Goldsmith, and Bonezzi 2015), neglect other
It is also possible that consumers view certain sale days as situational dimensions (Zwane 2012), and exhibit lower
sacred (or mythical; Tumbat and Belk 2011) events that self-control (Laran 2010). Taken together, under certain
warrant the suppression of appropriate behavior and license conditions that highlight competition between consumers,
consumers to behave antisocially. we propose that scarcity cues, such as limited-quantity pro-
In the current work, we not only extend the scarcity lit- motions for nonnecessity or luxury items, may foster a spe-
erature to show that consumers can aggress in resource- cific association with aggression that leads consumers to
rich environments, but we also challenge the assumption ignore the significant costs associated with violence, and
that marketplace violence is necessarily the result of a mob makes them more likely to engage in competitive, aggres-
mentality or idiosyncratic to certain days, which push sive actions (Campbell 1999; Taylor et al. 2000; Wilson
otherwise nonviolent individuals over the edge in the heat and Daly 1985).
686 JOURNAL OF CONSUMER RESEARCH
Scarcity Promotions and Aggressive Behavior: associated with aggression, and lead to aggressive behavior
The Role of Perceived Competitive Threat even outside of the competitive context. Formally,
But what underlying psychological mechanism could H1: Consumers exposed to a scarcity promotion high-
lead consumers to act aggressively in response to scarcity lighting limited quantity (vs. a promotion that
promotions? We propose that when quantities are limited, does not highlight limited quantity) will behave
scarcity promotions prompt consumers to perceive other more aggressively.
shoppers as competitive threats to obtaining the highly at- H2: Exposure to a scarcity promotion highlighting
tractive products. In other words, a perceived competitive limited quantity (vs. a promotion that does not
threat mediates the effect of limited-quantity promotions highlight limited quantity) leads consumers to
on aggression. By advertising the limited quantity of the perceive other consumers as potential competi-
featured products, the scarcity promotions necessarily tive threats to obtaining the focal product.
highlight that few people will actually be able to purchase Perceived threat will mediate the relationship be-
the goods, thereby fostering competition between shoppers. tween limited-quantity scarcity promotion ex-
Furthermore, we contend that perceiving other consumers posure and increased aggression.
as competitive threats following exposure to scarcity pro-
H3: Consumers exposed to a scarcity promotion high-
motions elicits a physiological response that prepares con- lighting limited quantity (vs. a promotion that
sumers to aggress. Specifically, we argue that exposure to does not highlight limited quantity) will exhibit
limited-quantity scarcity promotions causes an increase in higher testosterone levels.
physiological hormone levels associated with aggression
in particular, testosterone. Scarcity Promotion Type and Subsequent Violent
Previous evolutionary research has established a positive
relationship between testosterone and aggressive behavior
Action
in both humans (Book, Starzyk, and Quinsey 2001) and But when might scarcity promotions not lead to such a
nonhuman animals (Rines and vom Saal 1984), and destructive outcome? We propose that factors that reduce
increased testosterone levels have been shown to prepare the perceived competitive threat other consumers pose to
the body to act in an aggressive manner (Terberg and van obtaining the desired product will interrupt the activation
Honk 2013). In the consumer domain, Wood, McInnes, of aggression-related responses to scarcity and not result in
and Norton (2011) speculated that sports games with a such behavior. In other words, when cues are present that
close final score (where competition is fiercest) may lead directly minimize competition with other consumers, ag-
to increased testosterone levels in spectators, subsequently gressive tendencies in response to scarcity promotions
resulting in more aggressive, violent driving behavior. should be muted.
Relatedly, neuroendocrine research has demonstrated that One critical factor of both theoretical and practical sig-
in highly competitive situations, testosterone levels can in- nificance that directly affects perceived competitive threat
crease even when the competitive situation is merely is the type of scarcity promotion employed. Marketers use
anticipated (Gonzalez-Bono et al. 1999; Mazur, Booth, and scarcity promotions in two primary ways: either by limit-
Dabbs 1992; Mazur and Booth 1998). Prior work has also ing the number of products available (quantity: Only 5
shown that changes in hormone levels can occur in re- Available) or the time period for which the sale lasts
sponse to anticipated or imagined interactions not of a (time: One Day Only, Sale Lasts Until Noon).
competitive nature (Cesario, Plaks, and Higgins 2006; According to our conceptualization, if perceived competi-
Goldey and van Anders 2011); however, such work has ex- tive threat of other consumers is the process through which
plicitly asked participants to imagine an interpersonal exposure to scarcity promotions drives aggression, then
interaction (we do not). Further, to our knowledge, no re- this deleterious consequence should result only when the
search to date has demonstrated any hormonal change in scarcity promotion limits promotional quantity (vs. time).
response to a marketing appeal. This is because promotions that limit product quantity in-
Indeed, in the current research we propose that exposure herently pit consumers against each other and heighten the
to limited-quantity scarcity promotions can lead to an ob- competitive threat others pose to securing the desired
servable behavioral change (increased aggression) that is good. Put another way, when product quantity is limited,
the result of both a psychological (perceived competitive consumers will miss out if they do not get to the product
threat), as well as an internal, physiological change before other consumers. Previous research supports this
(increased testosterone). More specifically, we propose claim, showing that limited-quantity promotions increase
that a scarcity cue that highlights other people as competi- both uncertainty about successfully obtaining the scarce
tion, such as a limited-quantity promotional ad, can product and the locus of causality in acquisition
heighten the perceived threat other consumers play in ob- (Aggarwal, Jun, and Huh 2011; Inman, Peter, and
taining the target good, activate a physiological response Raghubir 1997; Lynn 1993; Meyer 1980). Conversely, in
KRISTOFFERSON ET AL. 687
promotions that limit time, all consumers who want to se- and brand image association (study 6). Finally, study 7
cure the promotional product will do so as long as they ar- examines type of scarcity promotion (quantity vs. time, hy-
rive within the allotted time, making the perceived pothesis 4), again supporting the proposed mechanism, and
competitive threat other consumers pose in inhibiting prod- further highlights when aggressive competitive responses
uct acquisition minimal (i.e., consumers are competing can manifest compared to nonaggressive competitive
only against the clock, not each other). Thus, we predict behaviors.
that the type of scarcity promotion will moderate the rela-
tionship between scarcity promotions and subsequent ag- STUDY 1
gressive behavior by impacting the perceived competitive
threat of other consumers, such that aggression will result Study 1 provided preliminary support for the hypothesis
only when the promotion is limited quantity but not limited that limited-quantity scarcity promotions for a desirable
time. Formally, product can lead to increased aggression (hypothesis 1). In
this study, we contrasted a scarcity promotion ad with a
H4: The type of scarcity promotion will moderate the control promotion ad for the same product in which no re-
relationship between scarcity promotions and ag- strictions on product quantity availability are made.
gression, such that exposure to scarcity promo-
Because of the difficulty (and ethical implications) of
tions will lead to increased aggressive behavior
studying interpersonal violence in a controlled lab setting,
when the promotion limits available product
quantity, but not when the promotion limits avail- we examined aggressive behavior in this study using vio-
able time to obtain the product. lent video games (Anderson and Bushman 2002;
Englehardt et al. 2015).
Aggressive versus Nonaggressive Competitive
Reactions to Scarcity Promotions Procedure
One hundred forty marketing undergraduates from the
Finally, our framework predicts that exposure to limited-
University of British Columbia (ages 1856, Mage 22.7,
quantity scarcity promotions will lead specifically to ag-
56.3% female) participated in this study in exchange for
gressive (vs. nonaggressive) competitive behaviors. This is
course credit, and were randomly assigned to one of two
because scarcity is specifically associated with aggression.
conditions (promotional ad: scarcity, control), manipulated
Thus, we contend that while competition triggered by scar-
between-participants.
city promotions will increase the likelihood of consumers
Participants arrived at the lab in groups of four and com-
engaging in aggressive competitive behaviors like shoot-
pleted the study at individual computer terminals. They
ing, hitting, and kicking, it will not have the same effect on
were told that they would be completing a number of unre-
nonaggressive competitive actions like working or thinking
lated studies, the first of which would be to provide feed-
harder. That is, because of the perceived competitive threat
back to the universitys bookstore regarding a promotion it
and physiological responses that are activated upon expos-
planned to run in the near future. Participants were then
ure to limited-quantity scarcity promotions, aggressive
directed to an online survey, given a sealed envelope that
competitive actions become more likely as consumers are
contained the promotional ad stimuli, and directed to open
predisposed to aggress, but nonaggressive behaviors should
the envelope when prompted by the survey.
remain relatively unaffected.
Scarcity Manipulation. Participants reviewed one of
Overview of Studies two promotional ads from the UBC bookstore that served
as our scarcity manipulation. The ads described a promo-
We test our proposed framework in seven studies using tional sale featuring the (then new) 64 GB Apple iPhone 5
multiple behavioral measures of aggression. Study 1 dem- for only $50 (retail value of $650 crossed out; see appendix
onstrates that exposure to limited-quantity scarcity promo- A for stimuli and web appendixes A and B for product and
tion advertising leads to increased aggressive behavior stimuli pre-tests). The heading Promotional (vs. Black
(hypothesis 1). Study 2 replicates this effect in a different Friday) Sale was selected to ensure reactions were not due
consumption context and begins to disentangle the aggres- to the perception that Black Friday is a sacred or mythical
sive response from more general competitive (but nonag- event. It also served to enhance the generalizability of the
gressive) behavior. Study 3 shows that exposure to a potential effect. To manipulate scarcity, we varied the
scarcity promotion leads to physiological increases in tes- quantity of iPhones available to consumers via the promo-
tosterone (hypothesis 3). Study 4 provides support for the tion. In the scarcity (control) promotion, only 3 (3,000)
underlying process of perceived competitive threat driving iPhones were available to consumers.1 This information
aggression (hypothesis 2) through measured mediation.
Studies 5 and 6 also document the process by manipulating 1
Retailers in the marketplace do advertise both large and small quan-
perceived threat directly through social affiliation (study 5) tities available during scarcity promotions.
688 JOURNAL OF CONSUMER RESEARCH
was presented in the main text of the ad, as well as at the which served as our dependent measure of aggressive be-
bottom. All other information was identical between the havior. Two coders, blind to the hypothesis, manually
two promotions. viewed each participants game play video file and re-
After reviewing the bookstores promotional ad, partici- corded the dependent measure using the Morae software.
pants answered cover story questions regarding anticipated
demand for the promotion and expected word of mouth
among the university population. Finally, participants com- Results
pleted the following manipulation check item on a scale Manipulation Check. The manipulation check revealed
from 1 (very scarce) to 7 (very abundant): How would that the scarcity manipulation was successful. Participants
you describe the University Book Stores iPhone 5 promo- perceived the promotional package quantity to be more
tional package quantity? scarce (lower values indicate more scarce) in the scarcity
Eliciting Aggressive Behavior. Immediately after re- condition than the control condition (MScarcity 1.72 vs.
viewing the scarcity ad, participants moved on to our MControl 3.97; F(1, 138) 68.5, p < .001).
measure of aggression, ostensibly described as a separate
study on classic video games: firearm shooting behavior. Dependent Variable. The number of shots fired was se-
Specifically, participants were told that marketing re- verely nonnormal (!Shots Fired 2.48, x(140) .72, p <
searchers were investigating the recent trend among video .001); thus, we log-transformed this variable to test our hy-
game players of downloading, purchasing, and playing pothesis. We note, however, that results are consistent if
classic video game systems (e.g., Super Nintendo, Atari), the analysis is performed using raw values. We report raw
and that they would be randomly assigned to play a classic score means and standard deviations for ease of under-
game and report on their experience. In actuality, all par- standing. We examined the shooting behavior of partici-
ticipants were directed to play an online first-person shoot- pants using a one-way ANOVA with shots fired as the
ing game called Deadeye. In this game, participants use the dependent variable. As predicted, participants exposed to
mouse to aim a gun and shoot at moving targets. The game the scarcity promotion fired significantly more bullets than
is referred to as first-person because the game play is de-
participants exposed to the control promotion (MScarcity
signed such that players aim the gun as if they are holding
42.3, SD 19.4 vs. MControl 37.0, SD 11.7; F(1, 138)
it in their hands and aiming with their own eyes.
4.02, p .047).2
One feature of the game is that the dominant strategy is
not to fire bullets recklessly. Upon starting the game, par-
ticipants were given basic on-screen instructions about Discussion
how to shoot and informed that game scoring is dependent
upon both the number of targets hit and the accuracy of Study 1 provided preliminary support for our hypothesis
shooting behavior. We chose to operationalize aggression that exposure to limited-quantity scarcity promotions can
as the number of bullets participants fired during game lead to increased aggressive behavior among consumers
play. Given that experience with first-person shooting (hypothesis 1). We found that consumers behave more ag-
games could impact measures such as accuracy and overall gressively after viewing a scarcity (vs. control) promo-
score, we felt the quantity of bullets fired was the cleanest tional ad for a desirable product. To our knowledge, this is
measure of aggressive behavior in this experimental con- the first study to empirically demonstrate that exposure to
text, although we also report accuracy results below. a marketing promotion can induce aggressive behavior.
Shooting behavior was unobtrusively recorded using a pro- Importantly, this increased aggression resulted from a scar-
gram called Morae Recorder (see web appendix C for a de- city promotion that featured a nonnecessity product
tailed description of the software). (iPhone) among participants operating in a resource-rich
As the overall difficulty and complexity of the game is environment (a university in a large North American city).
low, and in order to increase the number and speed of the This suggests that individuals may resort to aggression not
moving targets, participants were instructed to play the only to ensure their very survival, but also in other con-
game on the most difficult setting. Participants who played sumption contexts.
the game on an incorrect setting were removed prior to
analyses to ensure consistency. In this and all subsequent 2
Subsequent investigation revealed that scarcity participants were
studies exclusions did not vary systematically by condition. significantly less accurate in their shooting behavior than control par-
The game took approximately 1 minute to complete. Upon ticipants (p .052), although their overall scores did not differ (p >
.90). Analysis shows that participants were hitting the same number of
completion of the game, participants answered cover story targets in each condition, but were using more bullets to do so under
questions, and then proceeded to a separate study. scarcity (acting more aggressively). Given participant knowledge that
overall score was based in part on higher accuracy, this suggests that
Dependent Variable. Participant mouse clicks were re- scarcity was leading them to be more aggressive, even when they were
corded to measure the number of bullets participants fired, instructed that it was not beneficial to do so.
KRISTOFFERSON ET AL. 689
person at the vending machine (11), participant interacted limited-quantity scarcity promotions elicits physiological
with previous study participant while waiting to begin the responses shown previously in the literature to predict ag-
study (4), machine error due to foreign (US) currency gressive behavior.
usage (6). The pattern of results is consistent if all partici-
pants are included. STUDY 3
Manipulation Check. The scarcity manipulation
worked as intended between the three conditions. The goal of study 3 was to extend the behavioral find-
Participants exposed to the scarcity ad perceived the ings of studies 1 and 2 by examining whether exposure to a
iPhone promotional quantity to be more scarce than partici- limited-quantity scarcity promotion can elicit automatic
pants exposed to both controlquantity omitted (MScarcity physiological responses associated with aggressive behav-
2.00, SD 1.50 vs. MControlQuantity Omitted 3.33, ior (hypothesis 3). If consumers exhibited increased testos-
SD 1.48, F(1, 198) 23.2, p < .001) and control terone levels after exposure to a scarcity (vs. control)
(MControl 4.45, SD 1.76, F(1, 198) 80.2, p < .001) promotion, this would lend support to our claim that
ads. Moreover, participants in the controlquantity omitted limited-quantity scarcity promotions may lead to increased
condition perceived the iPhone promotional quantity to be likelihood of aggressive behavior.
more scarce than participants in the control condition (F(1, To test this proposition, we partnered with a leading sal-
198) 16.6, p < .001). ivary bioscience research institute, Arizona State
Universitys Institute for Interdisciplinary Salivary
Mood. No differences in participant mood emerged be- Bioscience Research, to conduct an experiment to measure
tween the three conditions. Moreover, mood did not predict testosterone levels after exposure to a scarcity (vs. control)
the dependent measure and will not be discussed further. promotion.
Dependent Variables. We standardized the three de-
pendent measures and averaged them to form a single ag- Procedure
gressive behavior score. We note that each of the three
One hundred fifty marketing undergraduates (ages 18
dependent measures is independently significant and
48, Mage 22.3, 48.1% female) from Arizona State
matches the results of the combined measure (we provide
University participated in this study in exchange for course
full details and statistics in web appendix D). To test
whether participants exposed to the scarcity promotion credit and were randomly assigned to one of two condi-
behaved more aggressively than participants exposed to tions (scarcity, control) in a between-participants design.
the control or controlquantity omitted promotions, we cre- Saliva testing was selected as the testosterone measure-
ated two dummy variables and entered them in a linear re- ment procedure to maximize measurement accuracy and to
gression to predict aggressive behavior. Supporting our minimize invasiveness to participants. The study lasted ap-
predictions, participants exposed to the scarcity promotion proximately 30 minutes.
behaved significantly more aggressively toward the ma- Our procedure and analysis were guided by the expertise
chine than participants exposed to both controlquantity of the bioscience institute and recent bioscience salivary
omitted (MScarcity .29, SD 1.11 vs. MControlQuantity research design findings (Bosch et al. 1996; Granger et al.
Omitted .09, SD .79, b .49, t(198) 3.14, p .002)
1999; Granger et al. 2004; Out et al. 2013). Participants
and control promotion conditions (vs. MControl .20, SD were instructed not to eat, drink, smoke, brush their teeth,
.78, b 39, t(198) 2.44, p .016), both of which did or use mouthwash for at least one hour prior to the session
not differ from each other (p > .50). (Dabbs 1991). Participants arrived at the lab in groups of
six to 10 and were seated at individual workstations. Upon
Discussion arrival, participants were given a cover story explaining
that researchers were investigating how different oral hy-
The results of study 2 provide further support for our giene products affect user saliva, and that three saliva sam-
claim that exposure to a limited-quantity scarcity promo- ples would be taken during the course of the session.
tion can lead to aggressive behavior (hypothesis 1). Following previous research involving salivary measure-
Specifically, in a real consumption setting, the results of ment, three saliva samples were taken to obtain reliable es-
study 2 showed that exposure to a limited-quantity scarcity timates of individual baseline hormone levels (Out et al.
promotional ad led to greater physical aggression. We also
2013). Thus, baseline testosterone levels were measured
found no differences in aggressive behavior between the
both at the beginning and end of the experimental session
control and controlquantity omitted conditions in this
(separated from target stimuli exposure; Bosch et al. 1996).
study, thus ruling out the alternative explanation that our
effects are due to a reduction in aggression when a large Baseline Measurement 1. The first of the two baseline
quantity is presented. In study 3, we begin to test our pro- measures was taken at the beginning of the session.
posed process claim by examining whether exposure to Participants were given an oral swab and instructed to
KRISTOFFERSON ET AL. 691
place it on the tongue for 2 minutes until saturation. Upon the two measurements was provided to us for data analysis.
saturation, the experimenter brought a glass test tube No other hormones besides testosterone were assayed.
(labeled with a bar code to track participant and measure-
Participants. Four participants experienced problems
ment timing) and the participant placed the swab inside.
during the session and were not able to complete the
The experimenter then stored the samples in the freezer at
measurements as directed (e.g., swab-induced gagging,
20 degrees Celsius.
target measurement was taken prior to ad exposure).
Scarcity Manipulation. After completing the first base- Further, six participants exhibited testosterone levels that
line testosterone measure, the experimenter informed par- exceeded 3 standard deviations from the mean, suggest-
ticipants that they would move on to an unrelated study ing measurement or procedural error, as these outliers far
before continuing with the oral hygiene products study. exceeded documented levels found in humans (Goldey
Specifically, participants were told that they would provide and van Anders 2011; Granger et al. 2013). Therefore,
feedback on a marketing promotion the university the analysis was conducted with 140 valid participant
bookstore was planning to run. The instructions and stimuli samples.
were identical to those used in study 1. Participants were
Manipulation Check. The scarcity manipulation check
exposed to either a scarcity or control ad for the
was successful. Participants perceived the iPhone promo-
bookstores iPhone promotion and answered the same
tional package quantity to be more scarce in the scarcity
cover story questions and manipulation checks.
versus the control ad (1 very scarce to 7 very abun-
Dependent Variable. After participants exposure to dant; MScarcity 2.19 vs. MControl 4.14; F(1, 138)
the scarcity manipulation, the target testosterone measure- 41.9, p < .001).
ment was taken, with participants receiving a second oral
Covariates. The following factors were included as
swab. Identical measurement and storage procedures were
covariates following previous salivary testing research:
followed.
baseline levels, gender, time of day, age, ethnicity, smok-
Demographic Covariates. After the target testosterone ing history, and dental habits (Shirtcliff et al. 2002). Only
measurement was taken, participants completed an oral hy- the two baseline testosterone levels emerged as significant.
giene health and demographic questionnaire for approxi- For model parsimony, we included only the significant
mately 10 minutes. In addition to supporting the cover covariates in our analysis.
story, this questionnaire collected demographic measures
that bioscience research has shown to affect testosterone Dependent Variable. Our target dependent measure
levels: time of day, gender, age, ethnicity, dental habits, was participant testosterone levels after exposure to the
and smoking history (Shirtcliff, Granger, and Likos 2002). scarcity stimulus. A one-way ANCOVA supported our
Upon completion of the questionnaire, participants were proposition. Participants exposed to the scarcity promotion
directed to watch a neutral video about tourism in a exhibited significantly higher testosterone levels than par-
European city and answered cover story questions for the ticipants exposed to the control promotion (MScarcity
remainder of the study (approximately 10 minutes). This 121.5, SD 66.9 vs. MControl 120.8, SD 52.9; F(1,
neutral task was performed to allow testosterone levels to 136) 3.82, p .053).3 This result suggests that scarcity
return to a baseline state following stimuli presentation promotions elicit physiological changes in hormones
(Bosch et al. 1996). shown to be associated with aggressive behavior.
the first research to show that exposure to a marketing pro- given on how participants should behave during the game.
motion can elicit a physiological change in consumers. The These basic instructions were provided to minimize any
results of study 3 also supported our process claim regard- uncertainty regarding how to punch using a Wii controller,
ing why limited-quantity scarcity promotions can lead to as our main dependent variable was number of punches
generalized aggressive behavior. Our framework proposes thrown.
that exposure to limited-quantity scarcity (vs. control) pro- Immediately after reviewing the scarcity manipulation
motions leads consumers to perceive others as competitive and instructions, the research assistant instructed the par-
threats to obtaining the target good, and physiologically ticipant to stand on a marked spot and to play a 1-minute
prepares the body to potentially aggress when given the op- round of the game. Each gaming session was recorded with
portunity to do so. We provide more process support in participant consent; no participants declined. Importantly,
study 4 by demonstrating the psychological mechanism to control for differences in gaming experiencing or prior
driving this behavior, while also examining another type of video game knowledge, we selected a training setting in
physical violence: punching. which the opponent did not fight back. This decision
allowed the participant to blatantly aggress against a de-
STUDY 4 fenseless human opponent, and ruled out provocation as
a driver of participant aggression. In addition, this study
Study 4 had two goals. First, we sought direct evidence addresses a minor limitation of study 1, where participants
for the psychological process underlying increased aggres- were given a score. While in that study, participants were
sive behavior in response to scarcity promotions. Our told that firing recklessly would not increase their score,
framework predicts that limited-quantity scarcity promo- seeing a score is still feedback, in that it is a signal about
tions can prompt individuals to perceive other consumers the normative value of aggression, and a cue of a competi-
as potential competitive threats to obtaining the desired tive setting. In this study, no score or performance feed-
product (hypothesis 2), and this increased threat mediates back was given in any way (it was merely a training
the relationship between the scarcity promotion and ag- session). Thus, throwing more punches yielded no strategic
gressive action. Second, in study 4 we sought to examine benefits, nor could any validation or disapproval (from the
yet another type of physical aggression: punching. As men- game) be inferred. Participants attacked a person that was
tioned in study 1, given the low likelihood of observing visually identical to the participants assigned character
not to mention the ethical impossibility of encouraging (i.e., white male with an average build).
punching behavior among lab participants, we return to our Dependent Variable. The number of punches thrown
violent video game paradigm to assess participants attack- served as our dependent measure of physical aggression.
ing of a target with human likeness. Two coders, blind to the hypotheses, watched each partici-
pants video and manually counted the number of punches
Procedure thrown during the 1-minute session. If disagreements arose,
One hundred seven marketing undergraduates (ages 18 both coders rewatched the participant video together until
41, Mage 21.9, 55.1% female) from Arizona State agreement on the number of punches thrown was reached.
University participated in this study in exchange for course Process Measure: Perceived Competitive Threat. After
credit, and were randomly assigned to one of two between- completing the gaming session, participants returned to
participants conditions (promotional ad: scarcity vs. their computer workstation and completed the following
control). Upon arrival to the lab, participants were seated question to measure perceived competitive threat on a scale
at the computer station and told that they would be com- from 1 (not at all threatening) to 7 (very threatening):
pleting a number of unrelated studies, the first of which How much do you perceive other people as a threat to you
was to provide feedback to the university bookstore on an obtaining the product in the promotion? After completing
upcoming promotion it planned to run. The instructions this item, participants moved on to a separate study.
were identical to those of previous studies, and the
university bookstore promotional ad served as our scarcity Results
promotion manipulation. Participants reviewed either the
scarcity or control ad from study 1, and answered the same Manipulation Check. The manipulation check revealed
cover story and manipulation check questions. Upon com- that the scarcity manipulation was successful. Participants
pletion of the scarcity ad manipulation, participants were perceived the iPhone promotional package quantity to be
given the same classic video game cover story as study 1, more scarce in the scarcity versus the control ad (MScarcity
1.95 vs. MControl 3.47; F(1, 105) 22.8, p < .001).
but were instead assigned to play the Wii boxing game
Ready 2 Rumble Revolution. Participants were provided Dependent Variable. Supporting hypothesis 1, partici-
with an information sheet that outlined how to punch using pants threw marginally more punches at the defenseless
the Wii controller, but no additional instructions were opponent after exposure to the scarcity promotion than
KRISTOFFERSON ET AL. 693
after exposure to the control promotion (MScarcity 68.2, affiliation. Research has established that similarity and af-
SD 33.4 vs. MControl 57.3, SD 26.3; F(1, 105) filiation with others can reduce a perceived threat to the
3.48, p .065). self by reducing uncertainty and informing consumers how
they should feel toward, view, and treat others (Park and
Mediation Test. To test the indirect effect, we followed
Maner 2009). Specifically, feelings of affiliation with
Preacher, Rucker, and Hayes (2007)s bootstrapping pro-
others are related to positive affect and subsequent behav-
cedure of 10,000 resamples with replacement. Participants
ior (Hogg et al. 2007). Thus, we manipulated perceived
exposed to the scarcity promotion perceived other con-
threat by reinforcing a social affiliation between the self
sumers as higher potential threats to obtaining the target
and fellow consumers using a manipulation from previous
product than participants exposed to the control promotion
research (Wellen, Hogg, and Terry 1998). Participants
(MScarcity 5.88, SD 1.50 vs. MControl 4.78,
wrote down two ways in which they were similar to (low
SD 1.84; F(1, 105) 11.4, p .001). Next, controlling
threat) or different from (high threat) other consumers that
for scarcity condition, perceived threat significantly pre- lived in their city. A separate pre-test (n 60) confirmed
dicted the number of punches thrown (b 3.86, t(104) that writing about being similar to (vs. different from) peo-
2.23, p .028). Supporting hypothesis 2 and our overall ple in their city led participants to view fellow consumers
framework, the indirect effect of scarcity promotion on ag- as less threatening (see web appendix E for details).
gressive behavior through perceived threat was significant Second, we wished to extend the generalizability of our
(b 2.11, SE 3.02, CI95: 4.85, 0.55). effects by using a more diverse sample and an additional
measure of generalized aggression: preference for violent
Discussion experiences. In this study, participants were presented with
The results of study 4 support our framework and repli- seven pairs of classic video games (one violent and one
cate previous studies, again showing that exposure to a nonviolent) and chose which of the two games they would
limited-quantity scarcity promotion can lead to more aggres- like to play at that moment. The proportion of violent games
sive behavior (hypothesis 1). In addition, we extend the find- selected served as our measure of generalized aggression.
ings of the previous studies in two ways. First, we This study also served a third purpose by further disen-
demonstrate support for our proposed process: perceived tangling our proposed aggressive-competitive response
competitive threat to obtaining the target good. We found from a nonaggressive competitive goal by utilizing a de-
that exposure to limited-quantity scarcity promotions leads pendent measure that detects differences in aggression,
consumers to view other consumers as potential threats to while holding competition constant (choice of two com-
obtaining the target product (hypothesis 2), and this petitive video games). If our effects are specific to an ag-
increased threat mediates the relationship between the pro- gressive manifestation of competition, we should observe
motional ad and aggressive behavior. Second, study 4 ex- higher preferences for violent experiences after exposure
tended the generalizability of our effects by examining yet to the limited-quantity scarcity promotion (vs. control)
another type of aggression: physical violence through punch- among participants who did not reduce the competitive
ing. We found that individuals exposed to a limited-quantity threat via the affiliation task.
scarcity promotion physically threw more punches at a de-
fenseless, human-like individual than participants exposed to Procedure
a control ad. In study 5 we provide further support for our One hundred ninety-four participants (ages 1967,
proposed process by manipulating perceived threat. Mage 33.6, 43% female) from Amazon Mechanical
Turk participated in this study for $1, and were randomly
STUDY 5 assigned to conditions in a 2 (promotional ad: scarcity,
control) 2 (perceived competitive threat: high, low)
The goal of study 5 was three-fold. First, we wished to between-participants design. Participants were told they
provide further support for our proposed process (hypoth- would complete two unrelated studies: one on a consumer
esis 2) by manipulating perceived threat. If our theorizing promotion, and one on classic video games. Participants
is correct, factors that strengthen the degree to which other first completed the perceived threat manipulation, and
consumers are perceived as competitive threats will amp- then were immediately presented with our scarcity ma-
lify the degree to which aggressive tendencies are activated nipulation. Specifically, they viewed one of the two pro-
upon exposure to scarcity promotions. On the other hand, motional ads from study 1 and were told this promotion
anything that mitigates the degree to which other con- was from their local electronics retail store (no name or
sumers are perceived as competitive threats should reduce logo was present on the promotional ad). Next they read a
the activation of such tendencies, making consumers less scenario about lining up to participate in the local re-
likely to aggress after exposure. We manipulated perceived tailers sale with fellow local consumers. Participants
threat in study 5 by means of an established proxy: social read that they had arrived before the store opened and
694 JOURNAL OF CONSUMER RESEARCH
FIGURE 1
were positioned ahead of other local consumers near the MScarcity 1.49 vs. MControl 4.84; F(1, 156) 209.4, p
front of the line to enter once the doors opened. After < .001). No main effect (p > .75) or interactions with per-
completing cover story questions, participants completed ceived threat emerged (p > .13).
an attention check item that asked whether the person
next to them in line in the scenario was from their city or
Dependent Variable. We contrast-coded both promo-
a different city (0 person beside me was from my city,
tion ad (1 scarcity, 1 control) and perceived threat
1 person beside me was from a different city;
(1 low, 1 high) independent variables and entered
Oppenheimer, Meyvis, and Davidenko 2009). Finally,
them in a 2 2 ANOVA with the proportion of violent
participants moved on to the dependent measure: prefer-
ence for violent experiences. games selected as the dependent variable. The ANOVA re-
vealed a main effect of perceived threat (higher threat led to
Dependent Variable. In study 5, we operationalized ag- more violent game choices; PHigh Threat .37, SD .23 vs.
gression as a preference for violent experiences via video PLow Threat .30, SD .22, F(1, 156) 4.14, p .044),
game choice. Participants were presented with seven pairs but no effect of scarcity promotion (p > .20). However, the
of Super Nintendo games and chose which of the two main effect was qualified by the expected interaction (F(1,
games they wished to play right now. Pre-testing con- 156) 3.96, p .048).4 As predicted, participants in the
firmed that game pairings differed on perceived violence scarcityhigh-threat condition expressed a significantly
and not other attributes (see web appendix F for details). higher preference for violent games than participants in the
The number (i.e., proportion) of violent games selected controlhigh-threat condition (PScarcityHigh Threat .42 vs.
served as our dependent measure, which ranged from 0 (no PControlHigh Threat .32, F(1, 156) 4.42, p .032).
violent choices) to 1 (all violent choices). However, no differences between scarcity and control ads
emerged among participants in the low-threat condition
Results (PScarcityLow Threat .28 vs. PControlLow Threat .31, F(1,
Participants. Seven participants skipped though the 156) .46, p .499; see figure 1).5
scenario without reading and 27 participants failed 4
the instructional attention check and are excluded from the When participants who failed the attention check were included, the
predicted interaction remains consistent (F(1,183) 3.82, p .052) as
analysis. Thus, the analysis is conducted using 160 does the high-threat contrast (F(1,183) 4.65, p .032).
participants who completed the study as designed. The 5
We also note that participants in the scarcityhigh-threat condition
pattern of results is consistent if all participants are used. expressed a significantly higher preference for violent video games
than participants in each of the other three conditions individually
Manipulation Check. The manipulation check was suc- (PScarcityHigh Threat .42 vs. PScarcityLow Threat .28, F(1, 156)
cessful. Participants perceived the iPhone promotional 8.24, p .005; PScarcityHigh Threat .42 vs. PControlHigh Threat .32,
F(1, 156) 4.42, p .037; PScarcityHigh Threat .42 vs. PControlLow
package quantity to be more scarce in the scarcity versus Threat .31, F(1, 156) 4.69, p .032) as well as against the average
the control ad (1 very scarce to 7 very abundant; of other three conditions (F(1,156) 8.49, p .004).
KRISTOFFERSON ET AL. 695
STUDY 6 Procedure
Study 6 had two goals. The first was to provide further Two-hundred seventy-seven marketing undergraduates
conceptual support for the observed phenomenon and (ages 1849, Mage 21.9, 49.8% female) from Arizona
State University completed this study in exchange for
course credit and were randomly assigned to one of four
6
We also conducted a follow-up study (Amazon MTurk, n 176) to conditions in a 2 (promotional ad: scarcity, control) 2
ensure that participants made the explicit connection between affili-
ation with individuals in ones community and the potential threat of (perceived threat via brand association: high, low)
participating in the promotion. This study mirrored the main study between-participants design. Study 6 utilized the firearm
with one difference: the perceived threat manipulation occurred imme- shooting experimental paradigm from study 1.
diately after scarcity promotion exposure but before the shopping scen-
ario. Replicating the main study results, a significant interaction Experimental Manipulations and Stimuli. To ensure
emerged (p .020) with the scarcityhigh-threat condition expressing our effect was robust across products and product catego-
a significantly higher preference for violence when tested against the
controlhigh-threat condition (p .042) and the other three conditions ries, we selected luxury watches as the target product in
(p .043). the promotional ad. We created four promotional ads that
696 JOURNAL OF CONSUMER RESEARCH
FIGURE 2
NUMBER OF SHOTS FIRED AS A FUNCTION OF PERCEIVED THREAT VIA AGGRESSIVE BRAND IMAGE ASSOCIATION (STUDY 6)
featured a Tag Heuer luxury watch for $50 (retail price Aggressive Behavior. As in previous studies, shots
was $1,000) from either Walmart (high threat) or fired was severely nonnormal (!Shots Fired 4.50, x(273)
Nordstrom (low threat).7 We manipulated scarcity by vary- .91, p < .001) and was log-transformed. We contrast-
ing the quantity of watches available (scarcity 3, control coded promotional ad (1 scarcity, 1 control) and
3,000). Upon exposure to the promotional ad, partici- perceived threat (1 high, 1 low) and entered them
pants completed the same scarcity manipulation check as in a 2 2 ANOVA to predict shots fired. Results revealed
in previous studies (1 very scarce, 7 very abundant; a main effect of promotional ad (MScarcity 34.2,
see appendix D for stimuli), as well as an attention check SD 6.67 vs. MControl 24.4, SD 6.63, F(1,269) 6.11,
(consisting of recalling the promoting brand; Oppenheimer p .014), but no effect of perceived threat via aggressive
et al. 2009). brand association (p > .50). However, the expected inter-
Dependent Variable. Participants completed the first- action emerged (F(1,269) 3.89, p .049).8 Supporting
person shooting game and the number of shots fired served our predictions, when perceived threat was high, we repli-
as our dependent variable. cated the results from the previous studies. Participants
exposed to the scarcity promotion fired significantly more
Results bullets than participants exposed to the control promotion
(MScarcity 35.3, SD 6.69 vs. MControl 32.1,
Participants. Four participants failed the attention SD 7.26, F(1,269) 9.53, p .002). However, when
check and are excluded. Thus, the analysis is conducted perceived threat was low, no differences emerged between
using 273 participants. The pattern of results is consistent scarcity and control promotional ads (MScarcity 33.2,
if all participants are included. SD 6.53 vs. MControl 32.7, SD 6.07, F(1,269) .13,
Manipulation Check. The scarcity manipulation was p > .70; see figure 2).
successful. Participants who received the scarcity promo-
tional ad perceived the promotional quantity to be signifi- Discussion
cantly more scarce than participants who received the Study 6 provided further insight into consumer aggres-
control ad (MScarcity 1.90, SD 1.48 vs. MControl 4.75, sive responses to limited-quantity scarcity promotions by
SD 1.67, F(1,269) 221.0, p < .001). No main effects manipulating perceived threat using a factor relevant to
or interactions with brand emerged (ps > .23, NS). both consumer behavior researchers and marketers: brand
image. Results showed that when the association between
7 8
Both Walmart and Nordstrom sell Tag Heuer watches. Male and fe- When participants who failed the attention check were included, the
male watch ads were created and stimuli were matched to participant predicted interaction remains consistent (F(1,273) 2.86, p .092),
gender in the experiment. as does the high-threat contrast (F(1,273) 7.58, p .006).
KRISTOFFERSON ET AL. 697
brand and perceived threat was high, exposure to a limited- secure the promotional product as long as they arrive
quantity scarcity promotion led to increased aggression within the allotted time, the competitive threat other con-
among consumers; however, when the association between sumers play in inhibiting product acquisition is attenu-
perceived threat and brand was low, scarcity did not elicit ated. Thus, our framework predicts that aggressive
aggressive actions. behavior should not result from scarcity promotion ex-
We also conducted an additional experiment that manip- posure when the promotion limits the time the product is
ulated perceived competitive threat in yet another way: available. A separate pre-test confirmed that the competi-
through the ability to participate in the promotion. We rea- tive threat other consumers pose is perceived to be sig-
soned that when ability to participate in the promotion is nificantly higher in limited-quantity promotions than both
low, such as when the promotion takes place in a distant lo- limited-time promotions and promotions in which no
cation, perceived competitive threat of other consumers is mention of quantity or time is present (both of which did
reduced, and we should not observe the aggressive re- not differ from each other; see web appendix I for full
sponse. Thus, we adapted the previous promotional stimuli details).
such that the promotion was conducted at the University of In addition, in study 7 we sought to provide further
Vermont (pre-tested to be viewed neutrally, not as a rival construct clarity between aggressive and nonaggressive
university but one physically distant from participants competitive responses to scarcity promotions, by includ-
home university). ASU Undergraduates (n 239) took ing dependent measures of both. Specifically, in study 7
part in a 2 (promotional ad: scarcity, control) 2 we presented participants with an additional task (an on-
(perceived threat: high, low) between-participants design line word search game) that is competitive, but not ag-
in which participants evaluated one of the two iPhone pro- gressive. If scarcity promotions activate a generalized
motions from either the home university (high perceived competitive or achievement goal (vs. a specific aggres-
threat) or from the University of Vermont (low perceived sive manifestation of competition), then results should
threat). After exposure to the promotion, participants were align across both the aggressive and nonaggressive com-
given the opportunity to aggress via the firearm-shooting petitive tasks. However, consistent with our theorizing,
paradigm from study 1. Results revealed the expected we contend that only limited-quantity scarcity promo-
interaction (p < .05) and moderation. Participants in the tions will lead to an increase in aggressive competitive
high-threat condition (home university) fired significantly tasks.
more bullets after exposure to the scarcity versus control Finally, study 7 again utilized a controlinformation
ad, but the effect was attenuated when perceived threat was omitted condition in which no reference to product quan-
low (University of Vermont promotion; see web appendix tity or promotional time is present (as in study 2) to add
H for full details). Taken together, the results from the further confidence that the increased aggression is due to
main and ancillary studies provide further evidence for our exposure to scarcity (and not abundance). Our framework
claim that perceived threat of other consumers is the pro- predicts that participants exposed to the limited-quantity
cess through which limited-quantity scarcity promotions promotional ad will behave more aggressively than partici-
elicit aggressive behavior. We provide additional support pants exposed to the control ad; however, we expect no dif-
for our conceptual process in study 7 by examining a ferences in aggressive behavior between participants
boundary condition for our effects: type of scarcity exposed to the limited-time and control ads.
promotion.
Procedure
STUDY 7 One hundred fifty-two undergraduates (ages 1838, Mage
20.3, 50.9% female) from Arizona State University par-
The primary goal of study 7 was to provide further evi-
ticipated in this study in exchange for course credit and
dence for our process claim that the type of scarcity pro-
were randomly assigned to one of five conditions in a 2
motion affects the perceived competitive threat of other
(promotional ad: scarcity, control) 2 (scarcity type:
consumers and thus the likelihood of aggression in re-
quantity, time) 1 (controlinformation omitted) between-
sponse to scarcity (hypothesis 4). Marketers primarily use
participants design. Participants arrived in groups of eight to
two types of scarcity promotions to attract consumers:
12 and were seated at individual computer terminals. The
limited quantity and limited time. For example, while the
procedure of study 7 mirrored that of study 1 (firearm shoot-
majority of Black Friday promotions utilize limited-quan-
ing paradigm).
tity scarcity promotions, retailers often run promotions
throughout the year that offer a deal for a specific time Scarcity Manipulation. Participants received one of
period, with no limit on product quantity (e.g., One Day five iPhone promotional ads that served as our manipula-
Only, Sale Lasts Until Noon). Given that all con- tion. Participants in the quantity conditions either
sumers who participate in a limited-time promotion can received a promotional ad in which 3 (scarcity) or
698 JOURNAL OF CONSUMER RESEARCH
3,000 (control) products were available. Participants in fired was severely nonnormal (!Shots Fired 7.25, x(152)
the time conditions received a promotional ad in which .82, p < .001) and was log-transformed. We contrast-
the sale lasted for 1 day only (scarcity) or 30 days (con- coded both the scarcity condition (1 scarcity, 1
trol; see appendix E for time stimuli). Participants in the control) and scarcity type (1 quantity, 1 time) in-
controlinformation omitted condition received the iden- dependent variables and entered them in a 2 2 ANOVA
tical ad, but no reference to quantity or time was made. to predict aggressive behavior.9 Results revealed no ef-
Participants next completed the same cover story ques- fect of scarcity condition (p > .95) but a main effect of
tions as in previous studies and the following manipula- scarcity type (MQuantity 33.0, SD 5.37 vs. MTime
tion check items on a scale from 1 (very scarce) to 7 30.5, SD 5.75, F(1,130) 5.18, p .025). However,
(very abundant): How would you describe the university this main effect was qualified by the predicted two-way
bookstores iPhone 6 promotional package quantity (time interaction (F(1,130) 8.37, p .004). Supporting hy-
length)? pothesis 4 and replicating the previous results, partici-
Aggressive versus Nonaggressive Competitive pants exposed to the limited-quantity promotional ad
Behavior. Upon exposure to the promotion participants fired significantly more bullets than participants exposed
were given the classic video game cover story as in study to the control-quantity promotional ad (MLimited-Quantity
1; however, we modified the previous procedure such that 34.2, SD 6.28 vs. MControl-Quantity 31.8, SD 3.85,
participants played two competitive games: one that was F(1,130) 4.69, p .032). However, the time condition
aggressive and one that was not aggressive. To measure ag- did not elicit the same aggressive outcome, as partici-
gressive behavior, we utilized the first-person shooting pants in the limited-time condition in fact fired margin-
game (Deadeye) as in study 1, with the number of shots ally fewer bullets than participants exposed to the
fired serving as our dependent measure. To measure a gen- control-time promotional ad (MLimited-Time 30.0,
eralized (nonaggressive) competitive goal, participants SD 3.24 vs. MControl-Time 31.2, SD 7.64, F(1,130)
played a 1-minute round of an online word search game on 3.75, p .055). Although unexpected, this result may
the most difficult setting. Participants were instructed that have occurred because the limited-time promotion served
the goal of the game was to find as many words as possible as an explicit cue that minimized the perceived threat of
in the 1-minute round, and the number of words found other consumers and subsequent aggressive action by sig-
served as our dependent measure of nonaggressive com- nalling that aggression is not necessary. Importantly, to
petitive orientation. Presentation order of the two games show that the observed aggressive actions are caused by
was counterbalanced. As in study 1, we used the Morae re- an increase in shots fired in the limited-quantity condi-
cording software to remotely record and quantitatively tion, we performed a planned contrast against the con-
measure participant behavior for both games. Two coders, trolinformation omitted condition. Supporting our
blind to the hypothesis, watched each individual participant claim, participants exposed to the limited-quantity pro-
video and recorded both the number of shots fired (aggres- motion fired significantly more bullets than participants
sive competitive) and number of words found (nonaggres- exposed to the controlinformation omitted condition
sive competitive). (MLimited-Quantity 34.2, SD 6.28 vs. MControlInformation
Omitted 31.6, SD 7.67, F(1,147) 4.08, p .045; see
Results figure 3).10
Manipulation Checks. The two manipulation checks Aggressive versus Nonaggressive Competitive
revealed that the manipulations worked as intended. Behavior. Next, we sought to show that the observed ef-
Participants in the limited-quantity condition perceived the fects were specific to aggressive competitive behaviors and
iPhone promotional quantity to be significantly more not simply an increase in competitive behavior more
scarce than participants in control-quantity conditions broadly. We conducted the same ANOVA with the number
(MLimited-Quantity 1.76, SD 1.34 vs. MControl-Quantity
4.00, SD 1.61, F(1,130) 38.4, p < .001). Participants
9
in the limited-time condition perceived the iPhone promo- We conducted an ANOVA with presentation order of the two tasks
as an additional factor. No interactions with the scarcity condition or
tional time length to be significantly more scarce than par- scarcity type variables emerged (ps > .16). Moreover, the main effect
ticipants in the control-time condition (MLimited-Time and reported interaction remain significant if order is left in the model.
3.00, SD 1.94 vs. MControl-Time 4.83, SD 1.85, Thus, we collapsed across the order factor and report the 2 2
F(1,136) 17.3, p < .001). ANOVA.
10
We also analyzed the data using a one-way design to utilize all five
Aggressive Behavior. We first sought to test our pre- cells of the design. The ANOVA yielded significant differences
diction that the type of promotion moderated the relation- across conditions (F(4,147) 3.39, p .011), with participants in the
limited-quantity condition exhibiting significantly more aggression
ship between scarcity promotions and aggressive compared to participants in the other conditions (t(147) 3.02, p
behavior. As in previous studies, the number of shots .003).
KRISTOFFERSON ET AL. 699
FIGURE 3
of words found as the dependent measure. If exposure to a our overall framework and process claims. In addition,
limited-quantity ad elicited a general increase in competi- study 7 helped provide construct clarity, suggesting that
tive behavior, we should have observed the same pattern of scarcity promotions can alter consumer aggressive com-
results as the firearm shooting measure. However, support- petitive responses, but do not lead to a generalized achieve-
ing our theorizing that scarcity promotion exposure mani- ment motivation mind-set.
fests in aggressive behavior, no differences in the number
of words found emerged across conditions (main effect ps GENERAL DISCUSSION
> .30, interaction p > .16). While strong conclusions can-
not be drawn from a null result, subsequent analysis Across seven studies, we demonstrate the existence of a
showed that the correlation between shots fired and num- dark side to scarcity promotions: aggression. We propose a
ber of words was negative (r .23, p .004), suggesting framework to outline the process that drives when and why
that the two outcomes were operating independently, and consumers may turn to violence and identify contexts
lending support for our claim that the effect is specific to when scarcity promotions will not lead to aggression. In
aggressive behavior. study 1, we found that consumers exposed to limited-
quantity scarcity promotions behave more aggressively
Discussion than participants exposed to a control ad featuring identical
highly desired products. Operationalizing aggression as
The results of study 7 provide further insight into ag- firearm shooting behavior, we found that exposure to a
gressive reactions to scarcity promotions by examining a limited-quantity scarcity promotion led to participants fir-
moderator of both theoretical and practical importance: ing significantly more bullets than exposure to a control
type of scarcity promotion. When participants were ad. In a real consumption context, study 2 showed that par-
exposed to promotional material that limited the quantity ticipants exposed to a limited-quantity scarcity promotion
of the desired product and emphasized competition be- physically assaulted a vending machine significantly more
tween consumers, we replicated the results of studies 16 than participants exposed to a control promotion, and a
such that participants reacted with increased aggression. promotion in which no mention of quantity was present.
However, promotional material that limited time did not Study 3 provided support for our proposed process and
elicit the same aggressive outcomes. Given that the threat demonstrated that exposure to a limited-quantity scarcity
of other consumers limiting ones ability to obtain the de- (vs. control) promotion led to increased levels of testoster-
sirable product is present in limited-quantity, but not lim- onea hormone shown to be predictive of aggressive be-
ited-time, promotions, the results of study 7 further support havioramong participants. Study 4 demonstrated support
700 JOURNAL OF CONSUMER RESEARCH
for our claim that realized aggression is driven by the per- focal product, and prepares consumers to aggress by
ceived competitive threat of other consumers. Using phys- increasing testosterone levels. To our knowledge, this is
ical punching as our behavioral measure of aggression, we the first research to empirically demonstrate that a market-
showed that exposure to limited-quantity scarcity promo- ing tactic can automatically elicit physiological changes
tion advertising led consumers to perceive others as poten- and, in turn, lead consumers to behave aggressively when
tial competitive threats to obtaining the target product. given the opportunity to do so. We provide evidence for
This increased threat, in turn, led to participants throwing this claim by both directly measuring testosterone levels
significantly more punches at a defenseless target. Study 5 and perceived threat, and by manipulating threat using so-
provided further support for our proposed process by cial connection, brand image, and ability to participate in
manipulating perceived threat using social affiliation as a the promotion as proxies. Importantly, we provide further
proxy for threat. Study 6 manipulated perceived competi- evidence for our proposed process examining the type of
tive threat via brand image association using two real scarcity promotion. Specifically, we find that aggressive
brands and a different focal product. Finally, by examining responses to scarcity result when the promotion limits
promotion type, study 7 showed that the aggressive reac- available quantity, but not available time. This is because
tion to scarcity promotions occurs when the promotion lim- under quantity restraints, obtaining the promotion before
its quantity, but not time. other consumers is imperative to successful acquisition;
thus, the perceived competitive threat of other consumers
Contributions is high. However, as long as all consumers who choose to
This research makes a number of theoretical and sub- participate in a limited-time promotion arrive within the
stantive contributions. Theoretically, we add to the scarcity allotted window to obtain the product, the competitive
literature by showing that the observed response of aggres- threat other consumers pose is attenuated and aggression
sion extends beyond life-threatening and retailer-induced does not result.
consumption contexts and can result from exposure to Finally, our research makes important substantive con-
advertising featuring nonessential goods. To our know- tributions to firms, consumers, and policy makers. For
ledge, this research is the first to show that advertising can practitioners, the knowledge that scarcity promotions may
drive consumers to aggress in response to scarcity of activate aggression among consumers allows firms to de-
nonnecessity luxury items. The present research contrib- sign promotions more effectively by better managing both
utes by showing that aggressive tendencies from scarcity product availability and potential altercations that can lead
not only happen in such contexts, but more importantly can to costly negative consumer-brand experiences (e.g., more
originate from mere exposure to scarcity marketing mater- efficient retail layouts, multiple sales channels). Further,
ials and generalize outside of the promotional context. marketers could choose to implement limited-time promo-
This suggests that marketplace aggression can actually be tions if they seek to reduce the likelihood of physical alter-
activated before a consumer even reaches the store. cations and the accompanying media hype. From a
We also theoretically contribute to recent scarcity find- consumer standpoint, those planning to purchase a limited-
ings (Shah et al. 2012; Roux et al. 2014) by demonstrating quantity product may seek out alternative points of pur-
that scarcity promotions can lead consumers to engage in chase, such as online or secondhand outlets, rather than
dangerous and antinormative behavior toward others. Our physically shopping inside the retail outlet. Moreover, if
results support previous work by showing that scarcity consumers choose to participate in the promotion alongside
cues can lead to an increased focus on ones present envir- other consumers, awareness of potential aggressive tenden-
onment, but extend knowledge by showing that consumers cies a priori can help to control ones responses and minim-
may be willing to risk significant costs to achieve their pre- ize the potential for harm. Finally, from a public policy
sent goal. Further, we show these tendencies can result in perspective, a better understanding of negative consumer
generalized aggressive actions, meaning that the target of responses to scarcity promotions can help policy makers
the aggression need not be a person who is actually com- better regulate these types of marketing tactics to pro-
peting for the scarce items, nor does the target need be a actively protect consumer welfare. This may be achieved
person at all. by ensuring adequate staff-to-consumer ratios, requiring
In addition to showing the negative behavioral outcome trained security staff in retail stores during promotional
of aggression, we provide evidence for the process driving periods, or directly regulating the use of promotional tac-
our effects. We find that exposure to limited-quantity scar- tics that employ scarcity methods. These types of policy
city promotion advertising can lead consumers to perceive changes are likely to ensure the safety of consumers and
others as potential competitive threats to obtaining the firm employees alike.
KRISTOFFERSON ET AL. 701
Limitations and Directions for Future Research compensatory shopping behavior. Another important ex-
tension for future research could be to examine the dur-
While we believe this research makes important contri-
ation of our documented aggressive responses to scarcity
butions to both theory and practice, it is not without its
promotions. In our investigation, aggressive behavior was
limitations. One such limitation is relationship clarity re-
measured shortly after exposure to the promotional ad, but
garding our psychological and physiological process
does this response dissipate or get worse over time? Given
claims that drive generalized aggression in response to
the multitude of scarcity promotions occurring during
scarcity promotions. Specifically, while we empirically
shopping-oriented holidays such as Black Friday, it might
demonstrate that limited-quantity scarcity promotions both also be interesting to examine how exposure to multiple
prompt consumers to perceive others as competitive threats ads within a short period of time interacts with aggressive
and elicit physiological responses associated with aggres- responses.
sion, we are unable to definitively isolate the relationship In conclusion, our research demonstrates that scarcity-
between these two processes. Do they operate in a causal driven aggression is not confined to life-threatening or
chain such that scarcity promotion exposure elicits the even retailer-induced consumption environments, but can
automatic physiological response of increased testosterone, result from mere exposure to scarcity promotional advertis-
which in turn leads to threat perception and subsequent ag- ing. Across seven studies, multiple contexts, and numerous
gression, or do they operate independently (e.g., in paral- behavioral measures of aggression, we find that exposure
lel) and jointly drive subsequent aggression? Currently, we to limited-quantity scarcity promotions leads consumers to
are unable to clarify this relationship and call on future re- behave more aggressively. We show that scarcity promo-
search to examine the relationship between these con- tion advertising drives consumers to perceive others as po-
structs more closely. tential threats to obtaining the desired product, biologically
Moreover, further clarity is needed regarding how con- prepares the body to aggress, and leads to violence when
sumers actually experience the heightened competitive the opportunity arises. Thus, when the doors open on Black
threat upon exposure to scarcity promotions. Do con- Friday and the consumers rush in, racing toward the few
sumers explicitly visualize or anticipate an aggressive al- discounted items, the aggression that ensues likely origi-
tercation upon exposure to the promotional ad, or do they nated long before they entered the store, as soon as they
just think more generally about the fact that other shop- saw the first Black Friday ad.
pers have also seen the ad and will want to obtain the
product as well? While our results appear to be consistent
DATA COLLECTION INFORMATION
with a preparation-to-interact/visualization type of pro-
cess documented to result from automatic primes The first author supervised the collection of data by re-
(Berkowitz and LePage 1967; Cesario, Plaks, and search assistants for study 1 at the University of British
Higgins 2006), it is also possible that consumers do not Columbia in spring 2013 and analyzed the data. The first
explicitly try to visualize a specific altercation between author supervised the collection of data by research assist-
them and other shoppers but instead form a more vague ants for study 2 at the University of British Columbia in
perception of other shoppers as a potential competitive spring 2015 and analyzed the data. The third author super-
threat. Our data does not definitively answer what it vised the collection of data by research assistants for study
means for participants to perceive other consumers as a 3 at Arizona State University in spring 2015. The Arizona
competitive threat, and future research is needed to un- State University Institute for Interdisciplinary Salivary
cover what this process entails psychologically, as well as Bioscience Research generated the data and all authors
physiologically. analyzed the data. The third author supervised the collec-
We believe our work provides researchers many oppor- tion of data by research assistants for study 4 at Arizona
tunities for future research. For example, a natural exten- State University in winter 2013 and all authors analyzed
sion of our work would be to examine the relationship the data. The first author collected the data for study 5 on
between scarcity promotions and aggression in online Amazon Mechanical Turk in spring 2014 and analyzed the
shopping environments. We used physical ads to manipu- data. The third author supervised the collection of data by
late scarcity, but might our findings differ in online shop- research assistants for study 6 at Arizona State University
ping contexts? Moreover, while we focused our in spring 2014 and the first author analyzed the data. The
investigation on aggressive reactions to scarcity promo- third author supervised the collection of data for study 7 at
tions, future research could examine other downstream Arizona State University in winter 2014 and the first author
consequences, such as store or brand evaluation or analyzed the data.
702 JOURNAL OF CONSUMER RESEARCH
APPENDIX A APPENDIX B
STUDY 1 STIMULI STUDY 2 CONTROL STIMULI
APPENDIX C
STUDY 2 PRODUCT JAM
KRISTOFFERSON ET AL. 703
APPENDIX D
WALMART/NORDSTROM PROMOTIONS
(STUDY 6)
APPENDIX E
TIME PROMOTION STIMULI (STUDY 7)
704 JOURNAL OF CONSUMER RESEARCH
Wellen, Jackie M., Michael A. Hogg, and Deborah J. Terry Relationship Between Close Sporting Events and Game-Day
(1998), Group Norms and AttitudeBehavior Consistency: Traffic Fatalities, Journal of Consumer Research, 38 (4),
The Role of Group Salience and Mood, Group Dynamics: 61121.
Theory, Research, and Practice, 2 (1), 4856. Zimbardo, Philip G. (1969), The Human Choice: Individuation,
Wilson, Margo and Martin Daly (1985), Competitiveness, Risk Reason, and Order versus Deindividuation, Impulse,
Taking, and Violence: The Young Male Syndrome, and Chaos, Nebraska Symposium on Motivation, 17,
Ethology and Sociobiology, 6 (1), 5973. 237307.
Wood, Stacy, Melayne Morgan McInnes, and David A. Norton Zwane, Alix Peterson (2012), Implications of Scarcity, Science,
(2011), The Bad Thing About Good Games: The 338 (6107), 61718.
Copyright of Journal of Consumer Research is the property of Oxford University Press / USA
and its content may not be copied or emailed to multiple sites or posted to a listserv without
the copyright holder's express written permission. However, users may print, download, or
email articles for individual use.