Professional Documents
Culture Documents
The Effects of The Media On Body Image A Meta
The Effects of The Media On Body Image A Meta
To cite this article: Amanda J. Holmstrom (2004) The Effects of the Media on Body Image: A Meta-Analysis, Journal of
Broadcasting & Electronic Media, 48:2, 196-217, DOI: 10.1207/s15506878jobem4802_3
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
The Effects of the Media on Body Image:
A Meta-Analysis
Amanda J. Holmstrom
The media have been criticized for depicting the thin woman as ideal.
Some argue these images create unrealistic expectations for young women
and cause body dissatisfaction and disordered eating. This study cumu-
Downloaded by [Newcastle University] at 02:30 23 April 2014
Amanda /. Holmstrom (M.A., Purdue University) is a doctoral sfudent in the Department of Communication
at Purdue University. Her research interests include the relationship between media and body image.
The author wishes to extend her appreciation to Dr. john Sherry for his assistance on this project.
0 2004 Broadcast Education Association journal of Broadcasting & Electronic Media 48(2), 2004, pp. 796-217
196
Holmstrom/MEDIA AND BODY IMAGE META-ANALYSIS 197
Some researchers claim that media have an effect on young women’s body image
(e.g., Baker, Sivyer, & Towell, 1998; Botta, 1999; Harrison & Cantor, 1997; Hof-
schire & Greenberg, 2002). Specifically, these researchers report that results from
their studies indicate a relationship between media consumption and eating disor-
ders, body dissatisfaction, or a number of other related outcomes. However, other
researchers have come to contradictory conclusions. In fact, results of studies by
Crouch and Degelman (1998) and Myers and Biocca (1992) have indicated that
rather than causing an increase in body dissatisfaction, media promote positive body
attitudes in young women. Other researchers claim that no relationship exists
between media and body image (e.g., Borzekowski, Robinson, & Killen, 2000;
Cusumano & Thompson, 1997; Jane, Hunter, & Lozzi, 1999; Stice, 1998).
Downloaded by [Newcastle University] at 02:30 23 April 2014
At this point, a meta-analysis is a useful step toward sorting out these conflicting
results. Meta-analytic reviews are more systematic, explicit, and exhaustive than
qualitative reviews (Rosenthal, 1991). The number of variables included in this
meta-analysis will allow for the examination of trends among studies including the
’ use of different media types, media exposure lengths, outcome measures, compar-
ison stimuli, and participant ages. Meta-analysis at this stage of research on the
relationship between media and body image can offer a greater understanding of the
consequences of the prevalence of thin images in the media, as well as provide
direction for future studies.
Literature Review
Theory
Researchers have used various theoretical underpinnings for studying the relation-
ship between media and body image. Here, I review the theories that have been used
by researchers in the area. The most common of these theories i s Festinger’s social
comparison theory (Botta, 1999, 2000; Dunkley, Wertheim, & Paxton, 2001; Gro-
gan, Williams, & Conner, 1996; Heinberg & Thompson, 1995; Irving, 1990; Kalod-
ner, 1997; Martin & Gentry, 1997; Martin & Kennedy, 1993; Myers & Biocca, 1992;
Ogden & Mundray, 1996; Posavac, Posavac, & Posavac, 1998; Richins, 1991; Shaw,
1995; Turner, Hamilton, Jacobs, Angood, & Dwyer, 1997; Wegener, Hartmann, &
Geist, 2000; Wilcox & Laird, 2000). Festinger (1954) argues that people evaluate
themselves through comparison with others and are more likely to compare them-
selves to those who are similar to them and who are attractive. This comparison is
supposed to motivate one to improve if she finds herself lacking. However, research-
ers who use social comparison theory as a backdrop for media and body image
studies suggest such comparisons could have potential negative effects. The average
model portrayed in the media is approximately 5’1 1” and 120 pounds. By contrast,
the average American woman is 5’4“ and 140 pounds (Wolf, 1991). Research has
shown that a discrepancy between the “actual” self (attributes you and others believe
198 Journal of Broadcasting & Electronic MedidJune 2004
you possess) and the "ideal" self (attributes you or others believe you should possess)
can produce negative emotional states such as sadness, discouragement, and de-
pression (Hatter, 1998). An average woman may be discouraged by the discrepancy
between her body and that of the media ideal. Therefore, if social comparison is the
mechanism at play, researchers would expect that a discrepancy between a woman's
ideal body shape and her actual body shape leads to negative feelings, including
body dissatisfaction.
Other researchers have both implicitly and explicitly explained a connection
between media exposure and body image using Gerbner's cultivation theory as a
framework (Borzekowski et al., 2000; Botta, 1999; Cusumano & Thompson, 1997;
Henderson-King & Henderson-King, 1997; Henderson-King, Henderson-King, &
Downloaded by [Newcastle University] at 02:30 23 April 2014
Hoffman, 2001; Jane et ai., 1999; Tiggeman & Pickering, 1996; Vartanian, Giant, &
Passino, 2001). Cultivation theory posits that the more television a person watches,
the more that person will believe television life is "real life" (Gerbner, Gross,
Morgan, & Signorielli, 1994). Those who believe cultivation theory offers an expla-
nation of the relationship between media and body image posit that thin images in
the media lead people to believe the thin form is both realistic and ideal. If people
do adopt the thin media ideal, researchers would expect a connection in survey
research between length of media exposure and endorsement of the most prevalent
body type portrayed by the media, the thin woman (Silverstein, Perdue, Peterson, &
Kelly, 1986).
A third theoretical backdrop for researchers studying the relationship between
media and body image is Bandura's social cognitive theory (Dunkley et al., 2001;
Harrison, 1997, 2000; Harrison & Cantor, 1997; Hofschire & Greenberg, 2002;
Waller, Hamilton, & Shaw, 1992). Social cognitive theory assumes that people learn
and model the behaviors of attractive others (Bandura, 1994). Proponents of social
cognitive theory posit that young women find thin models in the media attractive and
try to imitate them through dieting and, eventually, the development of eating
disorders. If young women do attempt to imitate the figures they see on television
and in magazines, they would exhibit a greater degree of eating pathology.
Methodological Issues
Thus far, researchers have made some contribution to the understanding of the
relationship between media and eating disorders. However, methodological incon-
sistencies persist, inhibiting our understanding of this relationship.
Construct clarity. A common inconsistency is the use of the term "body image."
There has been no consensus on its definition. Therefore, it has been measured in a
variety of ways. Some researchers have measured the degree to which subjects are
dissatisfied with their bodies in response to media exposure (e.g., Hofschire &
Greenberg, 2002; Irving, 1990). Other researchers have examined the degree to
which subjects overestimate their body size, using the discrepancy between per-
ceived and actual body size as an indication of body dissatisfaction ( e g , Myers &
Holmstrom/MEDIA AND BODY IMAGE META-ANALYSIS 199
Biocca, 1992; Vartanian et al., 2001). Others measure the degree of subjects‘
disordered eating using diagnostic scales for bulimia and anorexia (e.g., Dunkley et
al., 2001; Harrison, 2000; Harrison & Cantor, 1997; Jane et al., 1999). Still others
measure different constructs such as ”importance of appearance” (Borzekowski et
al., 2000) and “endorsement of the thin ideal” (e.g., Botta, 1999, 2000; Stice,
Schupak-Neuberg, Shaw, & Stein, 1994).
Inconsistencies in construct explication and measurement could mean that statis-
tically significant results among these studies suggest drastically different conse-
quences of media exposure, ranging from adopting a thin ideal to developing an
eating disorder. The major concern that has been voiced, however, is whether or not
media cause eating disorders. Therefore, it i s important to understand the relation of
Downloaded by [Newcastle University] at 02:30 23 April 2014
greater control over extraneous variables to study the effect of specifically defined
variable(s) on other variable(s) (Singleton & Straits, 1999), and to make more valid
claims about causality of events. A consistent difference in effect size between the
surveys and experiments will provide more specific information about the nature of
the relationship of media to body image. A high effect for surveys might indicate that
overall media use is related to body image, but does not provide information about
causality. That is, a high effect size for survey studies could mean that media have a
negative impact on body image. However, it also could be that for some reason,
those with poor body image or eating disorders view more media. The controlled
nature of experimental designs would provide more information about a high effect
for experiments. The direction of causality would be clear and a high effect could be
Downloaded by [Newcastle University] at 02:30 23 April 2014
interpreted with more certainty that media do have a negative impact on body
image.
Exposure length. In addition, the amount of time subjects are exposed to media
varies greatly between experimental studies. Viewing time i s as short as 45 seconds
(Crouch & Degelman, 1998) and as long as 26 minutes (Myers & Biocca, 1992). With
this variation in the amount of time subjects are exposed to media, treatment strength
varies widely. Because each study examines only a narrow portion of media for
varying lengths of time, combining the studies in this meta-analysis will increase
power by allowing for the examination of a variety of media measures and exposure
lengths.
Age. The age of subjects participating in these studies varies as well. Researchers
have studied elementary school students ( e g , Harrison, 2000; Martin & Kennedy,
1993), high school students (e.g., Botta, 1999; Botta, 2000; Borzekowski et al., 2000;
Shaw, 1995), and college students (e.g., Irving, 1990; Richins, 1991; Wegener et ai.,
2000). Examining the correlation between age and effect size has theoretical
implications. Cultivation theory would support the hypothesis that older subjects
would be more greatly affected by mediated images of thin women. Older women
would have been exposed to these images for a longer period of time and presum-
ably have had more time to accept them as realistic. Researchers utilizing social
comparison theory and social cognitive theory posit that the women who most
directly identify with the models would be most likely to be affected by the images.
Therefore, since most models are young women, young women are most likely to be
affected.
Comparison stimuli. Another inconsistency in methodology is the type of com-
parison stimuli to which control groups are exposed in experimental studies. These
conditions include a variety of stimuli, such as photos of normal-sized women or
heavy women; television programs with normal-sized or heavy women; and photos
of homes and gardens. Social comparison theory assumes that people will compare
themselves with others they consider similar. If social comparison theory is the
mechanism at work, the meta-analysis may shed light on the effects of subjects’
comparison with mediated images of thin women. It may also provide information
about the varying effects of subjects who compare themselves with photos provided
Holrnstrorn/MEDIA AND BODY IMAGE META-ANALYSIS 201
Meta-Analytic Objectives
A meta-analysis of the existing studies of the effects of media on body image will
provide important information and direction for future research. This article will
Downloaded by [Newcastle University] at 02:30 23 April 2014
address seven objectives: (a) provide a cumulative measure of effect size across all
studies, (b) test for consistent differences in effect size due to type of media measure,
(c) test for differences in effect size due to outcome measure, (d) test for differences
due to study type, (e) test for differences due to comparison stimulus, (0 demonstrate
any relationship between effect size and age of subjects, and (g) demonstrate any
relationship between effect size and treatment length.
Method
This section describes three phases of the meta-analysis: (a) the method of
selecting and coding studies for inclusion, (b) the method of deriving effect size
estimates from the studies, and (c) the procedure for statistical analysis of the effect
sizes.
Literature search. A search of literature from January 1920 to January 2002 was
undertaken using online data bases (ComAbstracts, Psyclnfo). Keywords for both
indexes generally included "body" or "eating" and "media," "television," or "mag-
azines," as well as variations of these terms such as "body image," "eat," or "N."
indexes were searched as far back as they reached in order to obtain all available
research on media and body image; the earliest usable study was published in 1990.
The searches resulted in excess of 3,300 citations, many of which were redundant
across indexes. Titles and abstracts were examined to locate literature reviews that
addressed the effects of media on body image. Literature reviews fitting this descrip-
tion were obtained and the reference sections were examined in order to locate
additional studies. These additional studies were added to the collection, and their
reference sections were scanned for other studies. The process continued until no
new studies emerged. In addition, six authors of dissertations which fit the selection
criteria were contacted for their data via e-mail, but none responded.
The narrowed search resulted in 54 independent studies in which some measure
202 Journalof Broadcasting & Electronic MedidJune2004
of media exposure was the independent variable, and some measure of participants'
weight-based body image, valenced feelings about thinness, or eating pathology was
the dependent variable. Of these, three studies were excluded because they were
content analyses (Levine et al., 1994; Mondini, Favaro, & Santonastaso, 1996;
Thompson & Heinberg, 1999). One experiment was excluded because participants
in all experimental conditions saw images of thin women and there was no control
group from which to calculate an effect size Clung, Lennon, & Rudd, 2001). Seven
studies were excluded because they contained no direct measure of visual media
exposure (Brewis, 1999; Cash, Cash, & Butters, 1983; Dittmar & Blayney, 1996;
Dittmar, Lloyd, Dugan, Halliwell, Jacobs, & Cramer, 2000; Lavin & Cash, 2001;
Ogden & Elder, 1998; Stice, 1998). Three studies were discarded because insuffi-
Downloaded by [Newcastle University] at 02:30 23 April 2014
cient data was available to calculate an effect size: For example, some studies did not
include reports of standard deviations (Champion & Furnham, 1999; Edwards-Hewitt
& Gray, 1993; Pinhas, Toner, Ali, Garfinkel, & Stuckless, 1999). Studies were also
excluded because they did not measure the relationship between exposure to thin
images in the media and the subjects' own body image, which is the definition of
"body image" for this study. For example, two studies were discarded because
subjects were asked to measure the attractiveness of a model rather than evaluating
their own attractiveness (Kenrick & Gutierres, 1996; King, Touyz, & Charles, 2000).
Another study was excluded because though it did include media exposure as an
independent variable and eating pathology as a dependent variable, the researchers
contrasted the eating pathology of two groups whose lifetime media exposure was
drastically different (Baker et al., 1998). Some participants were sighted, while others
had been blind from birth and had never seen thin images. It was ultimately decided
that this study was too much of a departure from the other studies. Two studies
were excluded because the outcome measure, body self-consciousness, measured
trait awareness of one's body (items measuring body self-consciousness include: "I
get up about the same time every day" and "1 eat about the same time every day").
To be included in this meta-analysis, the outcome measure must be some type of
evaluation of one's body, primarily based on weight, and these studies did not
include such measures (Kalodner, 1997; Wegener et al., 2000). One study was
excluded because the outcome measure was amount of food eaten in an experi-
mental study, which also did not directly measure feelings about one's body or
eating pathology (Seddon & Berry, 1996). A total of 34 studies remained for which
effect size estimates could be calculated (see Table 1 for descriptives of the 34
studies).
In some cases, a study contributed multiple effect sizes. For example, one study
might use multiple outcome measures or multiple media measures. Because many
studies included multiple hypotheses with multiple effect sizes, the effect sizes were
analyzed both by study (to avoid problems of non-independence) and when appro-
priate, by sub-analyses of individual hypotheses ( n = 146).
Coding. Study coding sheets included entries for the following characteristics:
study descriptives, sample characteristics, research design, media measure, and
Holmstrom/MEDIA AND BODY IMAGE META-ANALYSIS 203
Table 1
Descriptives of the 34 Included Studies
results. Study descriptives include: title, author(s), date of publication, and type of
report (e.g., journal article, dissertation). Sample characteristics were recorded for
treatment and control conditions across the following categories: sample size, age,
percentage female, number of subjects with anorexia nervosa, and number of
subjects with bulimia nervosa. The type of sampling used to select the participants
was also recorded.
Research design was recorded for all included studies. Thirteen studies were
surveys, accounting for 105 hypotheses. The remaining 2 1 studies were experimen-
tal designs (hypotheses, n = 41). The type of media measure was also recorded for
both surveys and experiments. These media fell into six general categories: maga-
zine, photo (images from magazines), television, generalized mass media, movies,
Downloaded by [Newcastle University] at 02:30 23 April 2014
present time. Some measures in this general category include the Body Esteem Scale
(Franzoi & Shields, 1984) and the body dissatisfaction subscale of the Eating
Disorders Inventory (Garner, et al., 1983). Another type of body dissatisfaction
measure is body size estimation. These measures include body image detection
devices (e.g., Ruff & Barrios, 1986). The subjects are asked to indicate the size they
believe parts of their body to be, including their chests and waists. Subjects’
estimation of body size is compared with the subjects’ actual size, and the discrep-
ancy between the two measures i s an indication of body dissatisfaction. A fourth
category of outcome measure is endorsement of the thin ideal. Stice et al. (1994) and
Stice and Shaw (1994) created this measure by asking subjects in a pilot study to
generate statements about the ideal female body. These statements were then used
Downloaded by [Newcastle University] at 02:30 23 April 2014
to create a measure including items like ”the ideal woman should be slender and
thin” and “the ideal woman should always be dieting.” This measure differs from
importance of appearance measures because it does not measure how important
thinness is to young women but whether or not they believe thinness i s ideal. This
scale was used to examine five hypotheses from three studies.
The type of images shown to the control group was also recorded. Categories of
comparison stimuli included: images of average weight women ( e g , pictures of
women at a local college; studies, n = 10; hypotheses, n = 14), images of
overweight women (e.g., pictures of plus-sized models; studies, n = 3; hypotheses,
n = 6), and comparison with “non-human images” (e.g., pictures of homes or
gardens; studies, n = 10; hypotheses, n = 18).
Effect size estimate. Pearson‘s r was used as the effect size estimate as it is more
flexible and easy to interpret (Rosenthal, 1991). In the survey studies, reported
correlations between media use and the body image outcome measure were used as
the effect size. For experimental studies, an effect size was calculated when a
comparison between treatment and control conditions was reported. When means,
standard deviations, and cell sizes were reported, Cohen’s d (Cohen, 1988) was
calculated and converted to Pearson’s r. Cohen’s d is the difference between
treatment mean (YE)and control mean (Yc) standardized by dividing by an estimate
of the within group standard deviationb) and was calculated using Equation 1:
d=- YE - Yc
S
The within-group estimate of variance for the denominator (5’) was calculated as
directed by Hunter, Schmidt, and Jackson (1982) to lower the standard error of the
estimate (Equation 2):
d
4 ( N - 2)
In the remaining studies that did not provide means, standard deviations, and cell
sizes, a variety of methods were used to extract an effect size estimate including
conversion from t-tests, and conversion from F with 1 degree of freedom in the
numerator (for formulas, see Rosenthal, 1991).
Downloaded by [Newcastle University] at 02:30 23 April 2014
Effect sizes for each hypothesis were calculated then averaged so that each study
contributed only one effect size. This estimation strategy guarantees that each sample
contributes one effect size, and those studies that include only one effect size
estimate carry the same weight as studies which include up to 10 or 20 effect sizes.
The overall, unweighted effect size was calculated for the 34 studies included in the
meta-analysis and i s reported as a Pearson correlation, r = . I 0. Mean weighted effect
size was then calculated for the 34 studies. The weighted effect size i s an average in ,
which each correlation is weighted by the number of participants in that study and
i s the best estimate of cumulative effect size (Hunter et al., 1982). Thus, a study with
300 subjects contributes more to the mean effect size than one with only 40 subjects.
The weighted effect sizes per study are reported in Table 1, and the overall mean
weighted effect size for all studies was r = .08. Next, the amount of variance that is
due to sampling error was calculated and subtracted from the residual variance.
Subtracting the sampling error helps to eliminate variance in the studies due to
imprecise measures (Hunter et al., 1982). Variance remained after sampling error
was subtracted (ap= .02), suggesting that moderating variables are present.
Though it does account for non-independence of effect sizes, averaging effect
sizes by study has some negative consequences. The methods, particularly the
outcome measures, utilized in these studies are very different, and for some sub-
analyses, averaging for one effect size per study does not provide as much detail
about the relationship between media and body image. A mean effect size repre-
senting the Pearson correlation between media use and body image measures was
calculated for each of the 146 hypotheses included in the meta-analysis. The
estimate of the unweighted mean effect size was r = .OO. Next, the estimate of the
weighted mean effect size was calculated by hypothesis, r = .05. Variance remained
after subtracting the effect of sampling error (up= .02).
The somewhat smaller overall effect size when calculated by hypothesis is likely
due to a few studies that included up to 20 very small effect sizes. Once those effect
sizes were cumulated by study, the overall effect size became slightly larger.
Holmstrom/MEDIA AND BODY IMAGE META-ANALYSIS 207
Categorical Analysis-Methodology
participants' overall, global body dissatisfaction (n = 13; items include "I am satisfied
with my body" and "My weight is appropriate for my heighf'). These measures include
Huddy's body dissatisfaction scale (Huddy, Nieman, & Johnson, 1993) and the Body
Attitudes Questionnaire (Ben-Tovim & Walker, 1991). The fourth group of body dissat-
isfaction measures include items measuring participants' satisfaction with individual
body parts (n = 30; items include "I am satisfied with my butt" and "1 am satisfied with
my waist"). These scales include the Body Esteem Scale (Franzoi & Shields, 1984) and
the Body Satisfaction Scale (Slade, Dewey, Newton, Brodie, & Kiemle, 1990). The last
group of body dissatisfaction measures include those called "body dissatisfactionmea-
sure" and do not provide enough information for further interpretation (n = 6).
Effect size estimates for each category vary (body dissatisfaction subscale of the
Downloaded by [Newcastle University] at 02:30 23 April 2014
Eating Disorders Inventory, r = -.02; body size estimation measures, r = .03; global
body dissatisfaction measures, r = .I 7; individual body parts measures, r = .02; and
general, non-specific measures, r = .16). Variation within groups should not be great
since the scales are designed not only to measure the same constructs but also do so
in the same way. However, variance remained for all groups (body dissatisfaction
subscale of the Eating Disorders Inventory, up= .01; body size estimation measures,
up = .03; global body dissatisfaction measures, up = .02; individual body parts
measures, up= .02; and general measures, up = .01).
Another categorical analysis was performed that compared the type of media
measure used. Studies were categorized according to exposure to the medium type.
The different media measure categories are: magazine, photo, television, mass
media, movies, and computer. The mean effect sizes differ by media measure
(magazine, r = .08, n = 39; photo, r = .lo, n = 29; television, r = .03, n = 67; mass
media, r = .04, n = 7; movies, r = .OO; n = 2; and computer, r = .OO, n = 2). There
i s residual unexplained variance remaining in the magazine, photo, television, and
mass media measures (magazine, up= .01; photo, up= .04; television, up = .01;
mass media, up= .01; movies, up = .OO; computer, up= .OO).
Next, a categorical analysis was performed for the type of comparison stimulus
used in experimental studies. Studies were categorized according to the type of
images the control group was exposed to. These include images of average weight
women (e.g., pictures of women at a local college), overweight women (e.g., pictures
of plus-sized models), and "nonhuman images" (e.g., pictures of homes or gardens).
When compared by study, mean effect sizes differed for the comparison stimuli
(average weight women, r = .11, n = 10; overweight women, r = .29, n = 3;
nonhuman images, r = .03, n = 10). There is residual unexplained variance
remaining for the average weight (up= .02), overweight (up= .02), and nonhuman
category stimuli (up= .04).
medium. These variables are ratio level data and can be examined correlationally by
study.
For survey and experimental studies combined, effect size i s negatively correlated
with the length of media exposure ( r = -.07), suggesting that the longer the subject
is exposed to media, the better she feels about her body. However, the correlation
between effect size and length of exposure differs for experiments and surveys. For
experimental studies-in which “length of time” denotes the amount of time partic-
ipants are exposed to images during experimental procedures (in this case, from 45
seconds to 26 minutes)-the relationship between effect size and length of exposure
is similar to the overall correlation between these variables ( r = -.07). For sur-
veys-in which “length of time” refers to the participants’ reports of weekly media
Downloaded by [Newcastle University] at 02:30 23 April 2014
Discussion
Overall Effect
More sophisticated analyses led to important insights about the relationship between
media and body image.
A great deal of variance remains for the experimental studies (up= .04). Most of
these studies are categorized in the "photo" experimental condition, in which
participants are exposed to advertisements from magazines that depict thin models.
Indeed, much variance remains in the photo condition (up= .04) and helps to
explain the variance remaining within experimental studies. The experiments utiliz-
ing a photo measure are among the most similar and most controlled of all included
Downloaded by [Newcastle University] at 02:30 23 April 2014
experimental condition could likely remain unfazed by the depictions since they are
so common.
If viewing images of thin women does not change the participants’ body image,
how can the significant effect size for those studies employing images of overweight
women as the control stimuli be explained? Social comparison theory also posits that
people engage in downward comparisons. By comparing themselves with those less
fortunate, people increase their satisfaction with themselves (Suls & Wheeler, 2000).
Since the participants are young women and therefore less likely to be overweight
than those in other age groups (Mokdad, Bowman, & Ford, 2003), a comparison with
images of heavy women should increase their satisfaction with their bodies. In such
a case, the fairly large effect size for those studies is not due to the women in the “thin
Downloaded by [Newcastle University] at 02:30 23 April 2014
images” condition feeling worse about themselves, but rather to the women in the
“overweight images” condition feeling better.
Though these findings are interesting, future research i s necessary to further
explore and help explain them. In addition, addressing the issue of media saturation
of thin images is important. Because these images are so prevalent, women may not
attend to them, even in an experimental setting. Exposing an experimental group to
either ultra-thin, anorexic women or very overweight women may have an effect on
their body image since these images are much less common in the media. Such
images might induce a downward social comparison, causing the participants to feel
better about their bodies. Including striking, out-of-the-ordinary stimuli may prove to
be valuable tactics for future research.
length of Exposure
The sub-analyses concerning length of media exposure led to results that warrant
speculation. Overall, the relationship between length of exposure and effect size was
very small and negative ( r = -.07). Experimental studies, in which length of
exposure denotes the length of time participants were exposed to stimuli, followed
this same trend ( r = -.07). Participants in experimental studies were exposed to
anywhere from 45 seconds to 26 minutes of images. The correlation indicates that
the longer the participants were exposed to mediated images, the better they felt
about their bodies. The correlation, however, is rather small. The relationship
between length of exposure and body image for survey studies is also negative but
much larger ( r = -.39). Length of exposure for survey studies refers to the amount
of time participants reported viewing media over the course of a week. This strong,
negative correlation indicates that the more time participants reported using media,
the better they felt about their bodies.
Overall, the effect size estimate for the relationship between media use and body
image is very small. However, the correlation between length of exposure and body
image for women who report watching more media suggests that for this subset of
women, there may be a more sizable, positive effect. This correlation i s counter to
the popular belief about the effect of media on body image. Cultivation theory posits
21 2 Journal of Broadcasting & Electronic MedidJune 2004
that the more media we consume, the more we will accept the content of media as
"real life." Therefore, women who consume a great deal of media should accept the
thin ideal as realistic. If women accept the thin ideal and do not measure up to this
ideal (most women do not), it follows that they would be less satisfied with their
bodies, not more satisfied.
Social comparison theory can help to explain why women who report more media
exposure feel better about their bodies. Social comparison theory has been used
often in the media and body image literature to explain why women might feel bad
about themselves when they do not compare favorably with thin models. However,
social comparison theory maintains that in order to enhance their motivation to
improve on a particular dimension (in this case, weight), people may compare
Downloaded by [Newcastle University] at 02:30 23 April 2014
themselves with others who are slightly better on that dimension. Social comparison
may increase satisfaction on that dimension because people can see the results
of an improvement and become motivated. Therefore, some women, particularly
those who report engaging in more media use, may seek out thin images in order
to gain motivation to lose weight and exercise. This inspiration could lead to
increased body satisfaction. For example, a young woman who reports reading a
number of fashion magazines may use the magazines to gather ideas to help her
make fashion and exercise choices, which in turn make her feel good about her
appearance.
In order to determine why women who report greater media consumption feel
better about their bodies, it is important to move from experimental and cross-
sectional survey designs. Focus groups may help us understand how media affects
frequent viewers' body image. Longitudinal designs may offer insight about what
effect media exposure has over time. Perhaps at one time, these viewers were
affected negatively by thin images, but over time that effect diminished. Only
longitudinal studies can answer this question.
endorsement of the thin ideal to r = .05 for body dissatisfaction. They range from r =
.03 to r = .08 by study.
There are several potential reasons for the low average effect sizes garnered by
these studies. The first i s that the media do not have an effect on body image. The
second i s that media do have an effect on body image, but it is not captured by
our measures. In particular, body dissatisfaction measures are problematic.
“Body dissatisfaction” was identified as the outcome measure in nearly all of the
studies included in the meta-analysis ( n = 2 7 ) , but i t i s the only outcome
measure for which variance remained after sampling error was subtracted. This
remaining variance is likely due to the variation within body dissatisfaction
measures themselves. When body dissatisfaction measures were broken down
Downloaded by [Newcastle University] at 02:30 23 April 2014
more specifically into groups, an interesting pattern emerged. Most body dissat-
isfaction measure groups had very low average effect size estimates (body
dissatisfaction subscale of the Eating Disorder Inventory, r = - . 0 2 , body size
estimation measures, r = .03, individual parts measures, r = .02). However,
. larger effect sizes were garnered when the body dissatisfaction measures were
designed to capture general body image ( r = .17) or were unspecified ( r = .16).
These varied results suggest that body dissatisfaction scales are probably not
measuring the same construct. Researchers should consider what type of measure
they are using and what statistically significant results garnered by their studies
mean. For example, body size estimation may not be the same as global body
dissatisfaction. In fact, body size estimation measures differ themselves and may
not be measuring the same construct. The results from this sub-analysis indicate
that media may impact global body satisfaction, and in the future, researchers
should consider using measures designed to capture that effect.
Interestingly, the body dissatisfaction subscale of the Eating Disorders Inventory
produced a very small effect size that indicates women feel more satisfied with their
bodies after media exposure. This scale has been developed to be an indicator of
eating disorder pathology. If researchers and the public are interested in the media’s
impact on eating disorder development, this effect size coupled with the small
overall effect size estimates calculated for eating disorder diagnostic scales ( r = .08
when analyzed by study; r = .05 when analyzed by hypothesis) may indicate that
there i s no effect. The common belief i s that media do have an effect on eating
disorder development, but that belief may be wrong.
Theoretically, type of outcome measure should be important. However, incon-
sistent scales prevent us from seeing this difference. Body dissatisfaction mea-
sures are particularly important if social comparison processes are at work. Social
comparison theory indicates that after comparison with another, we are likely to
experience a change in satisfaction with ourselves on a particular dimension.
Therefore, if social comparison processes do underlie the relationship between
media and body image, we need to have a firm grasp on what “body satisfaction”
truly means.
214 Journal of Broadcasting & Electronic MedidJune 2004
Conclusion
This meta-analysis has shed light on some important trends in the research on the
relationship between media and body image. First, the sub-analyses conducted on
type of comparison stimulus revealed that thin images seem to have little effect on
participants' body image. This result could indicate that women are so accustomed
to seeing thin images that these images no longer lead to changes in body satisfac-
tion. Overweight images, however, tended to substantially increase participants'
body satisfaction. Downward social comparison theory lends support for the hy-
pothesis that engaging in downward comparisons with overweight models has a
positive effect. However, it is important to note that women who report viewing
Downloaded by [Newcastle University] at 02:30 23 April 2014
media for longer periods of time also report greater body satisfaction. These women
may be engaging in social comparison processes that enhance their view of their
bodies. Longitudinal designs and focus groups can offer more information about
these findings.
Additionally, examination of the literature revealed methodological shortcomings
that may have theoretical significance. Specifically, differences in outcome mea-
sures, including body dissatisfaction, could be clouding results. Explicating con-
structs and carefully choosing appropriate measures will aid in interpretation of these
findings.
The findings of the meta-analysis should be interpreted carefully, however. A
relatively small number of studies in the sub-analyses limit the conclusiveness of the
findings. Future research i s needed to further explore the findings suggested by the
meta-analysis. Specifically, a manipulation of exposure length, outcome measures,
and stimulus materials will provide greater understanding of the relationship be-
tween media and body image.
References
American Psychiatric Association Work Group on Eating Disorders. (2000). Practice guideline
for the treatment of patients with eating disorders. American Journal of Psychiatry, 757,
1-39.
Baker, D., Sivyer, R., & Towell, T. (1998). Body image dissatisfaction and eating attitudes in
visually impaired women. hternationalJournal of fating Disorders, 24(3), 319-322.
Bandura, A. (1994). Social cognitive theory of mass communication. In I. Bryant and D. Zillman
(Eds.), Media effects: Advances in theory and research (pp. 61 -90). Hillsdale, NJ:Erlbaum.
Ben-Tovim, D. I., & Walker, M. K. (1991). The development of the Ben-Tovim Walker Body
Attitudes Questionnaire (BAQ), a new measure of women's attitudes towards their own
bodies. Psychological Medicine, 2 7, 775-784.
Borzekowski, D. L., Robinson, T. N., & Killen, J. D. (2000). Does the camera add 10 pounds?
Media use, perceived importance of appearance, and weight concerns among teenage
girls. Journal of Adolescent Health, 26(1), 36-41.
Botta, R. (1999). Television images and adolescent girls' body image disturbance. journal of
Communication, 49(2), 22-41.
Holmstrorn/MEDIA AND BODY IMAGE META-ANALYSIS 215
Botta, R. (2000). The mirror of television: A comparison of Black and White adolescents’ body
image. Journal of Communication, 50(3), 144-159.
Brewis, A. (1999). The accuracy of attractive body-size judgment. Current Anthropology, 40,
548-553.
Cash, T . F., Cash, D. W., & Butters, J. W. (1983). “Mirror, mirror, on the wall. . .?”: Contrast
effects and self-evaluations of physical attractiveness. Personality and Social Psychology
Bulletin, 9, 351-358.
Champion, H., & Furnham, H. (1999). The effect of the media on body satisfaction in
adolescent girls. European Eating Disorders Review, 7, 213-228.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nded.). Hillsdale, NJ:
Erlbaum.
Crouch, A,, & Degelman, D. (1998). Influence of female body images in printed advertising on
self-ratings of physical attractiveness by adolescent girls. Perceptual & Motor Skills, 87,
585-586.
Downloaded by [Newcastle University] at 02:30 23 April 2014
Cusumano, D. L., & Thompson, J. K. (1997). Body image and body shape ideals in magazines:
Exposure, awareness, and internalization. Sex Roles, 37, 701-721.
Dittmar, H., & Blayney, M. (1996). Women’s self-reported eating behaviours and their
responses to food and non-food television advertisements. European Eating Disorders
Review, 4, 217-231.
Dittmar, H., Lloyd, B., Dugan, S., Halliwell, E., Jacobs, N., & Cramer, H. (2000). The “body
beautiful”: English adolescents’ images of ideal bodies. Sex Roles, 42, 887-915.
Dunkley, T. L., Wertheim, E. M., & Paxton, S. 0. (2001). Examination of a model of multiple
sociocultural influences on adolescent girls‘ body dissatisfaction and dietary restraint.
Adolescence, 36, 265-279.
Edwards-Hewitt, T., & Gray, J. J. (1993). The prevalence of disordered eating attitudes and
behaviors in Black-American and White-American college women: Ethnic, regional, class,
and media differences. Eating Disorders Review, 7, 41-54.
Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 117-140.
Franzoi, S. L., & Shields, S. A. (1984). The Body Esteem Scale: Multidimensional structure and
sex differences in a college population. Journal of Personality Assessment, 48, 173-178.
Garner, D. M., Olmstead, M. P., & Polivy, J. (1983). Development and validation of a
multidimensional Eating Disorders Inventory for anorexia nervosa and bulimia. Interna-
tional journal of Eating Disorders, 2, 15-34.
Gerbner, G., Gross, L., Morgan, M., & Signorielli, N. (1994). Growing up with television: The
cultivation perspective. In J. Bryant and D. Zillman (Eds.), Media effects: Advances in
theory and research(pp. 61-90). Hillsdale, NJ: Erlbaum.
Grogan, S., Williams, Z., & Conner, M. (1996). The effects of viewing same-gender photo-
graphic models on body-esteem. Psychology of Women Quarterly, 20, 569-575.
Harrison, K. (1997). Does interpersonal attraction to thin media personalities promote eating
disorders?Journal of Broadcasting & Electronic Media, 4 7, 478-500.
Harrison, K. (2000). The body electric: Thin-ideal media and eating disorders in adolescents.
Journal of Communication, 50(3), 1 19-143.
Harrison, K., & Cantor, J. (1997). The relationship between media consumption and eating
disorders. Journal of Communication, 47(1), 40-67.
Harter, S. (1998). The development of self-representations. In W. Damon and N. Eisenberg
(Eds.), Handbook of child psychology (5rhed.). Vol. 3. Social, emotional, andpersonality
development (pp. 1017-1095). New ‘fork: Wiley.
Heinberg, L. J., & Thompson, J. K. (1995). Body image and televised images of thinness and
attractiveness: A controlled laboratory investigation. Journal of Social and Clinical Psy-
chology, 74, 325-338.
Henderson-King, D., Henderson-King, E., & Hoffman, L. (2001). Media images and women‘s
self-evaluations: Social context and importance of attractiveness as moderators. Personal-
ity and Social Psychology Bulletin, 7 7, 1407-1416.
Henderson-King, E., & Henderson-King, D. (1 997). Media effects on women’s body esteem:
216 Journal of Broadcasting& Electronic MedidJune 2004
Social and individual difference factors. Journal of Applied Social Psychology, 27, 399-
417.
Hofschire, L., & Greenberg, B. (2002). Media‘s impact on adolescents’ body dissatisfaction. In
J. Brown and J. Steele (Eds.), Sexual teens, sexual media: Investigatingmedia’s influence on
adolescentsexuality (pp. 125-149). Mahwah, NJ: Erlbaum.
Huddy, D. C., Nieman, D. C., &Johnson, R. L. (1993). Relationship between body image and
percent body fat among college male varsity athletes and nonathletes. Perceptual and
Motor Skills, 77, 851-857.
Hunter, J. E., Schmidt, F. L., & Jackson, C. 6. (1982). Meta-analysis: Cumulating research
findings across studies. Beverly Hills, CA: Sage.
Irving, L. M. (1990). Mirror images: Effects of the standard of beauty on the self- and
body-esteem of women exhibiting varying levels of bulimic symptoms. journal of Social
and Clinical Psychology, 9, 230-242.
Jane, D. M., Hunter, C. C., & Lozzi, B. M. (1999). Do Cuban American women suffer from
Downloaded by [Newcastle University] at 02:30 23 April 2014
Richins, M. L. (1991). Social comparison and the idealized images of advertising. journal of
Consumer Research, 78, 109-132.
Rosenthal, R. (1 991). Meta-analytic procedures for social research. Newbury Park, CA: Sage.
Ruff, G. A,, & Barrios, B. A. (1986). Realisticassessment of body image. Behavioral Assessment,
8, 237-251.
Seddon, L., & Berry, N. (1996). Media-induced disinhibition of dietary restraint. British journal
of Health Psychology, 7, 27-33.
Shaw, J. (1995). Effects of fashion magazines on body dissatisfaction and eating psychopathol-
ogy in adolescent and adult females. European Eating Disorders Review, 3, 15-23.
Silverstein, B., Perdue, L., Peterson, B., & Kelly, E. (1986). The role of the mass media in
promoting a thin standard of attractiveness for women. Sex Roles, 74,519-532.
Singleton, R. A,, & Straits, B. C. (1999). Approaches to social research. New York: Oxford
University Press.
Slade, P. D., Dewey, M. E., Newton, T., Brodie, D., & Kiemle, G. (1990). Development and
Downloaded by [Newcastle University] at 02:30 23 April 2014
preliminary validation of the Body Satisfaction Scale (BSS). Psychology and Health, 4,
213-226.
Smith, M. C., & Thelen, M. H. (1984). Development and validation for a test of bulimia.journal
of Consulting & Clinical Psychology, 52, 863-872.
Stice, E. (1998). Modeling of eating pathology and social reinforcement of the thin-ideal predict
onset of bulimic symptoms. Behavior Research and Therapy, 36, 931-944.
Stice, E., Schupak-Neuberg, E., Shaw, H E., & Stein, R. I. (1994). Relation of media exposure to
eating disorder symptomatology: An examination of mediating mechanisms. Journal of
Abnormal Psychology, 703, 836-840.
Stice, E.M., & Shaw, H.E. (1994). Adverse effects of the media portrayed thin-ideal on women
and linkages to bulimic symptomatology.journal of Social and Clinical Psychology, 13,
288-308.
Stunkard, A. J.,Sorenson, T., & Schulsinger, F. (1983). Use of the Danish Adoption Register for
the study of obesity and thinness. In S. Kety, L. Rowland, R. Sidman, and S. Matthysse
(Eds.), GeneticsofNeurologicaland Psychiatric Disorders(pp. 1 15-120). New York: Raven
Press.
Suls, J., & Wheeler, L. (2000). A selective history of classic and neo-social comparison theory.
In J. Suls and L. Wheeler (Eds.), Handbook of social comparison: Theory and research (pp.
3-19). New York: Plenum.
Thompson, J. K., & Heinberg, L. J. (1999). The media’s influence on body image disturbance
and eating disorders: We’ve reviled them, now can we rehabilitate them?journal ofsocial
Issues, 55, 339-353.
Tiggemann, M., & Pickering, A. S. (1996). Role of television in adolescent women’s body
dissatisfactionand drive for thinness. lnternationallournalof EatingDisorders, 20, 199-203.
Turner, S . L, Hamilton, H., Jacobs, M, Angood, L .M., & Dwyer, D. H. (1997). The influenceof
fashion magazines on the body image satisfaction of college women: An exploratory
analysis. Adolescence, 32, 603-614.
Vartanian, L. R., Giant, C. L., Passino, R. M. (2001). “Ally McBeal vs. Arnold Schwarzenegger”:
Comparing mass media, interpersonal feedback and gender as predictors of satisfaction
with body thinness and muscularity. Social Behavior & Personality, 29, 711-723.
Waller, C., Hamilton, K., & Shaw, J. (1 992). Media influences on body size estimation in eating
disordered and comparison subjects. British Review of Bulimia & Anorexia Nervosa, 6,
81-87.
Wegener, B. S., Hartmann, A. M, & Geist, C. R. (2000). Effect of exposureto photographsof thin
models on self-consciousness in female college students. Psychological Reports, 86,
1 149-1154.
Wilcox, K., & Laird, J. D. (2000). The impact of media images of super-slender women on
women‘s self-esteem: Identification, social comparison, and self-perception. Journal of
Research in Personality, 34, 278-286.
Wolf, N. (1991). The beauty myth. Toronto: Vintage Books.