obesity | VOLUME 19 NUMBER 10 | OCTOBER 2011 1957
nature publishing group articles
Behavior and Psychology
IntroductIon
More than two-thirds of Americans over the age of 20 are now
overweight and over one-third are obese (1). Yet, despite the
fact that the majority of Americans are now overweight, a sub-
stantial body of literature has documented weight bias among
health-care professionals, teachers, potential employers, family
members of obese individuals, and the media (2–7). In fact, the
prevalence of perceived weight discrimination has increased
by 66% since 1995 (8), and is now on par with rates of racial
discrimination, especially toward obese women, who are tar-
geted most frequently (9).
Driving such discrimination and bias are the stereotypes that
depict overweight individuals as sloppy, lazy, unmotivated, and
less competent, and posit that overweight individuals are solely
to blame for their weight status (4). The media is a particularly
powerful source of both weight-based stereotypes and negative
portrayals of obese individuals. In fact, overweight individuals
remain among the last acceptable targets of derogatory humor
in both television and film (2). When compared to thin tel-
evision characters, overweight characters are more likely to be
the targets of ridicule and humor, are commonly seen engag-
ing in stereotypical eating behaviors, and are rarely depicted
in romantic relationships (5,10). Weight bias in the media is
not subtle, and instances of derogatory weight-based humor in
television and film can be both verbal and direct (6). For exam-
ple, a study of 18 popular prime-time television shows revealed
a strong positive correlation between a female character’s
weight and the frequency of derisive comments made toward
her by others (11). Audience laughter was also positively cor-
related with the character’s weight, and laughter (both live and
canned) was highest when negative comments were directed to
the overweight female characters.
Taken together, these findings demonstrate both the prev-
alence of weight bias in the media and its social acceptabil-
ity. According to the latest Nielsen census data, the average
American now watches over 150 h of television in a given
month. The pervasiveness of weight bias coupled with such
high rates of television viewing implies that the public is
exposed to significant weight bias.
Weight stigma is responsible for a range of negative psy-
chological consequences for those who are targeted. Research
demonstrates that individuals who have experienced weight-
based stigmatization have increased risk of depression, low
self-esteem, anxiety, poor body image, suicidality, and disor-
dered eating (2,4,12). Importantly, many of these studies con-
trol for BMI, indicating that it is the stigmatizing experience
The Impact of Weight Stigma on Caloric
Consumption
Natasha A. Schvey1, Rebecca M. Puhl2 and Kelly D. Brownell1,2
The present study assessed the influence of exposure to weight stigma on energy intake in both overweight and
normal-weight adult women. Seventy-three women (mean age: 31.71 ± 12.72 years), both overweight (n = 34) and
normal weight (n = 39), were randomly assigned to view one of two videos depicting either weight stigmatizing
material or neutral material, after which they consumed snacks ad libitum. Pre- and post-video measures included
blood pressure, attitudes toward overweight individuals, and positive and negative affect. Participants’ body weight
was measured, as was the number of kilocalories consumed following video exposure. Overweight women who
watched the stigmatizing video ate more than three times as many kilocalories as overweight women who watched the
neutral video (302.82 vs. 89.00 kcal), and significantly more calories than the normal-weight individuals who watched
either the stigmatizing or the neutral video. A two-by-two analysis of covariance revealed that even after adjusting for
relevant covariates, there was a significant interaction between video type and weight status in that when overweight,
individuals consumed significantly more calories if they were in the stigmatizing condition vs. the neutral condition
(F(1,65) = 4.37, P = 0.04, η 2 = 0.03). These findings suggest that among overweight women, exposure to weight
stigmatizing material may lead to increased caloric consumption. This directly challenges the notion that pressure to
lose weight in the form of weight stigma will have a positive, motivating effect on overweight individuals.
Obesity (2011) 19, 1957–1962. doi:10.1038/oby.2011.204
1Department of Psychology, Yale University, New Haven, Connecticut, USA; 2Rudd Center for Food Policy and Obesity, Yale University, New Haven, Connecticut, USA.
Correspondence: Natasha A. Schvey (natasha.schvey@yale.edu)
Received 13 November 2010; accepted 5 June 2011; published online 14 July 2011. doi:10.1038/oby.2011.204
1958 VOLUME 19 NUMBER 10 | OCTOBER 2011 | www.obesityjournal.org
articles
Behavior and Psychology
itself, rather than body weight, that is contributing to adverse
psychological outcomes.
A number of studies have also demonstrated that weight
stigma is associated with unhealthy eating behaviors, such as
eating in secret, refusing to diet, loss of control during eating,
and binge-eating (12–14) as well as lower motivation to exercise
and avoidance of physical activity (15,16), even after control-
ling for variables such as BMI and body dissatisfaction. Some
research suggests that individuals who internalize weight bias
and negative stereotypes are more likely to engage in binge-
eating and are more likely to report coping with stigma by
refusing to diet and consuming more food (17). Thus, weight
stigma may augment eating pathology and overeating, and at
the same time, attenuate the desire to exercise. Furthermore,
a recent study demonstrated that weight stigmatization was
associated with greater caloric intake, higher program attri-
tion, lower energy expenditure, less exercise, and less weight
loss in a sample of treatment-seeking adults participating in
a behavioral weight loss program (18). Thus, it appears that
weight stigmatization may increase unhealthy behaviors that
contribute to obesity.
Despite increasing evidence linking weight bias to overeat-
ing, to our knowledge, no studies have explored this link using
actual food intake data. Since questionnaire methods are
oftentimes nonconcordant with actual behavior, and under-
reporting of energy intake is common (19), it is beneficial
to examine the effect of weight stigma on actual, rather than
reported intake.
Therefore, the specific aims of this study were to determine
whether exposure to a stigmatizing video (vs. a neutral video)
influenced caloric consumption, mood, and blood pressure
among both overweight and normal-weight women.
We hypothesized that overweight women exposed to the
stigmatizing video would consume the greatest number of kil-
ocalories, have elevated blood pressure and endorse increased
negative affect and fat phobia following the video, when com-
pared to the other groups.
Methods and Procedures
Participants
Overweight and normal-weight women were recruited for the study
from a university and the surrounding community (city population:
124,000). The study was advertised as research exploring “the effects
of video clips on mood, blood pressure, and other health indices.”
Participants were recruited via flyers and online advertisements (e.g.,
craigslist.com) targeting both overweight and normal-weight women.
To meet inclusion criteria, participants were required to be female, at
least 18 years of age and have no medical complications that might
affect energy consumption (e.g., pregnancy). In total, 102 individu-
als contacted the researchers about participating. Participants were
excluded for the following reasons: Male (N = 4), did not attend sched-
uled appointment (N = 12), declined participation upon hearing study
description or because of scheduling conflicts (N = 9), medical reasons
(N = 3), and reported a 24-h fast before study initiation (N = 1). The
final sample consisted of 73 women.
This study focused exclusively on women as previous research has
demonstrated that women may be more vulnerable to weight stigma
than men (3,9) and women tend to report eating in response to negative
affect more frequently than men (20,21).
Procedure
Participants were admitted individually to the laboratory between the
hours of 2:30 and 4:30 pm, and were asked to refrain from eating 3 h
before study initiation. Participants were told this was to ensure accu-
racy of blood pressure measurement; however, the true purpose was to
standardize hunger among participants. Following informed consent,
participants were randomized to either the stigmatizing video or the
neutral video condition, both of which were 10 min long. This created
four study groups: Group 1: Overweight participants who viewed the
stigmatizing video (Overweight/Stigma, n = 17), group 2: Normal-
weight individuals who watched the stigmatizing video (Normal
Weight/Stigma, n = 20), group 3: Overweight individuals who viewed
the neutral video (Overweight/Neutral, n = 17), and group 4: Normal
weight individuals who watched the neutral video (Normal Weight/
Neutral, n = 19). The stigmatizing video consisted of brief clips from
popular television and movies that depict overweight and obese women
and evoke negative weight-based stereotypes (e.g., clumsy, loud, and
lazy). The types of scenes depicted in the video include teasing in the
workplace, pratfalls involving obese individuals, and interpersonal
instances of weight bias, and reflect some of the most common weight-
based stereotypes that have been reported in the literature (2,4,22). The
video had been pretested in a prior study investigating the effects of
stigmatizing material on antifat attitudes in women (21). The control
video consisted of a series of clips depicting neutral scenes (e.g., insur-
ance commercials) and had similarly been pretested.
Following randomization, participants’ blood pressure was meas-
ured and questionnaires assessing demographics, positive and negative
affect, “fat phobia,” depression, susceptibility to hunger, disinhibition, and
restraint were administered. Prior research examining racial stigmatiza-
tion has documented an increase in blood pressure following exposure
to discriminatory material (23). Thus, blood pressure was assessed to
detect possible physiological changes following the exposure to stigma-
tizing stimuli.
Participants then watched the video. Following the video, blood pres-
sure was measured again. Participants were then given a second set of
questionnaires and provided with three bowls of calorie-dense snack
foods, which they were invited to consume freely. The snacks were 300 g
of plain M&Ms, 300 g of Jelly Belly Jellybeans, and 86 g of SunChips. These
foods were chosen because they are all highly palatable and contribute
three diverse tastes. Participants were told that the snacks were provided
because they had come in to the laboratory fasting. Participants were then
left alone in the room to complete the post-video questionnaires and eat
the snacks ad libitum. At the study conclusion, participants’ height and
weight was measured and they were debriefed and compensated. The
snack bowls were then weighed and number of grams consumed was
recorded and converted into kilocalories. This study was approved by the
Yale University Human Subjects Committee and all procedures were in
accordance with its ethical standards.
Measures
A Medical condition phone screen was administered to exclude partici-
pants with conditions that may affect energy intake.
A Demographic Information Questionnaire assessed age, race, and
ethnicity, educational background, and occupation.
Total caloric intake was calculated based on the nutrition information
on the packaging of the three snack foods. A food scale was used to weigh
the snacks (300 g of M&M’s, 300 g of JellyBelly Jellybeans, and 86 g of
SunChips) both before and after the participants were allowed access to
them. Total number of kilocalories consumed was calculated.
The Three-Factor Eating Questionnaire (TFEQ) (24,25) is a 51-item
self-report questionnaire that assesses restraint, disinhibition, and sus-
ceptibility to hunger. The restraint subscale is calculated from 21 items
and assesses both cognitive and behavioral dietary restriction. The dis-
inhibition subscale is derived from 16 items and measures the tendency
to overeat, and the susceptibility to hunger subscale is derived from 14
items. Scores ≥14 represent the clinical range for restraint, scores ≥12
indicate the clinical range for disinhibition, and scores ≥11 signify a
obesity | VOLUME 19 NUMBER 10 | OCTOBER 2011 1959
articles
Behavior and Psychology
clinical range of susceptibility to hunger (24). The TFEQ has been shown
to be a valid and reliable instrument (24) with internal consistency coeffi-
cients ranging from 0.70 to 0.90 (26). The TFEQ was administered before
the video clip. The reliability of the total measure in the present sample
was α = 0.90, and the reliability of the restraint, disinhibition, and hunger
subscales were 0.77, 0.84, and 0.85, respectively.
The Beck Depression Inventory (BDI) (27) consists of 21 items that
assess depressive symptoms (e.g., I feel utterly worthless) on a scale from
0 to 3. Higher scores reflect more severe levels of depression, with a BDI
score of ≥20 indicating moderate to severe depression. The BDI has
demonstrated high internal consistency in both psychiatric, as well as
community samples (mean coefficient = 0.87) (27,28). The BDI has also
shown strong test–retest reliability, and high construct validity (27). The
BDI was administered before the video clip and the reliability of this
measure in the present sample was α = 0.92.
The Fat Phobia Scale (shortened form) was adapted from the original
Fat Phobia Scale (29). The shortened version of the Fat Phobia Scale lists
14 pairs of adjectives that may be used to describe overweight or obese
individuals (e.g., attractive/unattractive, lazy/industrious). Participants
are asked to indicate how well each adjective describes overweight people
on a 5-point likert scale. The Fat Phobia Scale has exhibited strong reli-
ability, as well as concurrent validity with the original version (29). Total
scores range from 1 to 5; a score of 5 represents the greatest amount of
fat phobia. The Fat Phobia Scale was administered to participants both
before and following the video clips. The reliability of this measure in the
present sample was α = 0.89.
The Positive Affect Negative Affect Schedule (PANAS) (30) is a 20-item
scale that measures participants’ positive and negative affect along a
number of dimensions. Participants are asked to indicate to what extent
they feel a certain way (e.g., alert, enthusiastic, distressed, scared) on a
5-point likert scale. Both the positive affect and negative affect sub-
scales range from 10 to 50. The PANAS is highly internally consistent
and has strong convergent and discriminant validity (30). The PANAS
may be used as a trait version or may be used to assess state affect
(when prefaced with “right now” or “in this moment”). The PANAS
was administered to assess affect both before, and after, the video clips,
and thus, the “state” instructions were employed. Cronbach’s α of the
total measure in the present sample was 0.80; α’s = 0.90 and 0.80, for
the positive and negative subscales, respectively.
BMI was calculated from height and weight which were both measured
by the experimenter at the conclusion of the study using a stadiometer
and digital scale.
Blood pressure was assessed using an automatic blood pressure cuff.
statistical analysis
All analyses were conducted using SPSS for Windows, 17.0 (SPSS,
Chicago, IL). All tests were two-tailed; P values ≤0.05 were considered
significant. ANOVA’s were conducted to assess baseline differences
and to evaluate the success of randomization on balancing key vari-
ables between groups. All significant differences were followed up with
Bonferroni Hochberg post-hoc tests.
To maximize statistical power, planned contrasts were performed to
evaluate a priori hypotheses about group differences following the videos.
These contrasts tested whether overweight participants who watched the
stigmatizing video experienced increased negative affect, fat phobia, and
blood pressure, and decreased positive affect following the video when
compared to the other groups. A 2 × 2 analysis of covariance was used to
test the main effects of weight status and video type, as well as the interac-
tion between weight status and video type on caloric consumption while
controlling for key covariates.
results
sample characteristics
Seventy-four women (mean age 31.71 ± 12.72 years) com-
pleted the study. The sample was divided into both overweight
(BMI ≥25, n = 34) and normal-weight women (18.5 ≤ BMI ≤
24.9, n = 40). The mean BMI of the overweight women was
31.63 ± 6.17, whereas the mean BMI of the normal-weight
women was 21.34 ± 1.89. The demographic distribution was
50% white, 27% African American, 14% Asian American,
8% Hispanic, and 1% Native American. One individual in
the Normal Weight/Neutral condition revealed she had not
eaten anything in the past 24 h due to lack of food security;
therefore, her data were excluded from analyses. Another
individual from the Normal Weight/Neutral group had a
BMI (18.3) that categorized her as underweight. Analyses did
not differ when this individual was removed, thus her data
are included (see Table 1 for baseline descriptive statistics of
relevant variables).
To assess for outliers, relevant variables were converted to
standardized scores. Standardized scores ≥±2.5 were consid-
ered outliers (31). Conversion to standardized scores revealed
that in the caloric consumption variable, there were four uni-
variate outliers (three in the Overweight/Stigma group, and
one in the Overweight/Neutral group) who consumed far
more calories than the rest of their respective group. These out-
liers were transformed according to the convention outlined
by Tabachnick and Fidell (31). Their raw scores on caloric
consumption were transformed to the next highest score from
their respective group plus one unit to bring them closer to the
distribution. All subsequent analyses of caloric consumption
employ these four outliers’ transformed data.
Baseline differences between groups
Between-group differences in baseline variables were assessed
to determine relevant variables to control for in subsequent
analyses. Significant differences between the four groups were
found in BMI, age, depression, disinhibition, systolic blood
pressure, diastolic blood pressure, and pulse (see Table 1).
None of the other variables were found to be significantly dif-
ferent among the groups.
Bonferroni post-hoc tests revealed that participants in the
Overweight/Neutral group reported greater baseline depres-
sion than the Normal Weight/Neutral group (P = 0.036),
and higher baseline blood pressure than both normal weight
groups (P < 0.01). The Overweight/Stigma group had sig-
nificantly higher baseline pulse than the Normal Weight/
Neutral group (P = 0.01). Post-hoc tests did not reveal any
other significant baseline differences between groups (see
Table 1).
Although the groups were randomly assigned and strati-
fied by BMI (two groups were overweight (BMI ≥25 kg/m2)
and two groups were normal weight (BMI <25 kg/m2)), the
Overweight/Stigma group had a higher mean BMI than the
Overweight/Neutral group (P < 0.01) (see Table 1). The two
normal weight groups did not differ in mean BMI.
correlation analyses
Correlations be