Journal of Communication ISSN 0021-9916

RESEARCH ARTICLE

Effects of Sexually Objectifying Media on

Self-Objectification and Body Surveillance
in Undergraduates: Results of a 2-Year
Panel Study
Jennifer Stevens Aubrey
Department of Communication, The University of Missouri at Columbia, Columbia, MO 65211-2310

This study used objectification theory (B. L. Fredrickson & T.-A. Roberts, 1997) to pre-
dict that the medias insidious practice of objectifying bodies socializes individuals to
take an outsiders perspective on the physical self (i.e., self-objectify) and to habitually
monitor their appearance (i.e., engage in body surveillance). To test these hypotheses,
a 2-year panel study using an undergraduate sample was conducted. Cross-lagged path
models showed that exposure to sexually objectifying television measured during Year 1
increased trait self-objectification (trait SO) during Year 2 for both women and men.
At the same time, trait SO during Year 1 decreased exposure to sexually objectifying
television during Year 2, suggesting that both male and female participants selectively
avoided sexually objectifying television based on antecedent trait SO. Moreover, expo-
sure to sexually objectifying television and magazines increased body surveillance for
men only. The discussion focuses on the process by which the media create body-focused
perceptions.

doi:10.1111/j.1460-2466.2006.00024.x

The relationship between the body and sex is unambiguously portrayed in contem-
porary media. Conforming to a thin body ideal is crucial to sexual attractiveness. We
see this message in a variety of ways and in a variety of media, especially in popular
magazines and television programs. For example, womens magazine covers often
place weight loss messages next to messages about ones sex life, implying weight loss
will lead to a better sex life (Malkin, Wornian, & Chrisler, 1999). In one issue, Drop
8 Pounds this Month was on the same cover as 25 Ways to Make Your Marriage
Hot Again, and in another, Stay Skinny was on the same cover as What Men
Want Most (Malkin et al., 1999). Moreover, teen and womens magazines regularly
feature articles on attracting the opposite sex interspersed with advertisements for
beauty care products and fashion merchandise (Carpenter, 1998; Durham, 1998).

Corresponding author: Jennifer Stevens Aubrey; e-mail: aubreyj@missouri.edu.

366 Journal of Communication 56 (2006) 366386 2006 International Communication Association

J. S. Aubrey Sexual Objectification

Finally, in television shows popular with adolescents, the most common sexual theme
is that women are judged as romantic or sexual partners based on their physical
appearance (Ward, 1995).
Perhaps the most insidious way that the media emphasize physical attractiveness
is by objectifying bodies (Fredrickson & Roberts, 1997). Sexual objectification has
been conceptualized as the separating of a persons body, body parts, or sexual
functions from his or her person, reducing them to the status of mere instruments,
or regarding them as if they were capable of representing him or her (Bartky, 1990).
Content analyses have operationally defined sexual objectification as instances in
which the focus is on isolated body parts, such as a bare stomach, buttocks, cleavage,
or a bare chest, in the absence of a focus on the rest of the person (Kolbe & Albanese,
1996; Rudman & Hagiwara, 1992; Sommers-Flanagan, Sommers-Flanagan, & Davis,
1993). A general conclusion from this content-analytic work is that the media often
focus on bodies and appearance as the most important components of sexual desir-
ability. However, there are gender differences in how the media use sexual objecti-
fication. Some research has suggested that the difference in how the bodies of men
and women are portrayed is by the face-to-body proportions. For men, a face-ism
bias exists, whereby mens heads and faces are shown in greater detail than they are
for women (Archer, Iritani, Kimes, & Barrios, 1983). The corresponding bias for
women is body-ism; the focus is usually on womens bodies or body parts, some-
times eliminating their heads altogether (Unger & Crawford, 1996). Still, it should
not be assumed that men are never objectified. For example, one content analysis of
photographs in six male-oriented magazines (e.g., GQ, Rolling Stone) found that
there was a wide range in the frequency of images that focused on male sexual body
parts (Kolbe & Albanese, 1996). For example, in Playboy, 21.6% of the images of
men focused on a bare chest, whereas only 8.2% of images in Esquire focused on a
bare chest.
The effects of exposure to sexual objectification in the media on audiences self-
images are largely unexplored. Although much research has found that thin-ideal
media exposure is linked to body dissatisfaction (see Groesz, Levine, & Murnen,
2002, for a meta-analysis), the thin ideal is not the same as sexual objectification.
After all, a person might not conform to the thin ideal but nevertheless be objectified.
Similarly, some experimental studies have claimed to manipulate sexual objectifica-
tion by varying physical attractiveness, but such a manipulation clearly conflates
physical attractiveness with sexual objectification (Cattarin, Thompson, Thomas,
& Williams, 2000; Lavine, Sweeney, & Wagner, 1999).
To examine the effects of sexually objectifying media exposure on viewers body
perceptions, objectification theory (Fredrickson & Roberts, 1997) was used as a the-
oretical framework. The theorys main contention is that women in particular can be
acculturated to internalize a viewers perspective as a primary view of their physical
selves (p. 173), a perspective called self-objectification. Self-objectifying individuals
come to view themselves as objects or sights to be appreciated by others. Self-
objectification is the tendency to define the self in terms of how the body appears to

Journal of Communication 56 (2006) 366386 2006 International Communication Association 367

Sexual Objectification J. S. Aubrey

others, rather than what the body can do or how the body feels. Closely related to
self-objectification is body surveillance (McKinley & Hyde, 1996). Vigilance about
monitoring ones appearance is necessary to comply with sociocultural body stand-
ards and to avoid negative judgments.
In conceptualizing objectification theory, Fredrickson and Roberts (1997)
focused on the consequences of self-objectification, and as such, most of the empir-
ical research has examined the outcomes of self-objectification, including depleted
task performance, lower self-efficacy, lower intrinsic motivation, negative body emo-
tions, and eating disorder symptomatology (e.g., Fredrickson, Roberts, Noll, Quinn, &
Twenge, 1998; Noll & Fredrickson, 1998; Roberts & Gettman, 2004; Tiggemann &
Lynch, 2001). In contrast, the current study focuses on an antecedent to self-
objectification. Although influences on self-objectification might include a variety
of interpersonal, social, cultural, and even biological factors, an aggressive purveyor
of sexual objectification is undoubtedly the mass media (Fredrickson & Roberts,
1997). Thus, it stands to reason that media exposure that is high in sexual objecti-
fication can socialize individuals to treat their own bodies as objects. Individuals who
continuously see others bodies being objectified in the media learn the importance
of the body, which could encourage individuals to view themselves as objects to be
looked at by others. A theoretical explanation for this link is based on cultivation
theory (Gerbner, Gross, Morgan, & Signorielli, 1994). Whereas cultivation theory
purports that television teaches audiences to adopt certain cultivated views of the
world around them, the current study predicts that sexually objectifying media
content can teach audiences to adopt a certain perspective of the self, one that places
primary importance on physical appearance (Harrison & Fredrickson, 2003).
Although research has not yet investigated the medias role in socializing indi-
viduals to self-objectify in the long term, a recent experiment demonstrated that
sports media stimuli enhanced self-objectification in adolescent girls in the short
term (Harrison & Fredrickson, 2003). Exposure to sports clips featuring lean women
athletes activated a state level of self-objectification for European American girls,
whereas exposure to sports clips featuring nonlean women athletes activated a state
level of self-objectification in girls of color. In addition, another recent study has
implications for the medias influence on self-objectification. Roberts and Gettman
(2004) had participants unscramble sentences with objectification-related words
(e.g., sexiness, weight) or body competencerelated words (e.g., health, fitness).
The college-aged women and men who unscrambled the sentences with objectifica-
tion words reported more state self-objectification than those who unscrambled
sentences with body competence words. If self-objectification can be primed by such
a subtle taskmerely encountering words that highlight the bodys physical appear-
ancethen it is likely that the medias overt emphasis on the body will also activate
self-objectification, perhaps even more strongly than unscrambling sentences.
The predictions based on objectification theory are conceptualized to be appli-
cable to women primarily. A central assumption of the theory is that women exist in
a culture in which their bodies are looked at, evaluated, and always potentially

368 Journal of Communication 56 (2006) 366386 2006 International Communication Association

J. S. Aubrey Sexual Objectification

objectified (Fredrickson & Roberts, 1997, p. 177). Again, however, the theory is
focused on the outcomes of self-objectification. The question of whether the medias
impact on self-objectification only applies to women is still unanswered. Thus, both
men and women were included in this study for three main reasons. First, experi-
mental research has found that it is possible to induce self-objectification in both
men and women. For example, in a study ostensibly on consumer behaviors, both
men and women who tried on a swimsuit reported more self-objectification than
those who tried on a sweater (Fredrickson et al., 1998), and again, unscrambling
sentences with objectification-related words produced more self-objectification for
both men and women than neutral words (Roberts & Gettman, 2004). Second, there
seems to be an increased emphasis on the objectification of men in the media across
time (e.g., Thompson, 2000). Ubiquitous advertisements for mens appearance-
related products, television programs focusing on making over mens appearance
(e.g., Queer Eye for the Straight Guy), and the introduction of the word metro-
sexual (i.e., a slang word for a male who spends a great deal of time and money on
his appearance) into everyday vocabulary could be indicators that male appearance is
increasingly visible and important. Third, most studies that have examined medias
impact on body image have focused on restrictive eating and drive for thinness as
outcomes, which are undoubtedly more relevant to women than to men (McCabe &
Ricciardelli, 2001). Based on this research, men are considered to be exempt from the
negative effects of media on the body, even though research has generally ignored
male-relevant media content and outcomes. After all, manipulating thinness (e.g.,
Kalodner, 1997) is undoubtedly less damaging to men than manipulating muscu-
larity, for instance (McCabe & Ricciardelli, 2001). In short, the research that has thus
far been conducted on men does not rule out the possibility that the media can
socialize men to self-objectify.
The main hypothesis is that exposure to sexually objectifying media (either
magazines or television) will increase self-objectification for both women and
men. Furthermore, it is predicted that the relationship will hold after controlling
for variables that, according to previous research, are also likely to influence self-
objectification, such as a strong sociocultural attitude about appearance (Thompson,
Heinberg, Altabe, & Tantleff-Dunn, 1999), self-esteem (e.g., Befort et al., 2001), body
mass index (BMI; e.g., McLaren & Gauvin, 2002), race (e.g., Gluck & Geliebter,
2002), age (Tiggemann & Lynch, 2001), and parents education (McLaren &
Gauvin, 2002). The opposite direction in the hypothesis was also explored: Will
self-objectification influence whether individuals select sexually objectifying media
in the first place?

Method
Design and procedure
A longitudinal approach to test these hypotheses was fruitful for two main
reasons. First, it permitted an examination of the long-term relationships between

Journal of Communication 56 (2006) 366386 2006 International Communication Association 369

Sexual Objectification J. S. Aubrey

nonmanipulated media habits and self-objectification, as opposed to the short-term

effects of experimenter-selected media stimuli. Second, it permitted an examination
of the order of the variables in the relationship, examining whether media exposure
increases self-objectification or whether self-objectification drives the selection of
sexually objectifying media. For these reasons, all measures were taken at two time
points. The design was a two-wave panel study.

Participants
In total, 226 undergraduate participants from a large, Midwestern U.S. university
took part in both waves of the study. Participants were 65.9% (n = 149) women and
34.1% (n = 77) men. In the first wave, the participants were on average 19.6 years old
(SD = .76), and in the second wave, they were 20.6 years old (SD = .75). In total,
70.2% (n = 158) identified themselves as European American, 16.4% (n = 37) Asian
American, 5.3% (n = 12) African American, and 2.7% (n = 6) Latino. The remaining
5.3% (n = 12) did not identify with any category.
For the first wave, 41.6% (n = 94) of the participants completed the question-
naire to satisfy a requirement for an introductory communication studies course,
and 58.4% (n = 132) of the participants received $7 as compensation for completing
the questionnaire. For the second wave, all participants were recontacted 1 year after
they first completed the questionnaire and received $7 for participating. The ques-
tionnaire was administered via a web interface, which was used because research has
shown that research participants are more likely to report sensitive behaviors, such as
smoking, drinking, and drug use, via computer-assisted questionnaires than via face-
to-face interviews or paper-and-pencil, self-administered questionnaires (e.g.,
Wright, Aquilino, & Supple, 1998).
The retention rate between the waves was 58.9%. Assuming there is always
attrition in panel research, the critical issue is whether the participants who com-
pleted both waves of measurement differed from those who dropped out after the
first wave. In examining the Wave 1 means for the participants who stayed in versus
the ones who dropped out, the only significant difference regarded gender: 65.4% of
women originally recruited for this study came back for Wave 2, whereas 50.3% of
men who were recruited for Wave 1 came back for Wave 2, x2 = 9.98 (p , .01). For
all remaining variables, no significant differences were found. Overall, then, the
difference between participants who came back for Wave 2 and those who dropped
out after Wave 1 was relatively unsystematic.
One year was chosen as the interval between waves of data collection because
there is likely to be a considerable amount of variance in issues related to the body
during a 1-year interval. This assumption was based on research showing that college
in particular and emerging adulthood in general represent developmental periods
when individuals typically grapple with issues related to weight and appearance (e.g.,
Bishop, Bauer, & Becker, 1998).

370 Journal of Communication 56 (2006) 366386 2006 International Communication Association

J. S. Aubrey Sexual Objectification

Measures
Exposure to sexually objectifying television and magazines
A procedure similar to the one described by Harrison (2000) was used for measuring
and calculating exposure to sexually objectifying television and magazines. This
procedure has three steps. First, participants reported their habitual exposure to
popular television shows and magazines. Second, a separate, impartial sample of
judges (not research participants) rated each show and magazine according to
how sexually objectifying they perceived them to be. Third, the mean ratings sup-
plied by the judges were multiplied by participants frequency-of-viewing scores for
each show or magazine, and these cross products were averaged. The resulting
variables reflect both frequency of viewing and extremity of sexual objectification
in the shows and magazines.
Frequency of viewing (0 = never; 5 = always) was measured for 77 popular and
current television shows and 61 popular and current magazines. A separate ques-
tionnaire was given to 13 judges who were enrolled in an upper-level undergraduate
seminar on mass communication. In the seminar, the judges learned about objecti-
fication theory and ways to conceptualize and operationalize sexual objectification,
but they were not briefed on the specific goals of the current study. Reflecting the
gender composition of the seminar, all but one of the judges was female. The gender
breakdown was not ideal because it does not rule out the possibility that women
assess sexual objectification differently from men. However, because the judges were
specifically trained on how to detect sexual objectification from the standpoint of
objectification theory, any differences in their judgments that were inherent to their
gender were hopefully minimized.
The judges rated each show and magazine on an 11-point scale for how often
sexual objectification occurred (0 = never; 10 = all the time). If the judges had never
seen or had difficulty assessing the show or magazine, they marked a box designated
never seen or dont know. Only shows and magazines that had relatively definite
sexual objectification ratings were included in the final measure. To that end, shows
and magazines with widely varying ratings were eliminated, and only shows and
magazines that had five or more judges (out of a possible 13) rating them were
included in the final measure. Given these criteria, only 26 magazines and 26 pro-
grams were used in the final measure.1
The resulting variables were fairly stable across the two waves. For television, the
Time 1 and Time 2 measures were correlated at .51 (p , .001) for women and .60
(p , .001) for men. For magazines, the Time 1 and Time 2 measures were correlated
at .52 (p , .001) for women and .43 (p , .001) for men.

Trait self-objectification
The Trait Self-Objectification Questionnaire (Noll & Fredrickson, 1998) asked
individuals to rank body attributes in order of how important they were to their
physical self-concept. Ten items were included; five were appearance based (physical

Journal of Communication 56 (2006) 366386 2006 International Communication Association 371

Sexual Objectification J. S. Aubrey

attractiveness, weight, sex appeal, measurements, and muscle tone), and the other
five were competence based (muscular strength, physical coordination, stamina,
health, and physical fitness). Scores were computed by summing the ranks for the
appearance and competence attributes separately, and then computing a difference
score. The scores ranged from 125 (highest rating of self-objectification) to 225
(lowest rating of self-objectification). For women, the Time 1 and Time 2 measures
of trait self-objectification (trait SO) were highly correlated at .62 (p , .001), but for
men, the Time 1 and Time 2 measures were only correlated at .22 (p = .08). Thus, it
appears that for women, trait SO was highly stable between waves, but for men, trait
SO might be susceptible to situational factors and not best measured as a trait.

Body surveillance
A symptom of self-objectification is body monitoring (Fredrickson & Roberts, 1997).
Conceptually, body surveillance and trait SO are similar in that they both tap a con-
cern about appearance. However, it makes sense to think of body surveillance as
a cognitive (e.g., thinking and worrying about appearance) and behavioral (e.g.,
primping) outcome of trait SO. For this reason, trait SO and body surveillance were
assumed to measure similar yet nonequivalent constructs. In support of this assump-
tion, trait SO and body surveillance were correlated at .28 (p , .001) in Wave 1, and
at .42 (p , .001) in Wave 2.
Body surveillance was measured with the Surveillance Sub-Scale of the Objecti-
fied Body Consciousness Scale (McKinley & Hyde, 1996). The scale comprised eight
items (e.g., During the day, I think about how I look many times). Participants
indicated their agreement on a 5-point scale (0 = strongly disagree; 4 = strongly agree).
Cronbachs alpha for Wave 1 was .61 for women and .64 for men, and the alpha for
Wave 2 was .83 for women and .81 for men. Again, the stability of the measures was
quite different by gender. This time, however, Time 1 and Time 2 measures of body
surveillance were more highly correlated for men (r = .56, p , .001) than for women
(r = .31, p , .001).

Sociocultural attitudes about appearance

The Internalization Sub-Scale of the Sociocultural Attitudes Toward Appearance
Questionnaire (SATAQ, Heinberg, Thompson, & Stormer, 1995) was used to mea-
sure internalization of sociocultural beliefs of attractiveness. The items assessed the
extent to which participants accept the body ideals portrayed in the media (e.g.,
I tend to compare my body to TV and movie stars). Cronbachs alpha during
Wave 1 was .82 for both men and women, and .85 for women and .76 for men in
Wave 2. The Wave 1 and Wave 2 measures were correlated at .66 (p , .001) for
women and .55 (p , .001) for men.

Global self-esteem
For the Rosenberg Self-Esteem Scale (Rosenberg, 1965), participants were asked to
indicate their level of agreement with nine items (e.g., I feel that I have a number of

372 Journal of Communication 56 (2006) 366386 2006 International Communication Association

J. S. Aubrey Sexual Objectification

good qualities), using a 4-point scale (1 = strongly disagree; 4 = strongly agree). The
Wave 1 Cronbachs alpha for women was .88 and .80 for men; in Wave 2, it was .86
for women and .88 for men. The Wave 1 and Wave 2 correlation was .66 (p , .001)
for women and .55 (p , .001) for men.

Other demographic variables

Several other control variables were measured, including BMI (weight in kg/height
in m2), age (in years), race,2 and parental education (1 = less than high school to
6 = graduate degree completed).

Results
Table 1 reports the summary statistics for the primary predictor and criterion var-
iables. Differences by gender were evident. Women reported more exposure to
sexually objectifying television and more exposure to sexually objectifying magazines
than men. In addition, as expected, women reported more self-objectification and
more body surveillance than men. This gender difference is in support of objectifi-
cation theorys contention that women are more self-conscious of their bodies than
men (Fredrickson & Roberts, 1997).

Correlations of sexually objectifying media exposure with trait SO and

body surveillance
The zero-order correlations between the Time 1 predictor variables and Time 2
criterion variables were examined for women and men separately. The results are
shown in Table 2.
Exposure to sexually objectifying television at Time 1 was significantly correlated
with trait SO at Time 2 for women. Although the strength of the correlation did not
substantively differ between men and women, the correlation did not reach signif-
icance for men, due to the smaller sample size of male participants. Moreover, Time
1 exposure to sexually objectifying magazines was not significantly correlated with
Time 2 trait SO for men or women.
Sexually objectifying media exposure (both television and magazines) was more
strongly correlated with body surveillance for men than for women. For men, the
correlation between Time 1 exposure to sexually objectifying television and Time 2
body surveillance was positive and highly significant, whereas it was only marginally
significant for women. Similarly, Time 1 exposure to sexually objectifying magazines
was correlated with Time 2 body surveillance for men, but not at all correlated with
body surveillance for women.

Path models between exposure to sexually objectifying media and trait SO,
body surveillance
To further investigate the relations between the main variables, the data were
analyzed using cross-lagged path analyses with the Analysis of Moment Structures

Journal of Communication 56 (2006) 366386 2006 International Communication Association 373

Sexual Objectification J. S. Aubrey

program (AMOS). First, the question of gender was addressed. The zero-order
correlations suggested that there was not a substantive gender difference for the
correlations involving trait SO, but there was a considerable gender difference in
the correlations involving body surveillance. Thus, it appears that the model between
exposure to sexually objectifying television and trait SO should be run for all par-
ticipants, whereas the model between exposure to sexually objectifying media (tele-
vision and magazines) and body surveillance should be run separately by gender.
To confirm this reasoning, gender differences were checked by first running an
unconstrained path model in which the cross-lagged parameters between exposure
to sexually objectifying television and trait SO were allowed to differ for male and
female participants (Byrne, 2001). This model was compared to a constrained
model, in which the cross-lagged parameters were constrained to be equivalent for
male and female participants. As expected, the fit of the model that constrained the
cross-lagged parameters versus the constrained model did not significantly differ,
x 2difference 2 5 :96, p = .62, suggesting that the paths of interest did not differ by
gender.3 Thus, in order to best represent the results of the path model investigating
the relations between exposure to sexually objectifying television and trait SO,

Table 1 Descriptive Statistics of Predictor and Criterion Variables, by Gender

Females Males t
M (SD) Range M (SD) Range
Time 1 exposure to 7.94 (3.77) 0 to 21.04 6.43 (3.25) 0 to 14.43 3.04**
sexually objectifying
television
Time 2 exposure to 6.91 (4.16) 0 to 41.13 4.49 (2.62) .29 to 10.86 4.58***
sexually objectifying
television
Time 1 exposure to 5.85 (4.50) 0 to 42.48 3.49 (2.66) 0 to 12.73 4.20***
sexually objectifying
magazines
Time 2 exposure to 5.45 (4.49) 0 to 42.48 3.10 (2.57) 0 to 14.74 4.22***
sexually objectifying
magazines
Time 1 trait self 1.73 (11.63) 225.00 to 23.19 (11.78) 225.00 to 2.72**
objectification 25.00 25.00
Time 2 trait self 1.75 (12.23) 225.00 to 22.47 (12.64) 225.00 to 2.38*
objectification 25.00 25.00
Time 1 body 2.68 (.54) .88 to 3.88 2.37 (.69) .88 to 3.75 3.72***
surveillance
Time 2 body 2.62 (.74) .13 to 4.00 2.12 (.69) .63 to 3.50 4.95***
surveillance
Note: T statistic based on independent sample t test.
* p , .05. ** p , .01. *** p , .001.

374 Journal of Communication 56 (2006) 366386 2006 International Communication Association

J. S. Aubrey Sexual Objectification

Table 2 Correlations Between Predictor and Criterion Variables

1 2 3 4 5 6 7 8
Female Participants
1. Time 1 exposure to
sexually objectifying
television 1.00
2. Time 2 exposure to
sexually objectifying
television .45*** 1.00
3. Time 1 exposure to
sexually objectifying
magazines .57*** .33*** 1.00
4. Time 2 exposure to
sexually objectifying
magazines .26** .74*** .52*** 1.00
5. Time 1 trait
self-objectification .09 2.13 2.01 2.10 1.00
6. Time 2 trait
self-objectification .19* .01 .05 .02 .62*** 1.00
7. Time 1 body surveillance .08 .03 .00 .04 .30*** .21* 1.00
8. Time 2 body surveillance .14 .07 .04 .10 .52*** .38*** .31*** 1.00
Male Participants
1. Time 1 exposure to
sexually objectifying
television 1.00
2. Time 2 exposure to
sexually objectifying
television .60*** 1.00
3. Time 1 exposure to
sexually objectifying
magazines .41*** .21 1.00
4. Time 2 exposure to
sexually objectifying
magazines .35** .43*** .43*** 1.00
5. Time 1 trait
self-objectification .11 2.08 .10 2.14 1.00
6. Time 2 trait
self-objectification .16 .11 .09 .14 .22 1.00
7. Time 1 body surveillance .18 .17 .14 .24* .18 .26* 1.00
8. Time 2 body surveillance .39** .31** .31** .30** .39** .45*** .56*** 1.00
p , .10. * p , .05. ** p , .01. *** p , .001.

Journal of Communication 56 (2006) 366386 2006 International Communication Association 375

Sexual Objectification J. S. Aubrey

a two-group path model was run, in which the cross-lagged paths between the pre-
dictor and criterion were constrained to be equal between men and women, and all
other paths in the model were unconstrained. The results are presented in Figure 1.
For both men and women, the path coefficient between Time 1 exposure to
sexually objectifying television and Time 2 trait SO was positive and significant,
suggesting that exposure to sexually objectifying television in the first wave of the
study predicted an increase in trait SO in the second wave of the study. The corre-
sponding coefficient between trait SO at Time 1 and exposure to sexually objectifying
television at Time 2 was statistically significant and negative, suggesting that trait
SO at Time 1 predicted a decrease in exposure to sexually objectifying television at
Time 2. One plausible explanation for this negative effect is that individuals who
exhibit a high trait level of self-objectification attempt to avoid sexually objectifying
television. One might think of this as selective avoidance of potentially damaging
television content.
In examining the gender differences in the model between exposure to sexually
objectifying television and body surveillance, x2 difference tests comparing the

Female Participants
.88
T1 exp to .46*** T2 exp to
s-ob TV s-ob TV

.11 -.12** (-.05) .13* (.43)

T1 trait .77
T2 trait
SO SO
.60***

Male Participants

.78
T1 exp to .62*** T2 exp to
s-ob TV s-ob TV

.10 .11* (.43)

-.20** (-.05)
T1 trait .97
T2 trait
SO SO
.20+
Chi-square (4, N = 226) = 1.86, p = .76,
RMSE = .00, NFI = .998

Figure 1 Path models predicting direct effects between exposure to sexually objectifying
television and trait self-objectification (two-group model constraining cross-lagged paths
between males and females).
The unstandardized path coefficients are shown in parentheses. Exp to s-ob
TV = exposure to sexually objectifying television; trait SO = trait self-objectification.
p , .10. * p , .05. ** p , .001.

376 Journal of Communication 56 (2006) 366386 2006 International Communication Association

J. S. Aubrey Sexual Objectification

unconstrained model versus the constrained model were again examined. This time,
the results suggested that separate models for males and females were needed,
x2difference 7 5 38:85, p , .001.4 For women, exposure to sexually objectifying tele-
vision at Time 1 was not related to body surveillance at Time 2, nor did Time 1 body
surveillance predict exposure to sexually objectifying television at Time 2. However,
for men, the path between Time 1 exposure to sexually objectifying television and
Time 2 body surveillance was positive and significant, suggesting that early exposure
to sexually objectifying television predicted an increase in later body surveillance.
The models are shown in Figure 2.
The x2 difference test suggested that the path model relating exposure to
sexually objectifying magazines to body surveillance also should be examined
separately for men and women, x2difference 7 5 35:83, p , .001. Because the cor-
relations showed that womens exposure to sexually objectifying magazines was

Female Participants
.89
T1 exp to .45*** T2 exp to
s-ob TV s-ob TV
.11
Chi-square (1, N = 149) = .05, p = .83,
.08 -.01
RMSE = .00, NFI =.999
T1 body T2 body .94
surv. .30*** surv.

Male Participants
.80
T1 exp to .59*** T2 exp to
s-ob TV s-ob TV
.30**

.17 .06
Chi-square (1, N = 77) = .64, p = .42,
.77
T1 body T2 body
surv. .51*** surv. RMSE = .00, NFI =.999

T1 exp to .43*** T2 exp to .87

s-ob s-ob
mags .23* mags
Chi-square (1, N = 77) = .89, p = .35,
.14
.18+
.79 RMSE = .00, NFI = .998
T1 body T2 body
surv. .53*** surv.

Figure 2 Path models predicting direct effects between exposure to sexually objectifying
media and body surveillance.
Path coefficients are standardized estimates. Exp to s-ob TV = exposure to sexually
objectifying television; exp to s-ob mags = exposure to sexually objectifying magazines; body
surv. = body surveillance. p , .10. * p , .05. ** p , .01. *** p , .001.

Journal of Communication 56 (2006) 366386 2006 International Communication Association 377

Sexual Objectification J. S. Aubrey

not related to body surveillance, the path model was run for men only and is
presented in Figure 2. Time 1 exposure to sexually objectifying magazines pre-
dicted an increase in Time 2 body surveillance for men. In addition, the corre-
sponding path coefficient between Time 1 body surveillance and Time 2 exposure
to sexually objectifying magazines was positive and marginally significant. Mens
exposure to sexually objectifying magazines predicted an increase in body sur-
veillance, and body surveillance also predicted an increase in exposure to sexually
objectifying magazines at a marginal level of significance.

Controlling for key third variables

Although the models show some significant longitudinal relations between the pre-
dictor and criterion variables, they leave open the question as to whether the results
might be due to some third variable that is related to both. To test for this
possibility, the structural models were recomputed, controlling for possible other
influences on trait SO and body surveillance.
Each control variable was introduced into the structural equation models shown
in Figures 1 and 2, with paths of influence leading to all variables.
Controlling for two self-related variables (SATAQ and global self-esteem) and
four demographic variables (BMI, parents education, race, and age) did not dimin-
ish the path from early TV viewing to later trait SO for men or women, as the
significant z values in Table 3 indicate. Although some of the paths between the
control variables and the predictor and criterion variables were substantive, they did
not account for the cross-lagged longitudinal relations. For example, the path coef-
ficient from Time 1 SATAQ to Time 2 trait SO was .12 (p = .09) for women and .20
(p = .08) for men. However, these relationships did not substantively reduce the path
from Time 1 exposure to sexually objectifying television and Time 2 trait SO (from
.13 to .12 for women and from .11 to .10 for men). In addition, the correspond-
ing paths between Time 1 trait SO and Time 2 exposure to sexually objectifying
television were not substantively altered by the introduction of the self or demo-
graphic variables.
The procedure of entering in the control variables was replicated for the models
in Figure 2, and the results are also presented in Table 3. All path coefficients pre-
dicting Time 1 exposure to sexually objectifying television to Time 2 body surveil-
lance remained statistically significant after controlling for the self and demographic
variables. Likewise, the introduction of the control variables did not substantively
alter the relationship between Time 1 exposure to sexually objectifying magazines
and Time 2 body surveillance in all cases, with one exception. The exception
occurred when SATAQ was introduced; the strength of the coefficient was reduced
from .23 to .16, a large enough change to reduce the coefficient to marginal sig-
nificance (p = .09). Overall, though, the major implication of these models is that
the longitudinal relations were largely unaltered by the introduction of the con-
trol variables.

378 Journal of Communication 56 (2006) 366386 2006 International Communication Association

 Table 3 Structural Models Cross-Lagged Path Coefficients After Controlling for Self and Demographic Variables
J. S. Aubrey

Time 1 Exposure to Sexually Objectifying Time 1 Trait Self-Objectification !Time 2 Exposure to

Television ! Time 2 Trait Self-Objectification Sexually Objectifying Television
Original Control z Original Control z
SATAQ
Females .13 .12 2.06* 2.12 2.14 22.79**
Males .11 .10 2.06* 2.20 2.21 22.79**
Global SE
Females .13 .13 2.27* 2.12 2.11 22.35*
Males .11 .11 2.27* 2.20 2.18 22.35*
BMI
Females .13 .13 2.10* 2.12 2.14 22.76**
Males .11 .10 2.10* 2.20 2.21 22.76**
Parents education
Females .13 .14 2.32* 2.12 2.12 22.61**
Males .11 .11 2.32* 2.20 2.19 22.61**
Race
Females .13 .10 1.75 2.12 2.12 22.62**
Males .11 .08 1.75 2.20 2.19 22.62**

Journal of Communication 56 (2006) 366386 2006 International Communication Association

Age
Females .13 .14 2.49* 2.12 2.12 22.49*
Males .11 .11 2.34* 2.20 2.19 22.49*

(continued)
Sexual Objectification

379
380
Table 3 Continued

Males Only Time 1 Exposure to Sexually Objectifying Time 1 Body Surveillance ! Time 2 Exposure
Television ! Time 2 Body Surveillance to Sexually Objectifying Television
Sexual Objectification

Original Control z Original Control z

SATAQ .30 .25 2.87** .06 .02 .18
Global SE .30 .30 3.23** .06 .06 .60
BMI .30 .30 3.27** .06 .06 .66
Parents education .30 .30 3.28** .06 .07 .78
Race .30 .28 3.13** .06 .06 .65
Age .30 .31 3.33** .06 .06 .64

Males Only Time 1 Exposure to Sexually Objectifying Time 1 Body Surveillance ! Time 2
Magazines ! Time 2 Body Surveillance Exposure to Sexually Objectifying Magazines
Original Control z Original Control z
SATAQ .23 .16 1.70 .18 .19 1.70
Global SE .23 .25 2.68** .18 .18 1.74
BMI .23 .23 2.40* .18 .18 1.82
Parents education .23 .24 2.54* .18 .19 1.92
Race .23 .24 2.57* .18 .19 1.87
Age .23 .24 2.55* .18 .19 1.83
Note: The z statistic tests the significance of the control coefficients. SATAQ = Sociocultural Attitudes Toward Appearance Questionnaire; Global
SE = Global self-esteem; BMI = body mass index.
p , .10. * p , .05. ** p , .01.

Journal of Communication 56 (2006) 366386 2006 International Communication Association

J. S. Aubrey
J. S. Aubrey Sexual Objectification

Discussion
As expected, exposure to sexually objectifying television during Year 1 predicted an
increase in trait SO during Year 2 for both women and men. Surprisingly, exposure
to sexually objectifying television predicted an increase in body surveillance for men
only. In addition, trait SO predicted an avoidance of sexually objectifying television
for both men and women. Sexually objectifying magazine exposure, on the other
hand, predicted an increase in body surveillance for men only, and, at a level
approaching significance, body surveillance also predicted an increase in exposure
to sexually objectifying magazines for men.
A benefit of a longitudinal approach is that it offers some credibility in inter-
preting temporal order in the relationships. Still, although critical self and demo-
graphic variables were controlled, it is possible these variables did not account for all
alternative hypotheses. Thus, the results of this study can be interpreted as suggestive
of a causal connection, but a firm conclusion on causality awaits further research.
A main finding is that exposure to sexually objectifying television shows was
associated with an increase in viewers definitions of their physical selves in terms of
externally perceivable traits (i.e., how the body appears) rather than internal traits
(i.e., what it can do). One explanation for this process is that exposure to televised
objectification cultivates a particular view of the self, a view that emphasizes the
importance of physical appearance. Another way of thinking about this process is to
think of exposure to sexually objectifying television as activating self-objectification
in the short term, much like what one would expect from a priming process (e.g.,
Bargh, Chen, & Burrows, 1996). Although priming is generally conceived to be
a short-term effect, one must consider the television diet of a typical college student.
An average daily dose of television might include reality television programs, soap
operas, talk shows, music videos, and, of course, lots of advertising. Throughout this
viewing, an individual might be exposed to objectification dozens if not hundreds
of times (Harrison & Fredrickson, 2003, p. 229). Thus, the most important question
in judging the medias influence on self-objectification is not how long a media-
induced state of self-objectification lasts, but how frequently it is induced (Harrison
& Fredrickson, 2003). If exposure to sexual objectification in television continuously
activates self-objectification, then the overall picture of a college students life is
chronic, trait-like self-objectification.
However, the results of this study suggest that exposure to televised objectifi-
cation does not necessarily translate to a sea change in audiences perceptions of the
self. Rather, the results have much in common with the results of other cultivation
studies; the measurable effects are modest but should not be dismissed as theoret-
ically insignificant (Gerbner et al., 1994). After all, it takes but a few degrees shift in
the average temperature to have an ice age or global warming (p. 26). Similarly, it
takes only a bit of self-objectification for a persons view of his or her body to
be fundamentally different from a person who does not chronically self-objectify
(Fredrickson & Roberts, 1997).

Journal of Communication 56 (2006) 366386 2006 International Communication Association 381

Sexual Objectification J. S. Aubrey

A surprising finding in this study was the lack of a gender difference in this
relationship. This is consistent with experimental evidence that has shown that
self-objectification can be primed in both men and women (Fredrickson et al.,
1998; Roberts & Gettman, 2004); thus, this study adds that the long-term influence
of media on self-objectification might also applicable to both men and women.
While Fredrickson and Roberts (1997) argue that objectification theory is about
the lived experience of women, the theory emphasizes the consequences of self-objec-
tification and does not focus on explicating the origins of self-objectification. This
study suggests that there is relative gender equity in the medias ability to cultivate
self-objectification.
In fact, judging that exposure to sexually objectifying television and magazines at
Year 1 increased body surveillance at Year 2 for men but not for women, one might
conclude that the medias ability to increase body surveillance was stronger for men
than for women. One possible explanation for this surprising gender difference is
that for women, body monitoring is normative and thus not as susceptible to influ-
ence by media exposure as it is for men. Indeed, because body monitoring activities
are considered deeply socialized components of femininity (McKinley & Hyde,
1996), it might be more influenced by interpersonal sources, such as friends, family,
and significant others.
Although exposure to sexually objectifying television enhanced trait SO for both
men and women, the stronger effect of media exposure for men was on body
surveillance. In addition, trait SO was a more stable construct for women, whereas
body surveillance was a more stable construct for men. Overall, then, one might
conclude that trait SO was a more relevant construct for women than for men,
whereas body surveillance was a more relevant construct for men than for women.
Further consideration of the conceptual differences between body surveillance and
trait SO helps to better understand these gender differences. First, body surveillance
basically taps a concern over how one appears to others. In contrast, trait SO meas-
ures a deeply internalized definition of the self. A second conceptual difference
between trait SO and body surveillance is that trait SO measures an implicit
trade-off between appearance and body competence, whereas body surveillance
explicitly measures the trade-off between appearance and comfort. Thus, the
trade-off implied by body surveillance is less serious because it does not compromise
the bodys well-being, such as the case when participants choose appearance over
body competence. Based on these differences between trait SO and body surveillance,
it appears that for women, the internalized message is the possibly more serious and
more damaging you are your body. For men, the message appears to be the less
serious need to adhere to a certain appearance standard that will yield positive
evaluations by others.
Some provocative results were found for the impact of trait SO on participants
viewing of sexually objectifying television. Trait SO during Year 1 predicted avoid-
ance to sexually objectifying television during Year 2, at the same time that exposure
to sexually objectifying television during Year 1 increased trait SO during Year 2.

382 Journal of Communication 56 (2006) 366386 2006 International Communication Association

J. S. Aubrey Sexual Objectification

How do these two relationships complement each other? One plausible explanation
is that individuals who exhibit a high level of trait SO attempt to avoid sexually
objectifying television. One might think of this as selective avoidance of potentially
damaging television content. However, sexually objectifying programs on television
are rather prevalent and popular, especially among young audiences (e.g., Collins,
1998); thus, despite the attempt to avoid sexually objectifying television, inevitable
exposure to programs high in sexual objectification might still have the effect of
increasing trait SO.
In addition to the selective avoidance effect, body surveillance during Year 1
increased exposure to sexually objectifying magazines during Year 2 (at a level
approaching significance) for men. This finding is akin to arguing that violent media
causes aggression, but also aggressive individuals seek violent media. These relation-
ships can be regarded as complementary. In this case, men who are concerned about
their appearance are drawn to sexually objectifying magazines.

Limitations and future directions

Participants were recruited on a volunteer basis. Strictly speaking, then, this studys
sample cannot be extrapolated to a larger population. Moreover, the attrition of
women was less than the attrition of men. It is possible that some of the men who
participated in the first wave did not want to participate in the second wave because
by then, they knew what kinds of questions were on the questionnaire and were not
interested in responding to what they might have considered de facto womens
issues. In addition, given that women in general score higher on measures of
altruism than men (Phares, 1984), it is possible that the female participants were
more amenable to helping out the researcher. Although unequal attrition is not
ideal, it should be noted that besides gender, the participants who stayed in the study
for both waves and the dropouts did not differ significantly on any of the measured
variables.
A limitation of the measures of exposure to sexually objectifying media is that the
judges who evaluated the degree of sexual objectification in the television programs
and magazines were almost exclusively women, a consequence of the upper-level
mass communication seminar, in which the judges trained to detect sexual objecti-
fication were almost exclusively women. It is possible that if an almost exclusively
male sample of judges evaluated the content, the results would be substantively
different. Despite this limitation, it is still argued that this trained group of mostly
women was more desirable than a gender-balanced group of novice judges. Sexual
objectification can have multiple conceptual or operational meanings; these differ-
ences in definitions were hopefully minimized by training judges on what to look for.
Another limitation of this measure is that it did not allow inferences about what
specific characteristics of the media enhance self-objectification; this is a topic for
future research. Also, it did not distinguish between the person who consumes a lot
of moderately sexually objectifying media from the person who consumes a little of
highly sexually objectifying media.

Journal of Communication 56 (2006) 366386 2006 International Communication Association 383

Sexual Objectification J. S. Aubrey

A future avenue of research could include a closer examination of how the

media affect self-objectification for men. For example, this study did not delineate
a difference between media that objectified womens bodies and media that objec-
tified mens bodies. One possibility is that objectification of mens bodies is so rare
that men are especially sensitized to these images. For men, there might be a drench
effect, whereby even a small dose of objectification of mens bodies has a relatively
large effect (Greenberg, 1988). In contrast, there might be a cumulative, drip, drip
effect for women because they are constantly inundated with objectified females in
the media.

Notes
1 A complete list of the shows and magazines, with their corresponding ratings, is available
from the author.
2 Race was coded as a dummy variable, one category for European Americans and the other
category for nonEuropean Americans.
3 First, a model in which all parameters were allowed to differ for male and female par-
ticipants was compared to a model in which all parameters were constrained to be equal
for male and female participants (Byrne, 2001). The x2 difference test was significant,
x2difference 7 5 43:52, p , .001, but the results of this test were misleading because the only
difference between the models was the stability coefficient for trait SO. More importantly,
there was no difference in the cross-lagged path coefficients. Thus, in a subsequent test,
only the cross-lagged paths were constrained to be equal for men and women. The fit of
this semi-constrained model versus the unconstrained model did not differ,
x2difference 2 5 :96, p = .62, suggesting that the specific paths of interest to the hypothesis
did not differ by gender.
4 A model in which all parameters were allowed to differ for male and female participants
was compared to a model in which all parameters were constrained to be equal for
male and female participants (Byrne, 2001). The x2 difference test was significant,
x2difference 7 5 38:85, p , .001, and the cross-lagged path coefficients were substantively
different for men and women.

References
Archer, D., Iritani, B., Kimes, D. D., & Barrios, M. (1983). Face-ism: Five studies of sex
differences in facial prominence. Journal of Personality and Social Psychology, 45, 725735.
Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social behavior: Direct effects of
trait construct and stereotype activation on action. Journal of Personality and Social
Psychology, 71, 230244.
Bartky, S. L. (1990). Femininity and domination: Studies in the phenomenology of oppression.
New York: Routledge.
Befort, C., Kurpius, S. E. R., Hull-Blanks, E., Nicpon, M. F., Huser, L., & Sollenberger, S.
(2001). Body image, self-esteem, and weight-related criticism from romantic partners.
Journal of College Student Development, 42, 407419.
Bishop, J. B., Bauer, K. W., & Becker, E. T. (1998). A survey of counseling needs of male and
female college students. Journal of College Student Development, 39, 205210.

384 Journal of Communication 56 (2006) 366386 2006 International Communication Association

J. S. Aubrey Sexual Objectification

Byrne, B. M. (2001). Structural equation modeling with AMOS: Basic concepts, applications,
and programming. Mahwah, NJ: Erlbaum.
Carpenter, L. M. (1998). From girls into women: Scripts for sexuality and romance in
Seventeen magazine, 19741994. Journal of Sex Research, 35, 158168.
Cattarin, J. A., Thompson, J. K., Thomas, C., & Williams, R. (2000). Body image, mood, and
televised images of attractiveness: The role of social comparison. Journal of Social and
Clinical Psychology, 19, 220239.
Collins, J. (1998, October 26). The youth brigade. Time, pp. 8890.
Durham, M. G. (1998). Dilemmas of desire: Representations of adolescent sexuality in two
teen magazines. Youth and Society, 29, 369389.
Fredrickson, B. L., & Roberts, T.-A. (1997). Objectification theory: Toward understanding
womens lived experiences and mental health risks. Psychology of Women Quarterly, 21,
173206.
Fredrickson, B. L., Roberts, T.-A., Noll, S. M., Quinn, D. M., & Twenge, J. M. (1998). That
swimsuit becomes you: Sex differences in self-objectification, restrained eating, and math
performance. Journal of Personality and Social Psychology, 75, 269284.
Gerbner, G., Gross, L., Morgan, M., & Signorielli, N. (1994). Growing up with television: The
cultivation perspective. In J. Bryant & D. Zillman (Eds.), Media effects: Advances in theory
and research (pp. 1742). Hillsdale, NJ: Erlbaum.
Gluck, M. E., & Geliebter, A. (2002). Racial/ethnic differences in body image and eating
behaviors. Eating Behaviors, 3, 143151.
Greenberg, B. S. (1988). Some uncommon television images and the drench hypothesis. In
S. Oskamp (Ed.), Television as a social issue (pp. 88102). Beverly Hills, CA: Sage.
Groesz, L. M., Levine, M. P., & Murnen, S. K. (2002). The effect of experimental presentation
of thin media images on body satisfaction: A meta-analytic review. International Journal of
Eating Disorders, 31, 116.
Harrison, K. (2000). The body electric: Thin-ideal media and eating disorders in adolescents.
Journal of Communication, 50, 119143.
Harrison, K., & Fredrickson, B. L. (2003). Womens sports media, self-objectification, and
mental health in black and white adolescent females. Journal of Communication, 53,
216232.
Heinberg, L. J., Thompson, J. K., & Stormer, S. (1995). Development and validation of the
sociocultural attitudes towards appearance questionnaire. International Journal of Eating
Disorders, 17, 8189.
Kalodner, C. R. (1997). Media influences on male and female non-eating disordered college
students: A significant issue. Eating Disorders, 5, 4757.
Kolbe, R. H., & Albanese, P. J. (1996). Man to man: A content analysis of sole-male images in
male-audience magazines. Journal of Advertising, 35(4), 120.
Lavine, H., Sweeney, D., & Wagner, S. H. (1999). Depicting women as sex objects in television
advertising: Effects on body dissatisfaction. Personality and Social Psychology Bulletin, 25,
10491058.
Malkin, A. R., Wornian, K., & Chrisler, J. C. (1999). Women and weight: Gendered messages
on magazines covers. Sex Roles, 40, 647655.
McCabe, M. P., & Ricciardelli, L. A. (2001). Parent, peer, and media influences on body image
and strategies to both increase and decrease body size among adolescent boys and girls.
Adolescence, 36, 225240.

Journal of Communication 56 (2006) 366386 2006 International Communication Association 385

Sexual Objectification J. S. Aubrey

McKinley, N. M., & Hyde, J. S. (1996). The objectified body consciousness scale:
Development and validation. Psychology of Women Quarterly, 20, 181215.
McLaren, L., & Gauvin, L. (2002). Neighbourhood level versus individual level correlates of
womens body dissatisfaction: Toward a multilevel understanding of the role of affluence.
Journal of Epidemiology & Community Health, 56, 193199.
Noll, S. M., & Fredrickson, B. L. (1998). A mediational model linking self-objectification,
body shame, and disordered eating. Psychology of Women Quarterly, 22, 623636.
Phares, E. J. (1984). Introduction to personality. Columbus, OH: C.E. Merrill.
Roberts, T-A., & Gettman, J. (2004). Mere exposure: Gender differences in the negative
consequences of priming a state of self-objectification. Sex Roles, 51, 1727.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton
University Press.
Rudman, W. J., & Hagiwara, A. F. (1992). Sexual exploitation in advertising health and
wellness products. Women & Health, 18(4), 7789.
Sommers-Flanagan, R., Sommers-Flanagan, J., & Davis, B. (1993). Whats happening on
music television? A gender role content analysis. Sex Roles, 28, 745753.
Thompson, M. J. (2000). Gender in magazine advertising: Skin sells best. Clothing and Textiles
Research Journal, 18, 178181.
Thompson, J. K., Heinberg, L. J., Altabe, M., & Tantleff-Dunn, S. (1999). Exacting beauty:
Theory, assessment, and treatment of body image disturbance. Washington, DC: APA.
Tiggemann, M., & Lynch, J. E. (2001). Body image across the life span in adult women: The
role of self-objectification. Developmental Psychology, 37, 243253.
Unger, R. K., & Crawford, M. E. (1996). Women and gender: A feminist psychology (2nd ed.).
New York: McGraw-Hill.
Ward, L. M. (1995). Talking about sex: Common themes about sexuality in prime-time
television programs children and adolescents view most. Journal of Youth and Adolescence,
24, 595615.
Wright, D. L., Aquilino, W. S., & Supple, A. J. (1998). A comparison of computer-assisted
paper-and-pencil self-administered questionnaires in a survey on smoking, alcohol, and
drug use. Public Opinion Quarterly, 3, 331353.

386 Journal of Communication 56 (2006) 366386 2006 International Communication Association