Content uploaded by Margaret C Campbell
Author content
All content in this area was uploaded by Margaret C Campbell on Nov 07, 2015
Content may be subject to copyright.
Journal of Consumer Research Inc.
Seeing Is Eating: How and When Activation of a Negative Stereotype Increases Stereotype-
Conducive Behavior
Author(s): Margaret C. Campbell and Gina S. Mohr
Reviewed work(s):
Source:
Journal of Consumer Research,
Vol. 38, No. 3 (October 2011), pp. 431-444
Published by: The University of Chicago Press
Stable URL: http://www.jstor.org/stable/10.1086/659754 .
Accessed: 31/10/2011 14:05
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
The University of Chicago Press and Journal of Consumer Research Inc. are collaborating with JSTOR to
digitize, preserve and extend access to Journal of Consumer Research.
http://www.jstor.org
431
䉷2011 by JOURNAL OF CONSUMER RESEARCH, Inc. ●Vol. 38 ●October 2011
All rights reserved. 0093-5301/2011/3803-0007$10.00. DOI: 10.1086/659754
Seeing Is Eating: How and When
Activation of a Negative Stereotype
Increases Stereotype-Conducive Behavior
MARGARET C. CAMPBELL
GINA S. MOHR
This research investigates the effect of activation of a negative stereotype on
behaviors that are perceived to increase the chance of becoming a member of the
stereotyped group. Activation of a negative stereotype (the overweight stereotype)
is shown to lead to stereotype-consistent goal commitment (low health goal com-
mitment), which partially explains increases in stereotype-conducive behavior(eat-
ing indulgent foods). Two theoretically relevant moderators are proposed and sup-
ported. Increased accessibility of the countervailing health goal and increased
accessibility of the link between the behavior and membership in the stereotyped
group both limit the effect of stereotype activation on stereotype-conducive be-
havior. Five experiments support the facilitative effect of stereotype activation on
stereotype-conducive behavior, the role of goal commitment, and both moderators.
Consider the following: Your friend Lucy, who is about
25 pounds overweight, e-mails you pictures from her
recent vacation. After you look at Lucy’s pictures, the office
secretary comes by with a plate of cookies. Will exposure
to someone who is overweight influence how many cookies
you eat? Nineteen adult consumers indicated how they
thought seeing someone overweight would influence their
consumption of indulgent food (e.g., cookies or candy).
About 42% said that they would take the same amount as
they would without exposure, 26.3% said that they would
take some but less than if they had not seen the overweight
person, and 31.6% said that they would not take any after
seeing someone overweight. The fact that many consumers
Margaret C. Campbell (meg.campbell@colorado.edu) is associate pro-
fessor of marketing at the Leeds School of Business, University of Colorado
at Boulder, 419 UCB, Boulder, CO 80309. Gina S. Mohr (slejko@colorado
.edu) received her PhD from the Leeds School of Business, University of
Colorado at Boulder. Address correspondence to Margaret (Meg) Campbell.
The authors thank Paul Herr, John Lynch, and participants in research
seminars at the University of Alberta, University of California, San Diego,
University of Missouri, University of Washington, and University of West-
ern Ontario for helpful comments at various stages of this research. In
addition, the authors acknowledge the helpful comments of the reviewers,
associate editor, and editor.
John Deighton served as editor and Gavan Fitzsimons served as associate
editor for this article.
Electronically published March 17, 2011
have a goal to be healthy and do not want to be overweight
may explain why none of the consumers selected the answer
“I would take more than if I hadn’t seen the overweight
person.” However, recent research suggests that exposure to
an overweight person may in fact lead to an increase in
food consumption. Christakis and Fowler (2007) indicate
that if a close other in a consumer’s social network gains
weight, the consumer is likely to gain weight. McFerran et
al. (2010a) show that consumers anchor on the food choices
of others, taking more snack food when another does even
when the other is obese. We contribute to this discussion
by (1) showing that beyond anchoring on others’ eating
behavior, merely being exposed to someone overweight can
result in increased food consumption and (2) proposing the
process and conditions that lead to this increase.
This research presents and tests a theoretical framework to
understand how exposure to someone a consumer does not
wish to emulate (e.g., someone overweight) can lead to an
increase in behavior that is perceived as causal in becoming
more like the person (e.g., eating). We draw from the stereo-
type priming literature to explain how exposure to a negative
social group member can increase causally related behavior
despite an underlying goal that could override such behavior.
Recent research demonstrates that stereotypes include goals
(Van Boven, Campbell, and Gilovich 2010) and can influence
goal-directed behavior (Aarts, Gollwitzer, and Hassin 2004).
Building on this, we propose that stereotypes can include
commitment to pursue associated goals, including counter-
432 JOURNAL OF CONSUMER RESEARCH
vailing goals. We argue that the effect of negative stereotype
activation on stereotype-conducive behavior (i.e., behavior
that is causally linked to group membership) can be explained
in part by the effect that stereotype activation has on a con-
sumer’s current commitment to an associated countervailing
goal.
Five experiments show that exposure to a negative ste-
reotype (i.e., seeing someone overweight) can increase ste-
reotype-conducive behavior (i.e., eating indulgent food). Ex-
periment 1 demonstrates increased stereotype-conducive
behavior following stereotype activation. Experiment 2 rep-
licates this and supports the proposed mediating role of
stereotype-consistent commitment to a countervailing goal
(i.e., the goal to be healthy). Experiment 3 provides addi-
tional support for the proposed process by demonstrating a
moderating effect of accessibility of the countervailing goal.
Experiments 4 and 5 demonstrate that increased accessibility
of the link between group membership and the behavior
attenuates the effect of the prime. Experiment 5 compares
two explanations, providing additional evidence in support
of the proposed process.
THEORETICAL BACKGROUND
Stereotypes are bundles of characteristics including traits,
attitudes, behavioral tendencies, and goals that are as-
sociated with the members of a social category (e.g.,
Aarts et al. 2005; Hilton and von Hippel 1996; Stangor
and Lange 1994). Activation of a social stereotype refers
to increased accessibility of associations with the social
group. Research demonstrates that both conscious and non-
conscious exposure to a stereotyped group member (e.g.,
by a photograph) can result in the automatic activation of
stereotypic characteristics (Blair and Banaji 1996; Kawak-
ami, Dion, and Dovidio 1998). Research consistently shows
that activation of a stereotype can influence behavior, typ-
ically resulting in an increase in stereotype-consistent be-
havior, even when the stereotype and related behavior are
somewhat negative (for reviews, see Dijksterhuis and Bargh
[2001] and Wheeler and Petty [2001]). For example, col-
lege students’ scores on general knowledge questions in-
creased after exposure to a professor but decreased after
exposure to a supermodel prime (Dijksterhuis and van
Knippenberg 1998).
It is noteworthy, however, that existing stereotype prime
research focuses on behaviors that are associated with an
activated social stereotype but not ones that are stereotype-
conducive, that is, perceived to be causally linked to group
membership. This lack of examination of stereotype-con-
ducive behaviors may be because most groups studied are
ones in which membership is deterministic, such that people
cannot typically engage in behaviors that will make them a
member of the group (e.g., a white cannot become Asian
by engaging in certain behaviors). However, even in the
case of research on nondeterministic groups, the behaviors
studied are not causal to group membership (e.g., a low
number of correct general knowledge answers does not
increase the probability of becoming a supermodel).
This gap in the literature raises the question of the extent
to which stereotype-conducive behaviors associated with a
negatively stereotyped group will show the same stereotype
prime effects as seen for stereotypical but nonconducive be-
haviors. The impact of stereotype activation on stereotype-
conducive behavior is particularly intriguing in the domain
of negative stereotypes because of the potential for goal sys-
tems involvement that could limit prime effects (Kruglanski
et al. 2002). In the case of negative stereotypes, people have
countervailing goals that could override stereotype-conducive
behavior. Consumers sometimes avoid actions, choices, and
products that are linked to out-groups because of goals to
avoid being associated with a negatively viewed group (Ber-
ger and Heath 2008; White and Dahl 2006). Research also
demonstrates that goals can limit response to tempting be-
haviors that run counter to goal achievement (e.g., Fishbach,
Friedman, and Kruglanski 2003). Thus, it seems possible that
consumers’ goals could limit prime effects on behaviors per-
ceived as conducive to membership in a dissociative social
group. For example, the countervailing goal to be a healthy
weight could override the stereotype-conducive behavior of
eating. This is the intuition demonstrated by the majority of
participants in the lay theory study described above who in-
dicated that they would eat less after seeing someone over-
weight.
However, for a countervailing goal to override activated,
goal-inconsistent behavior, the goal must be cognitively ac-
cessible and the person must feel committed to pursue the
goal (Shah and Kruglanski 2002; Shah, Kruglanski, and
Friedman 2003). Fishbach et al. (2003) show that exposure
to a temptation that could derail goal pursuit leads to in-
creased accessibility of the goal; the accessible goal appears
to enable resistance of the temptation, whereas giving into
temptation is more likely when the overriding goal is not
accessible. Goal commitment also has an important influ-
ence on goal pursuit: high goal commitment increases ac-
tions that favor the goal, whereas low goal commitment does
not (Fishbach and Dhar 2005). Given that stereotype acti-
vation is highly likely to result in increased accessibility of
stereotype-conducive behavior and the goal to pursue the
behavior, such behavior will increase unless a countervailing
goal is likewise accessible and the consumer is committed
to that goal. We propose that stereotype activation can result
in decreased commitment to the countervailing goal.
Stereotype Exposure and Stereotype-Consistent
Goal Commitment
We argue that stereotype activation increases accessibility
of goals and goal commitment associated with the stereotyped
group. Recent research demonstrates that stereotypes include
the goals that are stereotypical for the group (Van Boven et
al. 2010) and that stereotype activation can lead people to
automatically take on these goals themselves (Aarts et al.
2005). For example, people infer goals from others’ actions
and can exhibit “goal contagion” such that they pursue the
inferred goals themselves (Aarts et al. 2004, 2005). Building
INCREASED STEREOTYPE-CONDUCIVE BEHAVIOR 433
on this, we suggest that it is likely that exposure to a stereo-
typed person will activate knowledge that includes stereo-
typical goals and, additionally, motivation or commitment to
these goals. Thus, consumers will be likely to adopt the goals
and goal commitment considered to be stereotypical.
Stereotype knowledge can include high and/or low com-
mitment to particular goals. For example, the stockbroker
stereotype includes high and the journalist stereotype in-
cludes low commitment to the goal of making money (Aarts
et al. 2005), while the materialistic stereotype includes high
commitment to pursue external rewards as well as low com-
mitment to pursue intrinsic rewards (Van Boven et al. 2010).
For a negative, nondeterministic stereotype, it is likely that
the stereotype includes high commitment to pursue stereo-
type-conducive behaviors and low commitment to a coun-
tervailing goal that would limit those behaviors. That is, by
virtue of a person’s membership in a negatively stereotyped
group, others are likely to infer that the person has low
commitment to a goal that would prevent group member-
ship. Because having high commitment to a goal to be
healthy could limit actions that lead to overweight (Fishbach
and Dhar 2005), consumers infer that overweight people
have low commitment to the health goal.
We propose that stereotype activation results in increased
accessibility of interrelated knowledge that includes stereo-
type-consistent traits, behaviors, goals, and commitment to
goals. Goal-inconsistent behavior is affected by the interplay
among the accessibility of the behavior and the goal to
pursue the behavior, the countervailing goal, and current
commitment to the goal. Because stereotype activation in-
creases accessibility of the goal to pursue stereotype-con-
ducive behavior (Aarts et al. 2005) and low commitment to
the countervailing goal, we propose that stereotype priming
will increase stereotype-conducive behavior even when the
stereotype is negative. That is, we propose that exposure to
a negative stereotype increases stereotype-conducive be-
havior in part by affecting commitment to the goals linked
to the stereotype and behavior. Exposure to the stereotype
attenuates the consumer’s own commitment to pursue the
countervailing goal, resulting in an increase in the primed,
but goal-inconsistent, behavior. An important contribution
of our research is that we present and test a theoretical
framework that provides insight into the question of how
exposure to a negative stereotype can lead to increases in
behavior consistent with, and conducive to, the stereotype
while inconsistent with a continuing goal. This framework
also suggests two factors that can limit the effect of stereo-
type exposure on stereotype-conducive behavior.
Limits to the Impact of Negative Stereotype
Activation on Stereotype-Conducive Behavior
Behavioral assimilation to an active stereotype can be
beneficial for learning, group cohesion, and even safety
(Chartrand and Bargh 1999). However, Dijksterhuis and van
Knippenberg (2000) raise the issue that stereotype prime
effects can also be dysfunctional and that thus some factors
may limit such effects. The above theoretical development
suggests two possible limits to the proposed counterintuitive
increase in stereotype-conducive behavior linked to a neg-
atively stereotyped group.
First, this research proposes that the attenuating effect
of the stereotype on commitment to a countervailing goal
is one mediator of the effect of the stereotype on stereo-
type-conducive behavior. Based on this, we propose that
increasing the accessibility of the consumer’s underlying
countervailing goal will moderate the effect of stereotype
activation on stereotype-conducive behavior. Research
demonstrates that increased accessibility of an alternative
goal can interfere with behavior directed at a focal goal
(Shah and Kruglanski 2002) and that activation of a coun-
tervailing goal moderates the effect of priming temptations
(Fishbach et al. 2003). Macrae and Johnston (1998; ex-
periment 2) found that priming the concept of helpfulness
increased the extent to which participants helped an ex-
perimenter except when they were running late for another
experiment. An interpretation of these results is that the
heightened accessibility of the underlying goal (being
timely) limited the effect of the primed concept or goal
(to be helpful) on a behavior counter to the underlying
goal (taking time to help the experimenter). Thus, we pro-
pose that increasing the accessibility of a countervailing
goal can moderate the impact of social-stereotype acti-
vation on a stereotype-conducive behavior inconsistent
with the countervailing goal.
Second, one reason that the proposed increase in ste-
reotype-conducive behavior following negative stereotype
exposure is counterintuitive is that people are motivated
to avoid behaviors that bring them closer to dissociative
groups (Berger and Heath 2008; White and Dahl 2006).
This suggests that increasing the accessibility of the per-
ceived causal link between the stereotype-conducive be-
havior and group membership can limit the effect of ste-
reotype exposure on this behavior. We propose that when
the connection between the behavior and group member-
ship is highly accessible to a consumer, negative stereotype
activation will not lead to the same increase in stereotype-
conducive behavior as when the link is less accessible. In
support of this thinking, Johnston (2002) found that people
mimicked a stigmatized other’s behavior except when the
behavior was causally linked to the stigma. We propose
that this may be because seeing the person engage in the
behavior increased the accessibility of the causal link.
When the behavior-stereotype link is accessible, it can pro-
vide a stronger guide for behavior than the motivations
and behaviors associated with the stereotype prime (Dijk-
sterhuis and van Knippenberg 2000). Thus, we hypothesize
that heightened accessibility of the behavior-stereotype
link will limit the effect of a negative stereotype on ste-
reotype-conducive behavior.
RESEARCH OVERVIEW
Five experiments explore the proposed framework using
overweight as the social stereotype and eating indulgent
434 JOURNAL OF CONSUMER RESEARCH
(i.e., tasty, calorie-dense) food as the stereotype-conducive
behavior. There is a commonly held negative stereotype of
the overweight that includes eating indulgent foods and low
commitment to health goals; this is held by healthy weight,
overweight, and obese people (Crandall 1994; Roehling
1999; Schwartz et al. 2006). Research shows that it is com-
monly believed that (over)eating indulgent food is a cause
of being overweight (e.g., Bacon, Scheltema, and Robinson
2001). Importantly, recent studies show that over 70% of
Americans are concerned with their weight and that over
half of both men and women are actively pursuing the goal
to lose weight (e.g., International Food Information Council
2009; Spake 2004). The presented research contributes to
recent research showing that others’ weight can affect con-
sumers’ self-perceptions, food choices, and weight (e.g.,
Christakis and Fowler 2007; McFerran et al. 2010a, 2010b;
Smeesters, Mussweiler, and Mandel 2010) by demonstrating
that seeing someone who is overweight can increase food
consumption and showing how and when stereotype acti-
vation plays a role in the impact of others’ body weight on
food consumption.
EXPERIMENT 1: SEEING SOMEONE
OVERWEIGHT AND FOOD CHOICE
Experiment 1 examines whether exposure to a picture of
someone who is overweight leads participants to take more
food even though many consumers wish to avoid member-
ship in the stereotyped group (i.e., do not wish to be over-
weight) and believe that the behavior (eating indulgent
foods) is conducive to group membership. A pretest con-
firmed that a picture of someone who is overweight in-
creases the accessibility of overweight stereotypic associ-
ations. Participants (np55) randomly assigned to either
see an overweight person or not (between subjects) com-
pleted a lexical-decision task with stereotypic (e.g., indul-
gent, inactive, unhealthy) and nonstereotypic (e.g., rectan-
gle, briefcase, key) words (within subject). After eliminating
three suspicious participants and response latencies over 3
standard deviations above the mean,a2(overweightprime:
yes, no) #2 (word type: nonstereotypic, overweight ste-
reotypic) mixed-model ANOVA with word type treated as
a repeated measure revealed the expected significant inter-
action (F(1, 49) p5.61, p!.02). Participants who saw a
picture of an overweight person responded significantly fas-
ter to overweight stereotypic words (Mp589 milliseconds)
than did control participants (Mp713 milliseconds; t(49)
p2.46, p!.02); there was no significant effect of prime
on response times to nonstereotypic words (t(49) p1.63,
p1.12).
Method
Participants and Design. A researcher recruited 59 peo-
ple walking though a lobby at the University of Colorado,
Boulder (students, faculty, staff, and visitors), to participate
in a short survey. Ages ranged from 18 to 69 (Mp25),
and 64% were male; since there were neither main nor
interaction effects of gender, gender is not discussed fur-
ther. A pretest (np37) was conducted to identify pictures
of women perceived to be overweight (Mp12.6 on an
18-point weight scale) and healthy weight (Mp7; tp
7.4, p!.0001; see fig. 1). Participants were randomly
assigned to see one picture either of an overweight or a
healthy weight woman or of a lamp (the lamp provided a
no-person control).
Procedure. The surveys had a front page with the cover
story and a few questions (age, gender, and affiliation), al-
lowing the researcher to remain blind to condition. After
reading that they were helping to calibrate pictures for later
studies, participants turned the page, saw one of the three
pictures, and were asked to “list the first three things that
come to your mind” and rate the picture on typicality, qual-
ity, and clarity. They then closed the booklet, returned it to
the researcher, and helped themselves from a bowl of candy
as a thank you. The number of candies that each participant
took was recorded after they left, and this was the dependent
variable.
Results
The number of candies taken was analyzed as a function
of the three-level prime, with age and affiliation (staff, fac-
ulty, undergraduates, and visitors) as covariates. Results re-
vealed only a significant effect of prime (F(2, 53) p3.88,
p!.03). In support of our prediction, planned contrasts
revealed that those who saw the image of an overweight
person took significantly more candies (Mp2.2) than those
in the healthy weight (Mp1.4; F(1, 53) p7.03, p!.02)
or in the neutral control condition (Mp1.5; F(1, 53) p
4.77, p!.04). There was no difference between those in
the healthy weight and neutral conditions (F(1, 53) p.25,
NS).
Discussion
These results reveal that activating a negative stereotype
can increase stereotype-conducive behavior. When exposed
to someone overweight, people took more candies compared
to those who were exposed to either a neutral or a healthy
weight prime. Choice is an important variable, but we also
want to examine the impact of the stereotype on food con-
sumption. Further, the natural environment in experiment 1
precluded the use of process measures. Understanding the
process that drives the prime effect is an important next
step.
EXPERIMENT 2: THE MEDIATING ROLE
OF HEALTH GOAL COMMITMENT
The primary purpose of experiment 2 was to examine the
process underlying the effect of negative-stereotype activation
INCREASED STEREOTYPE-CONDUCIVE BEHAVIOR 435
FIGURE 1
PRETEST RESULTS FOR PERCEIVED WEIGHT OF HEALTHY WEIGHT AND OVERWEIGHT PHOTOS: EXPERIMENT 1
on stereotype-conducive behavior. We measured health goal
commitment to examine the prediction that overweight ste-
reotype activation decreases health goal commitment and this
decreased commitment mediates the effect of the overweight
prime on eating. We included a delay condition to examine
effect persistence and also measured attitude toward the food
to examine the alternative hypothesis that stereotype activa-
tion increases attitude toward the food, which leads to in-
creased food choice and consumption.
Method
Three pretests verified that low health goal commitment
is part of the overweight stereotype. Participants (np135)
saw pictures of four healthy weight and four overweight
people, rated each on a 7-point scale from unhealthy to
healthy, and indicated perceived weight on an 18-point scale.
Each of the overweight people was perceived to weigh more
and be less healthy than each of the healthy weight people
(all p’s !.0001). In a second pretest, 28 participants saw a
picture of an overweight or a healthy weight woman and
predicted how the woman would answer four health goal
commitment questions (i.e., “I reflect a lot about my health;
I’m constantly examining my health; I am very self-con-
scious about my health; I am not very involved with my
health” [reverse scored]; Gould 1990). Participants thought
that the healthy woman had higher health goal commitment
(Mp3.75) than the overweight woman (Mp3.14, t(26)
p2.0, p!.055). In a third pretest 19 participants indicated
the extent to which they perceived groups to be motivated
by several goals on a 9-point scale from not at all (1) to
very strongly (9), based on Aarts et al. (2005). The over-
weight group was perceived to be motivated by the goal to
eat tasty food (Mp7.58) and to have lower commitment
to the goal to be healthy (Mp3.37) than a nonoverweight
group (Mp4.45; t(18) p6.06, p!.0001).
Participants and Procedure. For partial course credit,
139 students (average age 21, age range 18–38; 71% male)
seated in private cubicles participated in several studies.
Participants were first given a study with a cover story on
understanding aspects of digital pictures. Participants looked
at three pictures on a computer display (counterbalanced for
order). Participants were randomly assigned to either the
overweight prime, in which they saw one neutral picture (a
tree) and two pictures of overweight people, or the neutral
prime condition, in which they saw three neutral pictures
(the tree, a fishbowl, and a lake). Participants wrote three
things about the picture and rated the color, typicality, qual-
ity, and clarity on 5-point scales.
436 JOURNAL OF CONSUMER RESEARCH
Participants were then given two additional studies, an
unrelated product study and a taste test, counterbalanced for
order between sessions. The product study involved pro-
viding evaluations of nine products (e.g., Band Aids, a side
table, pasta), and it took about 5 minutes to do. When this
study was first, it provided a delay between the prime and
the taste test. Either an increase or no change in the effect
over the delay provides evidence of goal-system involve-
ment, whereas a decline in the effect over the 5 minute delay
period suggests only semantic priming (e.g., Aarts et al.
2004; Bargh et al. 2001). Thus, manipulating delay allows
exploration of whether the stereotype activation includes
stereotypical goal commitment (as proposed) or only primes
semantic knowledge associated with the stereotype.
For the taste test study, participants were provided with a
bowl of eight small cookies and asked to eat at least one and
answer several questions about the cookies. Participants in-
dicated attitude toward the cookies on three 7-point scales
(bad/good, dislike/like, negative/positive; ap.89) and, in
keeping with the cover story, they wrote descriptions of the
cookies and indicated appropriate eating occasions. After the
taste test, participants completed a final questionnaire that
included the four-item measure of health goal commitment
used in the pretest (ap.85; Gould 1990), demographics,
and a funneled suspicion check. The number of cookies each
participant ate was recorded after all participants were thanked
and had left.
Results
Food Consumption. Responses from six participants
who indicated a connection between the priming task and
the taste test in the suspicion measure were removed (the
patterns of results are the same with and without these).
Participants in the overweight prime condition ate more
cookies (Mp2.6) than those in the neutral condition (M
p1.9; F(1, 131) p9.63, p!.01). Thus, the effect of the
stereotype prime seen in experiment 1 on food choice was
replicated with actual food consumption. (Neither gender
nor body mass index [BMI]—computed from self-reported
height and weight—had any significant effects, and they
are not discussed further.)
Process. We next examined possible routes for the influ-
ence of the prime on stereotype-conducive behavior. To test
the idea that exposure to a member of the social group tem-
porarily influenced participants’ own goal commitment, we
examined the effect of prime condition on health goal com-
mitment. In the neutral condition, participants rated their
health goal commitment at about the midpoint of the scale
(Mp3.9). As predicted, exposure to the overweight stereo-
type resulted in significantly lower health goal commitment
(Mp3.5; F(1, 129) p8.10, p!.04; differences in degrees
of freedom are due to missing data on scale items). This
provides initial support for the idea that consumers take on
the goal commitment of the stereotype.
Given these significant effects, we examined whether
health commitment mediated the effect of stereotype ex-
posure on food consumption, expecting “complementary
mediation,” with both a mediated (through commitment) and
a direct effect of stereotype activation on eating (Zhao,
Lynch, and Chen 2010). First, we conducted a series of
regression analyses (Baron and Kenny 1986). In particular,
we analyzed the effects of condition and health commitment
on the number of cookies eaten. This analysis revealed a
significant effect of the stereotype condition (Bp.575, p
!.02) and health commitment (Bp⫺.337, p!.007) on
the number of cookies eaten. Importantly, the effect of ste-
reotype condition was significantly reduced from .703 with-
out the inclusion of health goal commitment to .575 with
health goal commitment (Sobel p2.01, p!.044). Second,
we used bootstrapping (Preacher and Hayes 2004), revealing
that the mean indirect effect is positive and significant (M
p.1571), with a 95% confidence interval excluding zero
(.008 to .383). The direct effect is also significant (.575; p
!.014), supporting the combined role of an impact of the
prime on eating and commitment to the health goal. Both
significant mediation analyses indicate that the overweight
prime influences commitment to the health goal and that
low health goal commitment leads to increased consumption
of indulgent food.
An alternative explanation for the stereotype effect is that
exposure to the overweight prime results in more positive
attitudes toward food, which then leads to increased con-
sumption. Thus, we also examined the impact of prime con-
dition on attitude toward the cookies. Prime did not signif-
icantly impact attitude toward the cookies (M
OW
p5.28;
M
Control
p5.37; F!1), and so attitude is not examined
further.
Finally, we analyzed the effect of delay to provide some
insight into whether the goal system is affected by the
stereotype. Interestingly, delay had no effect (F(1, 132) p
1.43, p1.20, NS). The lack of a decline in the impact of
the prime on eating over a delay suggests a motivational
component to the process (Bargh et al. 2001). While this
null effect is not definitive, it is supportive of the proposed
idea that commitment to the health goal is part of the
process by which activation of the overweight stereotype
leads to increased food consumption.
Discussion
The results of experiment 2 support that activation of a
negative stereotype leads to an increase in stereotype-con-
ducive behavior in part by affecting consumers’ commit-
ment to countervailing goals. Activation of the overweight
stereotype leads to lower commitment to the health goal,
consistent with the stereotype. As a result, the consumer
engages in stereotype-conducive behavior despite his or her
countervailing goal to be healthy. This finding helps to an-
swer the question of why consumers may engage in behav-
iors that are causal to membership in a negatively stereo-
typed group despite a countervailing goal. Given the effects
of negative-stereotype activation on stereotype-conducive
behavior, it is important to investigate factors that may limit
these potentially “dysfunctional” behaviors (Dijksterhuis
INCREASED STEREOTYPE-CONDUCIVE BEHAVIOR 437
FIGURE 2
EXPERIMENT 3: THE INFLUENCE OF STEREOTYPE
EXPOSURE AND ACCESSIBILITY OF COUNTERVAILING GOAL
ON COOKIES EATEN
and van Knippenberg 2000). Next, we explore whether in-
creasing accessibility of the countervailing goal buffers a
consumer against the effects of social stereotype activation.
EXPERIMENT 3: COUNTERVAILING
GOAL ACCESSIBILITY
Experiment 3 was designed to examine whether increasing
accessibility of a consumer’s own countervailing goal off-
sets the effect of stereotype activation on stereotype-con-
ducive behavior. Only women were recruited for this ex-
periment because, while the first two studies showed no
effects of gender, they included more men than women;
matching the gender of participants and photos could reduce
additional variance (we thank a reviewer for this suggestion).
Using women is also consistent with the majority of con-
sumer weight research.
Design and Procedure
Women from an undergraduate subject pool (np106)
were randomly assigned to a 2 (prime: healthy weight vs.
overweight) #2 (goal accessibility: low vs. high) #2
(order: goal first vs. prime first) between-subjects design.
Participants, seated in individual cubicles, participated in
three studies; the first two were counterbalanced for order
to manipulate delay between the prime and the behavior
opportunity. Given that self-focus can moderate prime ef-
fects (Dijksterhuis and van Knippenberg 2000; Smeesters
et al. 2009), we manipulated goal accessibility while main-
taining thoughts related to the self. To manipulate goal ac-
cessibility, participants wrote for 3 minutes either about their
health goals (health goal accessibility high, related to the
self) or the geography of their home states (health goal
accessibility low, related to the self; from Wheeler and Ber-
ger 2007). Stereotype prime was manipulated in a picture
perception study in which participants saw two pictures: a
tree and either a healthy weight woman or an overweight
woman. A pretest (np66) verified weight perceptions
regarding two women different from those used in either of
the two previous studies (M
HW
p6.9, M
OW
p12.9; F(1,
64) p161.99, p!.0001). Participants evaluated each pic-
ture for color, originality, quality, and clarity. A taste test
was presented next; this was followed by a suspicion check
and background information questionnaire.
Results
Analyses with a full 2 (stereotype prime) #2 (goal ac-
cessibility) #2 (order) ANOVA showed no significant ef-
fects of order, so we collapsed across order for all other
analyses. Importantly, the lack of an effect of order shows
that the prime effect again persisted over the delay, indi-
cating that exposure to the overweight person activated re-
lated goals and goal commitment. A 2 (stereotype prime)
#2 (goal accessibility) ANOVA of the number of cookies
eaten revealed a significant effect of stereotype prime (F(1,
102) p4.83, p!.03) qualified by a significant interaction
(F(1, 102) p6.47, p!.01). When health goal accessibility
was low, participants ate more cookies after exposure to the
overweight person (Mp3.7) than the healthy weight person
(Mp2.1; F(1, 102) p11.45, p!.001). When, however,
health goal accessibility was high, participants ate the same
number of cookies regardless of the stereotype prime (M
HW
p2.5, M
OW
p2.4; F(1, 102) p.06, NS) and less than
those in the overweight, low health goal accessibility con-
dition (p!.01; see fig. 2). (Body mass index had no effect
and is thus not included.)
Discussion
These results replicate the basic finding of a stereotype
prime effect; participants ate more indulgent food after ex-
posure to an overweight person than to a healthy weight
person when health goal accessibility was not high. In-
creased accessibility of participants’ personal health goals,
however, moderated the effect. When health goal accessi-
bility was heightened, participants ate the same, relatively
low, number of cookies regardless of whether they were
exposed to a healthy weight or an overweight person. These
results reveal an important boundary for the effect of ste-
reotype activation when the primed behavior is inconsistent
with an underlying goal.
The first three studies demonstrate that exposure to some-
one in a negatively stereotyped group can lead to increased
stereotype-conducive behavior. Together, the studies support
the proposed theory that exposure to a stereotype can in-
fluence stereotype-consistent goal commitment, resulting in
low felt commitment, which leads to an increase in stereo-
type-conducive behavior. The mediating role of measured
438 JOURNAL OF CONSUMER RESEARCH
health goal commitment in experiment 2, the moderating
effect of manipulated health goal accessibility in experiment
3, and the lack of a decline in the effect over a delay provide
consistent evidence for goal systems involvement.
EXPERIMENT 4: ACCESSIBILITY OF THE
BEHAVIOR-STEROTYPE LINK
The primary purpose of experiment 4 is to examine whether
increasing the accessibility of the link between the behavior
and group membership can limit the stereotype priming ef-
fect. One reason that the demonstrated increase in behavior
believed to be causally linked to membership in the stereo-
typed group is nonintuitive is that consumers are likely to
want to avoid behaviors related to dissociative groups (Ber-
ger and Heath 2008; White and Dahl 2006). The underlying
desire to avoid these behaviors suggests that increasing the
accessibility of the link between the behavior (eating) and
group membership (being overweight) could limit the effect
of stereotype activation on stereotype-conducive behavior.
Design and Procedure
Following a similar methodology to experiments 2 and
3, 119 female students who were randomly assigned to a 2
(stereotype: healthy weight vs. overweight) #2 (accessi-
bility of behavior-stereotype link: low vs. high) between-
subjects design first participated in a digital picture study
in which they were asked to list the first three colors that
they noticed in the picture. Participants evaluated two pic-
tures: a neutral picture of a tree, followed by a picture of
either a healthy weight woman or an overweight woman.
Participants next took part in a cookie taste test, and they
then completed a filler task, a manipulation check, demo-
graphics, and a suspicion check.
The second photo in the picture study manipulated both
stereotype activation and accessibility of the stereotype-con-
ducive nature of the behavior. The weight (healthy weight
vs. overweight) of the person in the second photo manip-
ulated stereotype activation. Accessibility of the link be-
tween the behavior and stereotype membership was manip-
ulated by whether or not the person in the photo was eating.
Research shows that the body type of others influences the
amount of food that people take. For example, consumers
take and eat more when someone else takes a lot, both when
the other person is thin and overweight (McFerran et al.
2010a). However, nondieters do not eat more, and even eat
less, when served food by an obese server as compared to
a thin server (McFerran et al. 2010b). One reason could be
that being served food by an obese person increases the
accessibility of the causal link between eating and becoming
overweight; the direct connection between the obese server
and food could remind consumers of the link. Research also
showed that people did not mimic an obese person eating
indulgent food, although the eating behavior of a person
with a different stigma was mimicked (Johnston 2002). A
pretest (np63) verified that a picture of an overweight
person eating influenced perceptions of the link between
controlling eating and being overweight; participants who
saw a picture of an overweight woman eating thought that
people who are overweight have lower self-control (Mp
1.9) than did those who saw a picture of the same overweight
woman not eating (Mp2.49; F(1, 62) p5.45, p!.03).
Thus, pictures of the same overweight woman eating or not
eating were used to manipulate accessibility of the behavior-
stereotype link.
Results
The final questionnaire included a manipulation check
that asked participants to write the first thing that came to
mind for “I believe that the primary reason people are over-
weight is ___.” Answers were coded as eating-related (e.g.,
“eat fatty foods”) or not (e.g., “bad genes”) and were an-
alyzed with a 2 #2 logistic regression that showed only
a significant interaction (x
2
(1) p3.6, p!.05). Supporting
the successful manipulation of accessibility of the stereo-
type-conducive nature of the behavior, follow-up analysis
revealed a higher proportion of eating-related reasons after
seeing a picture of an overweight woman eating (75%) than
not eating (48%; x
2
(1) p4.6, p!.03). (There was no
difference in eating-related reasons when the healthy weight
target was eating [61%] or not eating [66%; x
2
(1) !1, NS].)
A 2 (stereotype) #2 (accessibility of behavior-stereotype
link) ANOVA of number of cookies eaten revealed a sig-
nificant effect of the stereotype condition (F(1, 115) p4.89,
p!.03), qualified by a significant interaction (F(1, 115) p
4.07, p!.05). Participants in the overweight, not-eating
condition ate significantly more cookies (Mp2.6) than
those in the overweight, eating condition (Mp1.9, F(1,
115) p5.04, p!.03), the healthy weight, not eating con-
dition (Mp1.8, F(1, 115) p8.17, p!.005), and the
healthy weight, eating condition (Mp1.9, F(1, 115) p
5.5, p!.02; see fig. 3). (Body mass index did not have any
effects.)
Discussion
These results provide a fourth replication, with different
photos, of the basic facilitative effect of overweight stereo-
type activation on the stereotype-conducive behavior of eat-
ing. In addition, these results show that increased accessi-
bility of the link between the behavior and stereotype group
membership serves as a boundary condition for the effect.
The interactive effect of the weight of the person and
whether or not the person was eating suggests that merely
making the idea of eating accessible does not limit the prime-
to-behavior effect. That is, rather than a main effect of eating
versus not, the results demonstrate that participants’ eating
is primed by seeing someone overweight but that the ste-
reotype effect is limited by seeing someone overweight eat-
ing. This supports that making the causal link between eating
and being overweight accessible limits the effect of the
prime on stereotype-conducive behavior.
Typical of stereotype prime research, the studies reported
thus far follow a procedure in which participants are some-
INCREASED STEREOTYPE-CONDUCIVE BEHAVIOR 439
FIGURE 3
EXPERIMENT 4: THE INFLUENCE OF STEREOTYPE
EXPOSURE AND ACCESSIBILITY OF BEHAVIOR-
STEREOTYPE LINK ON COOKIES EATEN
what distracted from the stereotype. In some published stud-
ies, the stereotype is primed subliminally, in others themeth-
odology obscures the prime (e.g., the scrambled sentence
task), and in others, like ours, instructions divert attention
from the stereotype to another aspect of the task (see
Wheeler and Petty 2001). This gives rise to an alternative
explanation for the moderated effects seen in experiment 4.
It is possible that a lack of attention on the stereotype is
necessary for behavioral prime effects to arise. That is, it
may be necessary for attention to be distracted from the
person’s weight; if a consumer considers stereotype mem-
bership (e.g., “that person is overweight”), the stereotype
effect on conducive behavior may be attenuated. Thus, in
experiment 4, it is possible that seeing someone overweight
eating brought attention to the person’s weight and that it
was this attention, rather than the accessibility of the ste-
reotype-conducive nature of the behavior, that eliminated
the stereotype prime effect (we thank a reviewer for noting
this). Thus, the primary purpose of experiment 5 is to tease
apart these two explanations for the attenuating effect of
eating seen in experiment 4.
There is also an alternative explanation for how the effect
of the overweight prime on the stereotype-conducive be-
havior of eating arises. While we propose that the stereotype
activates consistent traits, goals, and goal commitment,
which lead to increases in stereotype-conducive behavior,
another way in which the same effects could arise is by
affecting weight self-esteem. That is, social comparison to
an overweight person could make the participant feel better
about herself (Smeesters and Mandel 2006; Smeesters et al.
2010). If a consumer compares herself to someone more
overweight, she may feel that it is okay to eat more cookies
than when she compares herself to someone with a healthy
weight, or she does not engage in social comparison. The
fact that we do not get different effects for gender or level
of BMI in the earlier studies suggests that this alternative
explanation is unlikely, since women, versus men, would
be more likely to compare themselves to the pictured
women, and participants’ own BMI would be likely to affect
the outcome of a comparison process. However, it would
increase our understanding to examine this alternative more
directly; thus, such examination is an additional purpose of
experiment 5.
Finally, experiment 4 showed a moderating effect of the
accessibility of the behavior-stereotype link by manipulating
whether the people in the prime pictures were eating or not.
It is important to support that it is accessibility of the link,
rather than the behavior of eating, driving the moderation.
Thus, a third purpose of experiment 5 is to replicate the
findings of experiment 4 with a different method of increas-
ing the accessibility of the behavior-stereotype causal link.
EXPERIMENT 5: FOCUS OF ATTENTION
OR ACCESSIBILITY OF THE BEHAVIOR-
STEREOTYPE LINK?
Design and Procedure
Participants were 172 English-speaking undergraduates, who
were randomly assigned to the cells of a 2 (prime: overweight
vs. healthy weight) #2 (focus of attention: distracted vs. fo-
cused) #2 (accessibility of behavior-stereotype link: low vs.
high) between-subjects design. Students, both male and female
for broader generalizability, received course credit for partici-
pation. Ages ranged from 18 to 26; 82 were female, 81 were
male; and nine did not indicate gender. Participants completed
a digital picture study with the same prime procedure as used
in experiment 3; each saw a neutral picture (a tree) and a picture
of either a healthy weight woman or an overweight woman.
Focus of attention was manipulated by the study instruc-
tions. As in experiment 4, the “distracted focus” participants
were instructed to focus on the use of color and were asked:
“What are the first three colors that you notice in this
picture?” In the focused condition, participants were in-
structed to focus on the subject of each picture and were
asked: “What are the first three things you notice about the
person/object in this picture?” After viewing the picture and
writing the first three colors (things about the person), par-
ticipants were asked to rate either the quality of the photo
(distracted) or the weight of the person in the photo (fo-
cused). The instructions either focused the attention of the
participant on something different from the stereotype mem-
bership of the person (by focusing attention on the colors
in the photo) or focused attention on the stereotype mem-
bership (by focusing attention on the weight of the person).
If the alternative explanation is correct, focusing attention
on the weight of the person will attenuate the effect of the
overweight prime on the number of cookies eaten.
Next, the accessibility of the behavior-stereotype causal
link was manipulated. Participants in the low accessibility
condition had no further questions about the person. Those
in the high accessibility condition indicated how much they
thought that the person in the photo ate in a typical day (1
440 JOURNAL OF CONSUMER RESEARCH
pless than average, 7 pmore than average). Estimating
the amount of food an overweight person ate should increase
the accessibility of the link between eating and being over-
weight, attenuating the increase in consumption of cookies
from seeing the overweight prime. Participants next com-
pleted a cookie taste test and then a third study, including
the Twenty Statements Task (TST), a manipulation check,
and demographics. The TST, for which participants com-
pleted up to 20 self-descriptive statements by filling in “I
am _____” (Kuhn and McPartland 1954), was included as
a free-response measure of momentary health and weight
self-esteem (Smeesters and Mandel 2006).
Results and Discussion
We first examined models that included gender, whether
or not the participant was currently on a diet, and BMI.
None of these interacted with the independent variables, and
thus we collapsed across them for further analyses. We next
analyzed the accessibility manipulation check (participants’
answers to complete the sentence “I believe that the primary
reason people are overweight is _____”) coded as eating-
related or not. Analysis revealed a higher proportion of eat-
ing-related reasons after answering the accessibility question
about daily food consumption (45.6%) than not answering
this (30.1%; x
2
(1) p4.12, p!.043), supporting successful
manipulation of accessibility of the stereotype-conducive
nature of the behavior. It is possible that asking participants
to focus on the subject of the picture could also manipulate
the accessibility of the causal link between the stereotype-
conducive behavior and group membership. However, ex-
amination of the manipulation check revealed no difference
in eating versus other reasons for overweight in the focus
(35.7%) and no-focus (39.7%) conditions (x
2
(1) p.28, p
1.59). Examination of perceptions of how much the stim-
ulus person ate within the high accessibility condition with
a 2 (stereotype prime) #2 (focus of attention) ANOVA
found only a significant effect of stereotype prime (F(1, 93)
p143.96, p!.0001); the overweight person was perceived
to eat significantly more (Mp5.4) than the normal weight
person (Mp3.6).
We next verified that in the focus condition participants
did notice the weight of the stimulus person with a 2 (ste-
reotype prime) #2 (accessibility of behavior-stereotype
link) ANOVA. The only significant effect was that of ste-
reotype prime (F(1, 90) p334.36, p!.0001). As expected,
the healthy weight subject was perceived to weigh less (M
p4.58 on a 12-point scale) than the overweight subject (M
p8.41). Thus, the data suggest that the accessibility con-
ditions successfully manipulated the accessibility of the link
between the behavior and the stereotype as expected, while
the focus instructions did not have an impact on accessi-
bility. Importantly, participants in the focus condition did
notice the weight of the people in the photos.
The number of cookies eaten was analyzed with a 2 (ste-
reotype prime) #2 (attention focus) #2 (behavior-ste-
reotype accessibility) between-participant ANOVA. The al-
ternative explanation that the stereotype prime effects are
induced by passive processing and a lack of focus on the
weight of the person was tested by examining effects of the
focus manipulation, either alone or in interaction with the
other independent variables. No main effect or two-way or
three-way interaction was significant (all F’s !1). Overall,
these results provide no support for the alternative expla-
nation; whether participants were instructed to focus on the
picture and use of color or the subject of the picture and
her weight did not significantly affect the number of cookies
participants ate in the taste test. Even when instructed to
focus on the person and rate her weight, participants ate as
many cookies as when they focused on the colors of the
photo.
The only reliable effects in the 2 #2#2 ANOVA of
the number of cookies eaten was a main effect of the ac-
cessibility of the behavior-stereotype link (F(1, 162) p7.8,
p!.006), qualified by a significant interaction between the
prime and accessibility (F(1, 162) p4.93, p!.028). As
expected, the data provide a replication of the basic prime
effect. When nothing was done to heighten the accessibility
of the link between eating and being overweight, the over-
weight prime resulted in consumption of significantly more
cookies (Mp3.04) compared to the healthy weight prime
(Mp2.35; F(1, 162) p3.93, p!.049). However, when
accessibility of the behavior-stereotype link was heightened,
the prime had no effect on the number of cookies consumed
(Mp1.79, Mp2.21; F!1). Importantly, the impact of
exposure to the overweight prime on the number of cookies
eaten was significantly attenuated when the accessibility of
the behavior-stereotype link was increased (Mp3.04 vs.
Mp1.79; F(1, 162) p12.11, p!.001).
We next examined answers to the Twenty Statements Task
to gain insight to the alternative self-comparison process
explanation. If comparison to someone overweight leads to
a positive perception of relative health and weight, we would
expect to see an effect of the prime on the results of the
TST. Following Smeesters and Mandel (2006), two inde-
pendent coders who were blind to conditions scored re-
sponses to the TST. First, a coder identified each statement
that was about the person’s health and weight (e.g., “I am
athletic,” “I am overweight”). A different coder was then
given the identified statements and rated each participant’s
overall health and weight self-perception on a scale of 1–
5. These scores were analyzed with a 2 (prime) #2 (focus
of attention) #2 (accessibility of the behavior-stereotype
link) ANOVA (the sample is smaller because some partic-
ipants did not use any related self-descriptors in the TST).
The analysis revealed no significant effects, including no
effect of the prime #accessibility interaction (F(1, 138) !
1). Thus, comparison effects cannot explain the effects seen
in our data. To examine how this fits with existing research,
we conducted another analysis of the TST including gender
as a factor. While the prime and prime #accessibility in-
teraction remained insignificant (F’s !1), there was a sig-
nificant effect of prime #gender (F(1, 123) p10.86, p!
.005); women had higher health and weight self-esteem after
seeing the overweight woman (Mp3.70) than the healthy
INCREASED STEREOTYPE-CONDUCIVE BEHAVIOR 441
weight woman (Mp3.31; p!.056), while men had higher
health and weight self-esteem after seeing the healthy weight
(Mp3.84) than the overweight woman (Mp3.31; p!
.007). This effect fits with some earlier research on women
(see Smeesters et al. 2010) and suggests an interesting av-
enue for research on men, but it cannot explain the eating
effects in our data.
In addition to providing another replication of the facil-
itating impact of overweight others on consumption of low-
nutrient, energy-dense food, this experiment replicates the
experiment 4 finding that the prime effect of an overweight
other is limited when the behavior-stereotype link is acces-
sible using a different manipulation of accessibility. The
findings further suggest that distraction from the stereotype
is not essential to the effect of the prime and that the effect
of exposure to an overweight person is not driven by
changes in self-esteem. Thus, this research provides support
for the proposed process relative to possible alternative ex-
planations.
GENERAL DISCUSSION
The studies reported in this research provide a variety of
important findings. In five experiments, using four different
examples of stereotyped group members (four different
overweight women) with two different types of controls,
(1) neutral and (2) people not from the stereotyped group
(healthy weight women), and both male and female partic-
ipants, we demonstrate that an active, negative social ste-
reotype can lead to increases in stereotype-conducive be-
havior. We examine the process underlying this effect,
supporting that stereotype activation includes inferred goal
commitment that influences the current commitment of the
consumer to a continuing goal and that goal commitment
partially mediates the effect of the stereotype prime on ste-
reotype-conducive behavior (experiment 2). Experiments 3
and 4 examine two important boundary conditions: height-
ened accessibility of the (1) countervailing goal and (2) link
between the stereotype-conducive behavior and group mem-
bership can limit the stereotype prime effect on stereotype-
conducive behavior. Experiment 5 provides further evidence
for the proposed framework relative to either a process based
on the attention paid to the stereotype or a comparison and
self-esteem process.
Implications for Research on Stereotype-Activation
Effects
The present research provides four contributions to the
stereotype literature. First, five experiments demonstrate in-
creases in behaviors that are believed to be causal to mem-
bership in a negatively stereotyped group. While previous
work has shown increases in behaviors associated with a
group (Wheeler and Petty 2001), this appears to be the first
research that shows increases in behaviors that are conducive
to group membership. The fact that a negative stereotype
prime can increase stereotype-conducive behaviors despite
consumers’ countervailing goals contributes to understand-
ing of the powerful influence of stereotypes.
Second, building on recent research (Aarts et al. 2004;
Van Boven et al. 2010), this suggests that an activated ste-
reotype can have an impact on a consumer’s current com-
mitment to an ongoing goal. While research demonstrates
that a person can take on the stereotypical goals of a group
(Aarts et al. 2004), our work extends these findings, showing
that the level of commitment to a particular goal can also
be affected by a stereotype prime. This is the first demon-
stration of which we are aware that stereotype activation
can influence multiple goals and a person’s own felt goal
commitment. However, while the evidence supports the pro-
posed activation of stereotypical goal commitment, it is pos-
sible that the stereotype activation of the goal to pursue the
stereotype-conducive behavior (i.e., the goal to eat) sup-
presses the countervailing (i.e., health) goal (Shah, Fried-
man, and Kruglanski 2002). While we think that a direct
activation of stereotypical goal commitment is a more par-
simonious explanation, future research comparing activation
of goal commitment to a goal activation and shielding pro-
cess will be useful.
Recent research finds that self-focus can limit the effects
of stereotype activation on behavior and emphasizes the
need for research on other limiting factors (Dijksterhuis
and van Knippenberg 2000; Smeesters et al. 2009). A third
contribution of our research is the identification of two
additional moderating factors. Experiment 3 showed that
increasing the accessibility of a countervailing goal sig-
nificantly attenuates the effect of a stereotype prime. In-
terestingly, this experiment showed that thinking about one’s
home did not attenuate the effect of an overweight prime,
while focusing on one’s health goals did, suggesting that
exploration of the aspects of self-focus that limit stereotype
prime effects could be fruitful. The results of experiments
4 and 5 demonstrate that increased accessibility of the link
between a behavior and membership in the negatively ste-
reotyped group attenuates the effect of the stereotype prime,
demonstrating the importance of expanding consideration of
priming effects to nondeterministic groups and stereotype-
conducive behaviors. A question that our research does not
address is the extent to which the effect of goals is limited
to goals that directly compete with the potential outcome of
stereotype-conducive behavior. This is an intriguing ques-
tion for future research.
A fourth contribution of this research to the stereotype
literature stems from the examination of the role of focus
of attention in stereotype prime effects. Experiment 5 di-
rectly manipulates focus, showing that the effect of a ste-
reotype prime occurs even when participants consider, and
even rate, the person in terms of stereotype membership.
This research adds to the influential research on stereotype
prime effects by demonstrating that such effects can arise
with conscious recognition of the group membership of a
person in a stereotyped group.
442 JOURNAL OF CONSUMER RESEARCH
Implications for Consumers and Consumption
In addition to contributing to the theoretical literature
on stereotype prime effects, the specific domain of over-
weight and eating is of high practical importance. The
number of overweight consumers is rapidly increasing
worldwide. Over 67% of U.S. adult consumers are now
overweight or obese, which is associated with a variety of
adverse health consequences, including Type II diabetes,
heart disease, and multiple cancers (CDC 2007). While it
is clear that people are gaining weight because of an im-
balance between energy consumed versus expended, rea-
sons for the precipitous increase in this imbalance are un-
clear. Recent research emphasizes a need to explore social
and environmental factors that influence weight gain. In-
triguing research suggests that there are person-to-person,
social network influences—analysis demonstrated that a
person’s chance of becoming obese significantly increased
when a socially close other (e.g., friend, sibling, or spouse)
became obese, even if the other lived at a distance (Chris-
takis and Fowler 2007). While this research provided evi-
dence of social influences on weight gain, the nature of
the data limited examination of how this might occur. The
findings of our research are consistent with the spread of
overweight through social networks and, in fact, the be-
havioral and goal commitment assimilation from seeing
people who are overweight could be one way in which the
social network effects arise. People see, both in person and
in photos, the people with whom they have close social
ties. When close others are overweight, our research sug-
gests that stereotype activation could lead to increased food
consumption relative to when close others are healthy
weight since merely seeing someone overweight can in-
crease eating.
At first glance, the increase in food choice and con-
sumption upon exposure to someone overweight demon-
strated across our studies may appear inconsistent with
recent research. McFerran et al. (2010b) showed that non-
dieters (vs. dieters) ate fewer snacks when a server was
obese versus thin, while we did not find effects of BMI,
current dieting behavior, or gender in our studies.However,
as noted earlier, the fact that the overweight other was
serving food may have increased the accessibility of the
idea that eating leads to becoming overweight. Addition-
ally, a critical difference between our work and that of
McFerran et al. (2010b) is that their participants saw some-
one thin or obese while ours saw someone who was per-
ceived to be overweight (not obese) or of healthy weight
(not thin). As indicated by recent research, people who are
obese and overweight may be perceived to be different
(Smeesters et al. 2010). It is possible that seeing an obese
person, compared to an overweight person, increases ac-
cessibility of the link between eating and weight gain, which,
as shown in experiments 4 and 5, could limit the extent to
which consumers eat in the presence of an obese person
versus an overweight person.
We explored this possibility by conducting a study with
three manipulations of weight prime: healthy weight, over-
weight, or obese. A pretest using a 19-point scale identified
photos of women perceived to be obese (Mp15.62), over-
weight (Mp14), and of healthy weight (Mp6.68); dif-
ferences were significant (obese vs. overweight: F(1, 34) p
9.62, p!.004; overweight vs. healthy weight: F(1, 32) p
280.84, p!.0001). Participants, 73 people walking through
a lobby at the University of Colorado (average age p23,
range from 17 to 47; 58% male), were randomly assigned
to one of the prime conditions. After completing a short
questionnaire that included evaluating one of the three pic-
tures, participants were offered candy as a thank you. The
ANOVA of the number of candies taken revealed a signif-
icant prime effect (F(2, 70) p3.45; p!.04). The number
of candies was regressed on prime condition and the prime
#prime interaction; significant effects, in opposite direc-
tions, of the prime (t(70) p2.57, p!.01) and the interaction
(t(70 ) p⫺2.62) indicate a curvilinear relationship. Indi-
viduals who saw the overweight person took significantly
more candies (Mp1.4) than those who saw the healthy
weight person (Mp.83; t(70) p2.35, p!.02). Importantly,
individuals who saw the obese person took significantly less
candy compared to the overweight condition (Mp.86;
t(70) p2.17, p!.03) and the same amount as in the healthy
weight condition (t(70) p.17, p1.80).
These findings help resolve the apparent discrepancy be-
tween our findings and those of McFerran and colleagues by
demonstrating that extremity of weight may lead to differ-
ential effects on eating behavior. Thesealso provide additional
evidence that the process is not one of social comparison
since comparison to someone obese should lead to even better
feelings about the self and own-weight than comparison to
someone who is overweight. These results suggest directions
for future research on stereotype priming, as well as on weight
of others and consumption. Research examining the differ-
ences in the content activated by an overweight, versus obese,
stereotype may offer insights into the process by which the
weight of a person affects a consumer’s own food choices.
Smeesters et al. (2010) look at a range of body weights and
show different effects on females’ self-esteem depending
upon the extremity of the weight of models in ads and the
weight of the female herself. It is clear that there is still much
to be understood about how the weight of someone else in a
person’s social environment—friends, servers, models, char-
acters in TV shows, and so forth—is likely to have an impact
on eating and other health-related choices. While some re-
searchers who are interested in decreasing overweight bias
call for increased inclusion of overweight people in the pop-
ular media (e.g., Greenberg et al. 2002; Wykes and Gunter
2005), the current research suggests the possibility of unin-
tended consequences. Further research into the effects of
greater representation of overweight people on a variety of
variables including bias, goals, and eating is warranted, with
particular consideration being given to avoiding the ironic
effect of inadvertently causing consumers to eat more indul-
gent food.
INCREASED STEREOTYPE-CONDUCIVE BEHAVIOR 443
Implications for Future Research
A limitation of this research is that all of its experiments
examine a single stereotype-conducive behavior and one
negatively stereotyped group. It would be useful to examine
additional stereotype-conducive behaviors and stereotypes.
The behavior examined here is consumption of indulgent
foods; it would be useful to see whether the overweight
stereotype is such that eating all types of foods, healthy as
well as indulgent, would show the same increase or whether
this would lead to a decrease in active behavior and exercise.
Likewise, future research could examine other types of ste-
reotype-conducive behaviors, such as exercising, or different
negatively stereotyped groups. For example, does exposure
to a smoker result in decreased commitment to health goals
and thus a concomitant increase in the potential to smoke?
Or does exposure to the materialistic people often seen in
TV shows affect a person’s own extrinsic (vs. intrinsic)
motivations? An important direction for future research is
to examine how the strong effects seen in this research gen-
eralize beyond the overweight stereotype.
Another direction for future research is to examine the
impact of stereotype exposure on different consumer groups,
particularly vulnerable groups such as children or the elderly.
There is currently little stereotype prime research with chil-
dren. Examining the extent of these effects with children at
different developmental stages would add to our overall un-
derstanding of stereotype priming. The alarming increase in
overweight and obesity of young children, with obesity in
children doubling over the last 30 years (CDC 2007), makes
it particularly important to gain further understanding of fac-
tors that influence the eating choices that children make.
There is a growing literature on the important effects of
subtle environmental factors on consumers’ behavior (e.g.,
Meyers-Levy and Zhu 2007) and environmental factors that
exert unconscious influence on eating (see Wansink 2006).
In order for people to correct for the influence of such fac-
tors, they need to at least understand that there is potential
influence (Bargh and Chartrand 1999). Our study on con-
sumers’ naive theories about the influence of exposure to
overweight others described in the introduction suggests that
consumers are unaware of the possibility that others’ over-
weight could increase their own eating and thus are likely
to be vulnerable to these effects because they do not see a
need for vigilance. By identifying the need to be vigilant
and to consciously remember personal health goals when
considering eating choices, especially in the presence of
overweight others, we hope that our results can be helpful
to consumers, public policy makers, and researchers alike.
REFERENCES
Aarts, Henk, Tanya L. Chartrand, Ruud Custers, Unna Danner,
Giel Dik, Valerie E. Jefferis, and Clara Michelle Cheng
(2005), “Social Stereotypes and Automatic Goal Pursuit,”
Social Cognition, 23 (6), 465–90.
Aarts, Henk, Peter M. Gollwitzer, and Ran R. Hassin (2004),“Goal
Contagion: Perceiving Is for Pursuing,” Journal of Person-
ality and Social Psychology,87(1),23–37.
Bacon, Jane G., Karen E. Scheltema, and Beatrice E. Robinson
(2001), “Fat Phobia Scale Revisited: The Short Form,”
International Journal of Obesity, 25 (2), 252–57.
Bargh, John A. and Tanya L. Chartrand (1999), “The Unbear-
able Automaticity of Being,” American Psychologist,54
(7), 462–79.
Bargh, John A., Peter M. Gollwitzer, Annette Lee-Chai, Kimberly
Barndollar, and Roman Trotschel (2001), “The Automated
Will: Nonconscious Activation and Pursuit of Behavioral
Goals,” Journal of Personality and Social Psychology,81(6),
1014–27.
Baron, Rueben M., and David A. Kenny (1986), “The Moderator-
Mediator Variable Distinction in Social Psychological Research:
Conceptual, Strategic, and Statistical Considerations,” Journal of
Personality and Social Psychology, 51 (6), 1173–82.
Berger, Jonah and Chip Heath (2008), “Who Drives Divergence?
Identity-Signaling, Outgroup Dissimiliarity, and the Aban-
donment of Cultural Tastes,” Journal of Personality and
Social Psychology, 95 (3), 593–607.
Blair, Irene V. and Mahzarin R. Banaji (1996), “Automatic and
Controlled Processes in Stereotype Priming,” Journal of Per-
sonality and Social Psychology, 70 (6), 1142–63.
Centers for Disease Control and Prevention (CDC) (2007), “Over-
weight and Obesity,” http://0-www.cdc.gov.mill1.sjlibrary.org/
nccdphp/dnpa/obesity/index.htm.
Chartrand, Tanya L. and John A. Bargh (1999), “The Chameleon
Effect: The Perception-Behavior Link and Social Interaction,”
Journal of Personality and Social Psychology,76(6),
893–910.
Christakis, Nicholas A. and James H. Fowler (2007), “The Spread
of Obesity in a Large Social Network over 32 Years,” New
England Journal of Medicine, 357 (4), 370–79.
Crandall, Christian S. (1994), “Prejudice against Fat People: Ide-
ology and Self-Interest,” Journal of Personality and Social
Psychology, 66 (5), 882–94.
Dijksterhuis, Ap and John A. Bargh (2001), “The Perception-
Behavior Expressway: Automatic Effects of Social Perception
on Social Behavior,” in Advances in Experimental SocialPsy-
chology, Vol. 33, ed. Mark P. Zanna, San Diego, CA: Aca-
demic Press, 1–40.
Dijksterhuis, Ap and Ad van Knippenberg (1998), “The Relation
between Perception and Behavior, or How to Win a Game of
Trivial Pursuit,” Journal of Personality and Social Psychol-
ogy, 74 (April), 865–77.
——— (2000), “Behavioral Indecision: Effects of Self-Focus on
Automatic Behavior,” Social Cognition,18(1),55–74.
Fishbach, Ayelet and Ravi Dhar (2005), “Goals as Excuses or
Guides: The Liberating Effect of Perceived Goal Progress on
Choice,” Journal of Consumer Research, 32 (3), 370–77.
Fishbach, Ayelet, Ronald S. Friedman, and Arie W. Kruglanski
(2003), “Leading Us Not into Temptation: Momentary Al-
lurements Elicit Overriding Goal Activation,” Journal of Per-
sonality and Social Psychology, 84 (2), 296–309.
Gould, Steven J. (1990), “Health Consciousness and Health Be-
havior: The Application of a New Health Consciousness
Scale,” American Journal of Preventive Medicine, 6 (4),
228–37.
Greenberg, Bradley S., Dana Mastro, and Jeffrey E. Brand (2002),
“Minorities and the Mass Media: Television into the 21st
Century,” in Media Effects: Advances in Theory and Re-
444 JOURNAL OF CONSUMER RESEARCH
search, 2nd ed., LEA’s Communication Series, Mahwah, NJ:
Erlbaum, 333–51.
Hilton, James L. and William von Hippel (1996), “Stereotypes,”
Annual Review of Psychology, 47, 237–71.
International Food Information Council (2009), “2009 Food and
Health Survey: Consumer Attitudes toward Food, Nutrition,
and Health,” http://www.foodinsight.org/.
Johnston, Lucy (2002), “Behavioral Mimicry and Stigmatization,”
Social Cognition,20(1),18–35.
Kawakami, Kerry, Kenneth L. Dion, and John F. Dovidio (1998),
“Racial Prejudice and Stereotype Activation,” Personality and
Social Psychology Bulletin, 24 (4), 407–16.
Kruglanski, Arie W., James Y. Shah, Ayelet Fishbach, Ron Fried-
man, Woo Young Chun, and David Sleeth-Keppler (2002),
“A Theory of Goal Systems,” in Advances in Experimental
Social Psychology, Vol. 34, ed. Mark P. Zanna, San Diego,
CA: Academic Press, 331–37.
Kuhn, Manford H. and Thomas S. McPartland (1954), “An Em-
pirical Investigation of Self-Attitudes,” American Sociologi-
cal Review,19(1),68–76.
Macrae, C. Neil and Lucy Johnston (1998), “Help, I Need Some-
body: Automatic Action and Inaction,” Social Cognition,6
(4), 400–417.
McFerran, Brent, Darren W. Dahl, Gavan J. Fitzsimons, and Andrea
C. Morales (2010a), “I’ll Have What She Is Having: Effect
of Social Influence and Body Type on the Food Choices of
Others,” Journal of Consumer Research, 36 (6), 915–29.
——— (2010b), “Might an Overweight Waitress Make You Eat
More? How the Body Type of Others Is Sufficient to Alter
Our Food Consumption,” Journal of Consumer Psychology,
20 (2), 146–51.
Myers-Levy, Joan and Rui (Juliet) Zhu (2007), “The Influence of
Ceiling Height: The Effect of Priming on the Type of Pro-
cessing That People Use,” Journal of Consumer Research,
34 (2), 174–86.
Preacher, Kristopher J. and Andrew F. Hayes (2004), “SPSS and
SAS Procedures for Estimating Indirect Effects in Simple
Mediation Models,” Behavior Research Methods, Instruments
and Computers, special issue: Web-Based Archive of Norms,
Stimuli, and Data: Part 2, 36 (4), 717–31.
Roehling, Mark V. (1999), “Weight-Based Discrimination in Em-
ployment: Psychological and Legal Aspects,” Personnel Psy-
chology, 52 (4), 969–1016.
Schwartz, Marlene B., Lenny R. Vartanian, Brian A. Nosek, and
Kelly D. Brownell (2006), “The Influence of One’s OwnBody
Weight on Implicit and Explicit Anti-fat Bias,” Obesity,14
(March), 440–47.
Shah, James Y., Ron Friedman, and Arie W. Kruglanski (2002),
“Forgetting All Else: On the Antecedents and Consequences
of Goal Shielding,” Journal of Personality and Social Psy-
chology, 83 (6), 1261–80.
Shah, James Y. and Arie W. Kruglanski (2002), “Priming against
Your Will: How Accessible Alternatives Affect Goal Pursuit,”
Journal of Experimental Social Psychology, 38 (4), 368–83.
Shah, James Y., Arie W. Kruglanski, and Ron Friedman (2003),
“Goal Systems Theory: Integrating the Cognitive and Moti-
vational Aspects of Self-Regulation,” in Motivated Social Per-
ception: The Ontario Symposium, ed. S. Spencer, S. Fein, and
M. P. Zanna, NJ: Erlbaum, 247–76.
Smeesters, Dirk and Naomi Mandel (2006), “Positive and Negative
Media Image Effects on the Self,” Journal of Consumer Re-
search, 32 (March), 576–82.
Smeesters, Dirk, Thomas Mussweiler, and Naomi Mandel (2010),
“The Effects of Thin and Heavy Media Images on Overweight
and Underweight Consumers: Social Comparison Processes
and Behavioral Implications,” Journal of Consumer Research,
36 (6), 930–49.
Smeesters, Dirk, Vincent Y. Yzerbyt, Olivier Corneille, and Luk
Warlop (2009), “When Do Primes Prime? The Moderating
Role of the Self-Concept in Individuals’ Susceptibility to
Priming Effects on Social Behavior,”Journal of Experimental
Social Psychology, 45 (1), 211–16.
Spake, Amanda (2004), “Rethinking Weight,” U.S. News and
World Report, 136 (5), 7.
Stangor, Charles and James E. Lange (1994), “Mental Represen-
tations of Social Groups: Advances in Understanding Stereo-
types and Stereotyping,” in Advances in Experimental Social
Psychology, Vol. 26, San Diego, CA: Academic Press.
Van Boven, Leaf, Margaret C. Campbell, and Thomas Gilovich
(2010), “Stigmatizing Materialism: On Stereotypes and Im-
pressions of Materialistic and Experiential Pursuits,” Person-
ality and Social Psychology Bulletin, 36 (4), 551–63.
Wansink, Brian (2006), Mindless Eating: Why We Eat More than
We Think, New York: Bantam-Dell.
Wheeler, Christian S. and Johan Berger (2007), “When the Same
Prime Leads to Different Effects,” Journal of Consumer Re-
search, 34 (October), 357–68.
Wheeler, Christian S. and Richard E. Petty (2001), “The Effects
of Stereotype Activation on Behavior: A Review of Possible
Mechanisms,” Psychological Bulletin, 127 (6), 797–826.
White, Katherine and Darren W. Dahl (2006), “To Be or Not to
Be? The Influence of Dissociative Reference Groups on Con-
sumer Preferences,” Journal of Consumer Psychology,16(4),
404–14.
Wykes, Maggie and Barrie Gunter, (2005), The Media and Body
Image, Thousand Oaks, CA: Sage.
Zhao, Xinshu, John G. Lynch Jr., and Qimei Chen (2010), “Re-
considering Baron and Kenny: Myths and Truths about Me-
diation Analysis,” Journal of Consumer Research,37(Au-
gust), 197–206.
