Contributions of Entertainment Television to Adolescents' Sexual Attitudes and Expectations: The Role of Viewing Amount Versus Viewer Involvement.

Television's role as a sexual educator in our culture is one of contradictions. On the one hand, TV's accessibility, frankness, and popular appeal make it an excellent instructor, offering a convenient way to learn about sex without embarrassment. While TV's sexual messages are not necessarily visually explicit, they are abundant, and often provide information youth do not get elsewhere. Through its dialogue, characterizations, storylines, and themes (Roberts, 1980), television presents adolescents with numerous verbal and visual examples of how dating, intimacy, relationships, and sex are handled. On the other hand, concern is often expressed that the messages TV sends about sexuality are limited, stereotypical, and potentially harmful (see Huston, Wartella, & Donnerstein, 1998, for review). One concern is that the prevalence of sexual content on TV inadvertently overemphasizes the role of sex in male-female relationships. Of equal concern is the impression that television provides a one-dimensional picture of sexual relationships, one in which sex is only for the young, single, and beautiful, and sexual encounters are always spontaneous, romantic, and risk free. Thus, because of the prevalent yet limited nature of TV's sexual content, researchers and educators have become interested in \whether heavy viewing of these portrayals is associated with distorted expectations, irresponsible sexual decision-making, and permissive or stereotypical sexual attitudes.

Although the research in this area has been sparse, several findings do associate amount of TV viewing with viewers' sexual attitudes, expectations, and behavior. First, greater TV exposure has been linked to viewers' attitudes about sex and sexual relationships. Both heavy regular consumption of and experimental exposure to sexually-oriented genres, such as soap operas and music videos, have been related to expressing more liberal sexual attitudes (Calfin, Carroll, & Shmidt, 1993; Greeson & Williams, 1987; Strouse & Buerkel-Rothfuss, 1987), to being more accepting of sexual improprieties (Bryant & Rockwell, 1994), and to more negative attitudes toward remaining a virgin (Courtright & Baran, 1980). Second, greater TV exposure has been linked with viewers' expectations about the prevalence of certain sexual outcomes frequently depicted on television (Buerkel-Rothfuss & Strouse, 1993). For example, undergraduates who frequently view soap operas offer higher estimates of the numbers of real people who divorce or have illegitimate children than less frequent viewers do (Buerkel-Rothfuss & Mayes, 1981; Carveth & Alexander, 1985). Finally, there is tentative evidence that greater TV exposure is linked to viewers' sexual behavior. Although the amount of general television viewing typically has not been related to viewers' level of sexual activity (Brown & Newcomer, 1991; Peterson, Moore, & Furstenberg, 1991), links between exposure and greater sexual experience emerge when more sexually-oriented programming is examined (Brown & Newcomer, 1991; Strouse & Buerkel-Rothfuss, 1987; Strouse, Buerkel-Rothfuss, & Long, 1995). Together, these findings provide tentative evidence of a link between the amount of time spent watching sexually-oriented programming and viewers' own sexual attitudes, expectations, and behavior.

Up to this point, the dominant theoretical model supporting this research has been the cultivation theory. The cultivation theory (Gerbner, Gross, Morgan, & Signorielli, 1994), proposes that television's consistent images and portrayals construct a specific portrait of reality, and as viewers watch more and more television, they gradually come to cultivate or adopt attitudes and expectations about the world that coincide with this portrait. Although this model has typically been employed to explain the impact of TV violence (e.g., Gerbner & Gross, 1976), it has been applied successfully to the cultivation of attitudes about gender roles (e.g., Morgan, 1982; Signorielli, 1989), politics (e.g., Morgan & Shanahan, 1991), and marriage (Signorielli, 1991). Accordingly, if content analyses indicate that sex on TV is glamorous, prevalent, recreational, and relatively risk free, the cultivation model predicts that frequent viewers will be more inclined than sporadic viewers to hold and accept this perspective of sexuality.

However, while the results reported thus far are provocative and informative, and do support the premises of this model, they offer a limited analysis of television's role in the sexual socialization process. It is argued here that viewing amounts should be seen as only one avenue through which TV exposure might affect viewers. Indeed, the processes by which television viewing shapes viewers' attitudes and expectations are varied and complex, with several factors contributing at the same time (Bryant & Zillman, 1994). There is a richness in our experience with TV programs and characters that may not be easily represented by the amount that is watched. Therefore, to better examine contributions of television exposure, other dimensions of viewing need to be considered. In particular, a key element missing has been the role of viewer involvement. We believe that viewers' perceptions of the portrayals and their connections with the material are likely to be critical factors in the socialization process. Drawing on the premises of several theoretical perspectives, we have outlined five aspects of viewer involvement which we believe may play an important role. It is our argument that viewers who feel more connected with the material and who are more involved in the viewing experience will be most affected by it.

One dimension of involvement to be considered is viewing motivation, or viewers' expected uses of television. Proponents of uses and gratifications theories contend that people use and select media to gratify specific needs, and that specific uses lead to differences in viewing behavior and outcomes (see Rubin, 1994 for review). While some viewing is believed to be more ritualized, in which a medium is used to fill time, escape, or provide company, other media use is more instrumental, in which viewing is purposeful and goal-directed, perhaps driven by information-or arousal-seeking. Our focus is on instrumental motives, more specifically on viewing TV to learn about the world. The expectation is that the effects of TV's sexual content will be stronger among viewers who are intentionally using TV to learn about the world.

A second dimension of involvement focuses on the level of active viewing. According to this construct, people exhibit varying levels of activity when using the media, differing in their selectivity, attention to, and involvement with the content viewed. However, while the expectation is that the effect of TV content will therefore depend on one's level of active involvement, the direction of this effect is unclear. Some findings suggest that active viewers are more affected by television's messages (e.g., Levy & Windahl, 1984; Rubin & Perse, 1987); others indicate that passive viewers are most vulnerable (e.g., Hawkins & Pingree, 1982; Rouner, 1984). In mediating this debate, a current perspective suggests that different ways of being active contribute to different outcomes (Kim & Rubin, 1997). It is argued that facilitative activity, which includes selectivity, attention, and involvement, is more likely to enhance media effects, whereas inhibitory activity, which includes avoidance, distraction, and skepticism, should deter them. Focusing here on facilitative activity only, we expect that the socializing effects of TV's sexual content will be stronger among viewers who tend to engage in more facilitative activity.

A third dimension of involvement considers the perceived realism of the portrayals. Although most television programs are fictional, there are many aspects of the portrayals (such as the clothing, physical settings, dialogue, and situations) that closely resemble those in real life. Consequently, most theories predict that television's impact will increase the more realistic its content is perceived to be (Huston et al., 1998; Truglio, 1990). Evidence supports this premise, with cultivation effects appearing to work stronger for those who view the portrayals as realistic (e.g., Perse, 1986; Potter, 1986). Thus, it is expected that the socializing effect of television's sexual content will be stronger among viewers who perceive its portrayals as more realistic.

Two final dimensions of involvement focus on viewers' personal connections with the portrayals via identification and perceived relevance to the self. These dimensions are extensions of two theories that emphasize television characters as models of behavior: Greenberg's drench hypothesis (1988) and Bandura's cognitive social learning theory (1986; 1994). Greenberg's drench hypothesis proposes that specific critical portrayals may exert a stronger force on impression-formation and image-building than might the sheer frequency of television characters and behaviors viewed. This notion emphasizes the power of individual performances to affect viewers, most likely portrayals that speak to the viewer in some way. Similarly, Bandura's theory predicts that individuals will imitate or model what others do when those models are rewarded or are not punished for their behavior, and are perceived as attractive, powerful, and similar (Brown & Steele, 1995). Indeed, evidence suggests that perceiving TV figures as sexual role models is associated with more permissive sexual attitudes, more extensive sexual experience (Fabes & Strouse, 1987), and greater dissatisfaction with one's sexual status and sexual experiences (Baran, 1976a; 1976b). Thus, it is expected here that the contributions of TV's sexual content will be stronger among viewers who more closely identify or connect with the portrayals.

In summary, our goal was to begin to determine how television operates as a sex educator by exploring which aspects of TV use are associated with teens' sexual attitudes and expectations. This study expands on previous efforts by including both viewing amount and viewer involvement as central variables. Drawing on premises of several theoretical approaches, we defined viewer involvement to include viewing motivation, active viewing, perceived realism, perceived relevance, and identification. Our critical research questions and hypotheses read as follows:

   1. Do viewing amounts correlate with sexual attitudes and expectations? In
   general correlational analyses we expected to see a replication of
   previously-reported cultivation effects, such that as viewing amounts
   increased, there would also be increases in viewers' acceptance of sex as
   recreation and in their expectations about the prevalence of sexual
   activity.

   2. Does viewer involvement correlate with viewers' sexual attitudes and
   expectations? Here it was expected that viewers who are more involved in
   viewing (e.g., actively engaged, strongly identified with the portrayals)
   would exhibit more liberal sexual attitudes and expectations. The belief
   was that if TV's sexual content is as prevalent yet superficial as studies
   report, then viewers who are more involved with this content, regardless of
   their actual amount of viewing, are likely to be affected in these ways.

   3. Which TV factors best predict adolescent sexual outcomes: amount viewed
   or viewer involvement? This question provides a direct test of cultivation
   theory versus more constructivist models of media effects. In regression
   analyses that include several types of TV variables, it was expected that
   viewer involvement would contribute more to sexual outcomes than would
   viewing amounts.

Although we acknowledge that the relationships described are bi-directional, and that viewers' preexisting sexual attitudes and expectations affect their viewing choices, the focus of this study will be television's contribution to various sexual outcomes. By breaking TV exposure into more specific variables, we hoped to better document its reach and impact.

METHOD

Participants

Participants were 314 university undergraduates (55% female) from both the University of Michigan (N = 158) and the University of California, Los Angeles (N = 156) who agreed to participate as part of their Introduction to Psychology course requirement. Participation in this particular study was a random decision, and students had no a priori knowledge of the study's content. Although the students ranged in age from 18 to 24 (Mean = 19.36), 97% were between the ages of 18 and 20. Students' responses to an open-ended question about their racial/ethnic background/identity revealed that 54% were White/European-American/Caucasian, 26% were Asian-American/Pacific Islander/Middle Eastern, 12% were Latino/a, and 8% were Black/African-American. Approximately 95% of the participants had spent the majority of their formative years (ages 5 to 15) in the United States.

A variable representing socioeconomic status (SES) was constructed for each participant based on information about his or her mother's education (Mean = 15.3 years), father's education (Mean = 16.2 years), and each parent's occupation. Occupations were first rated for prestige using the 1989 Socioeconomic Index of Occupations (Nakao & Treas, 1992). A z-score was then calculated for each of the parental education and occupational prestige variables. Finally, these four z-scores were summed to create a z-score representing the relative SES of each student.

Measures of Television Viewing Behaviors

This study focused on adolescents' viewing of TV comedies and dramas. These particular genres were selected above other formats, such as music videos, because our focus was on sexual content presented via recurring characters to whom viewers can connect or identify. Three aspects of viewing were assessed: frequency, general involvement, and perceptions of sexual content.

Frequency. To measure viewing amounts, a list was provided of all network comedies and dramas that had aired regularly (at least four times) on one of the six major networks (ABC, CBS, NBC, Fox, UPN, and WB) during the previous school term. Although adolescents sample from many cable networks and channels, we chose to focus on the six major networks because these still attract the bulk of TV viewers (e.g., Bauder, 1998). Using a 5-point scale, participants indicated how often they had viewed each of the more than 150 programs listed during an average month during the previous school term. The prime-time and weekend daytime syndicated programs were rated using five frequency markers ranging from every week to never/not this season. The soap operas and weekday syndicated programs (which aired five times a week) were rated using five frequency markers ranging from once a day/almost every day to never/not this season. Average monthly viewing amounts were then calculated based on the frequencies with which each program was viewed and the length of the program. The key variables of interest resulting for each participant were the Monthly Hours of Prime Time Viewing, the Monthly Hours of Soap Opera Viewing, and the Total Monthly Viewing Hours (of comedies and dramas).

General involvement. The second group of measures assessed participants' general involvement in the viewing of comedies and dramas. One measure examined the degree to which viewers watch TV to learn about the world. Here, participants were asked to rate each of eleven possible motivations for watching television comedies and dramas. Individual motives reflecting this learning perspective were extracted from previous research (Carveth & Alexander, 1985; Rubin, 1984; Sun & Lull, 1986), and were compiled into one measure. To rate each motivation participants used a 6-point scale that ranged from 1 (strongly disagree) to 6 (strongly agree). Examples of motivations include "because they help me learn about myself and about others" and "to help me understand the world." Ratings of agreement with the eleven motives were summed to produce one Learning Motive score for each participant.

A second measure assessed how active or passive each participant was as a television viewer. This instrument was based on three types of audience activity outlined by Levy (Levy, 1983; Levy & Windahl, 1984) and others (e.g., Rubin & Perse, 1987) that tap some of the dimensions of facilitative activity. The first aspect, selectivity-in-exposure-seeking, reflects the degree to which viewers intend to enter the communication setting. The second, involvement-during-exposure, reflects the degree to which viewers either attend to the communication situation or engage in information processing of the messages presented. The third dimension, post-exposure use, reflects the extent to which viewers use the messages acquired in their social or psychological life.

To assess all of these dimensions in one measure, the Active Viewing Measure was developed based partly on items from the research cited above and from personal experiences. An initial list of 31 statements was developed reflecting the different aspects of active viewing, and was pilot tested with 74 undergraduates. Based on their responses, three additional statements were added and the wording of ambiguous items was changed. In the final group of 34 statements, 8 reflected selectivity in exposure-seeking (e.g., "I often plan my day around the TV shows I like to watch"), 17 reflected involvement during exposure (e.g., "I often try to guess what will happen next or how an episode will end"), and 9 reflected post-exposure use (e.g., "I frequently talk to others about what I have recently seen on TV shows"). Agreement with each statement was rated using a 6-point scale that ranged from not at all like me to very much like me.

Finally, reliability analyses were conducted using the sample of 314 undergraduates in this study. Although the initial Cronbach alpha reported for each subscale was greater than .7, seven individual items were removed where results had indicated that doing so would strengthen the cohesiveness of the scales. The final alphas exceeded .8 for each subscale (Selectivity = .83, Involvement = .84, Use = .81), and reached .91 when the subscales were combined. Thus, this remaining set of 27 items served as the measure of Active Viewing in this study.

Perceptions of televised sexual content. The remaining viewer involvement variables assessed participants' perceptions of the type of sexual content frequently presented on television. Serving as stimuli were four clips selected from the Ward (1995) analysis of TV's sexual content. The programs from which the clips were chosen (Roseanne, Martin, Home Improvement, and Family Matters) were among the 10 most widely viewed by adolescents aged 12-17 during the 1992-93 broadcast season, and were still airing during this study, either in prime-time or in syndication. Clips from programs that were currently most popular (e.g., Seinfeld, Friends) were not used, in order to reduce the likelihood that the scenes had been recently viewed and discussed. However, because the programs selected were among the most popular shows five years ago, study participants are likely to have been exposed to the material at some point.

Each clip was approximately 2 to 3 minutes in length, and dealt with one or more of the following relationship issues: jealousy and insecurity, lust, misunderstandings, fidelity, and temptation. These themes were representative of the types of sexual messages frequently presented on programs popular among adolescents (Ward, 1995). In addition, pilot testing with several clips and with 74 undergraduates had revealed that the four scenes selected did, indeed, convey the themes in question. These clips also represented a range of ethnic groups, relationship stages, and ages of the focal characters.

Participants rated the situation presented in each clip for its realism, likelihood to happen in their own lives, appropriateness of the main character's behavior, and level of identification with the main character. Perceptions were rated on scales of 1 to 4, with a score of 1 reflecting less of each construct. Mean scores were then computed across the four clips, resulting in an Average Realism, Average Identification, and Average Relevance score (i.e., likelihood to happen in own life) for each participant. These averages correlated highly (from .61 to .71) with the ratings for each of the four individual clips. However, for the fourth construct, in which participants rated the appropriateness of the main character's behavior, ratings did not correlate across the four clips. Instead, there was great variability in how positively and negatively the behavior of the main character was perceived to be. Consequently, a mean score was not computed for Perceived Appropriateness, and this construct was excluded from further analyses.

Sexual Outcome Measures

A set of seven sexual outcome measures was selected and/or created to reflect three dimensions of participants' sexuality: their attitudes, their expectations about other's behavior, and their own level of sexual experience.

Attitudes about sexuality. Participants' attitudes about dating, sexual roles, and romantic relationships were assessed using the Attitudes Toward Dating and Relationships Measure created for this study. Because a critical goal of this study was to determine whether television exposure increases viewers' acceptance of the types of attitudes and values frequently presented on TV, it was decided that the best outcome measure would be one that incorporated the exact types of values typically portrayed. Accordingly, the coding system used in the Ward (1995) analysis of television's sexual content was adapted into a survey measure designed to capture respondents' attitudes toward 17 themes about sexuality common in the larger culture and on television. A list of 33 statements reflecting these themes was developed and organized into three larger categories based on Ward's (1995) analysis. A Recreational Orientation to sexuality was represented by 18 statements in which dating and sexual relations are portrayed as a game or competition. Here, the sexual double standard prevails, and males and females are viewed as being on opposing teams ("battle-of-the-sexes"), taking different roles, and using sex as the object of exchange. A Procreational or Traditional Orientation to sexuality was represented by eight statements in which sexuality centers around religious dictates and traditional courtship norms. Here, sex is seen to belong in marital relationships only. Finally, a Relational Orientation to sexuality was represented by seven statements in which sex is seen to be one part of a loving relationship based on friendship, mutuality, trust, responsibility, and open communication. Participants rated their level of agreement with each statement using a 6-point scale that ranged from 1 (strongly disagree) to 6 (strongly agree).

Reliability analyses of this measure and its three subscales were conducted using the sample of 314 undergraduates in this study. For the Recreational Orientation and the Procreational Orientation subscales, the initial Cronbach alphas reported were each above .7; however, one item was removed from each subset to improve the cohesiveness of these scales, bringing the final alphas to .84 and .75 respectively. For the Relational Orientation subscale, both the initial and revised alphas fell below .7. Because nearly all subjects (typically 90%) either agreed or totally agreed with these statements, there was little variability in the scores of these items. This subscale was therefore dropped from the study.

Because gender roles shape sexuality (e.g., Schwartz & Rutter, 1998), and holding traditional gender role attitudes has been linked both with more TV viewing (e.g., Morgan, 1982; Signorielli, 1989) and with higher levels of risk-related sexual behavior (e.g., Fox, 1977; Pleck, Sonenstein, & Ku, 1993; Whitley, 1988), an attempt was made to assess participants' attitudes about the roles and appropriate behaviors of women and men. For these purposes, the Attitudes Toward Women Scale for Adolescents (Galambos, Petersen, Richards, & Gitelson, 1985) was used. Respondents rated their level of agreement with each of twelve statements using a 4-point scale. Sample statements include "Swearing is worse for a girl than for a boy" and "Boys are better leaders than girls." Scores were summed across the 12 items to yield a total score ranging from 12 to 48, with higher scores reflecting more traditional attitudes toward gender. The authors report alpha levels of .78 and .72 for males and females in grades 6 through 12, which are comparable to the coefficients of .80 and .79 obtained among the undergraduates in our sample.

Sexual expectations and behavior. To assess participants' expectations about the level of sexual activity among their peers, respondents were asked to estimate the number of males and females who had engaged in a list of 12 risk-related or recreational sexual behaviors (e.g., "have had sex on the first date," "will usually have sex without using birth control"). These behaviors had been selected from general literature on sexual risks (e.g., Metzler, Noell, & Biglan, 1992). For each behavior, participants were asked to circle a number from 0 to 10 that reflected their sense of the percentage of males and the percentage of females who had engaged in that behavior by age 18. Three scores were extracted from this measure. One score reflected the sum of the estimates for both sexes (across the 24 behaviors), and provided an overview of the level of risk-related sexual experience participants expected of their peers. Also extracted from this measure were participants' estimates of the percentage of females and the percentage of males who have had sex by age 18. These two particular questions were selected because their responses could be compared to the actual levels of these behaviors in the real world.

The final sexual outcome measure assessed participants' own levels of sexual experience. Because this study focuses on the links between TV exposure and viewers' sexual attitudes and expectations, the inclusion of a behavioral variable was purely exploratory. During the second phase of the study (i.e., the testing of the UMICH sample), the decision was made to expand assessment by including the following question about sexual behavior: "How would you describe your current level of experience with dating and sexual relationships?" Responses were made on an 11-point scale that ranged from just starting out and some dating to have had several sexual relationships.

Procedure

Students were tested in small groups in a university research room during the spring of 1996 and the spring of 1997. Upon arrival to the session, participants were informed that this was a study about viewers' perceptions of how dating and relationships are portrayed on television. After the study was introduced and consent was obtained, each participant received a packet to complete containing all of the measures for the study. At one point during the study, the four clips were presented. Participants were given time in between each clip to answer questions about it. Once all of the measures were completed, participants returned their packets, were thanked for their participation, and received a written debriefing information sheet (UMICH sample only) providing more information about the study.

Several steps were taken during the execution of this study to reduce experimental effects. First, two viewing conditions were established to allow us to investigate possible connections between watching/responding to TV clips and answering questions about one's sexual attitudes. In one condition (N = 163; 52% of sample) participants watched and responded to the clips first, and then completed questions assessing their sexual attitudes and expectations. In the second condition (N = 151; 48% of sample), these events were reversed. Second, the ordering of the questionnaires was randomized across the entire sample. Finally, the clips were presented in one of four orders, and these orderings were equally represented across the two experimental viewing conditions.

Analyses were conducted to examine the extent of any experimental effects of the Viewing Condition (i.e., viewing clips before or after completing sexual outcome measures) and the Clip Order on the main sexual outcome variables (excluding Sexual Experience). Six 2 (Viewing Condition) x 4 (Clip Order) analyses of covariance were run, with relevant demographic variables serving as covariates and six of the seven sexual outcome variables serving as the dependent variables. Two significant results emerged. First, there was a significant effect of the Clip Order on participants' gender role attitudes, F (3, 310) = 3.224, p = .023. However, subsequent post-hoc tests revealed that scores from no one Clip Order differed significantly from scores from any other. Consequently, this outcome was considered to be informative yet somewhat spurious, and was not analyzed further. Second, there was a significant effect of the Viewing Condition on participants' Endorsement of Traditional Dating Norms, F (1, 313) = 3.002, p = .032. Here, participants who watched the clips before completing the sexual outcome measures were less likely to endorse Traditional Dating Norms than were participants who viewed the clips later in the session. This suggests that watching the clips first may have had a liberalizing effect on participants' sexual attitudes. To control for this outcome, Viewing Condition was entered as a control in subsequent analyses of participants' Endorsement of Traditional Dating Norms.

RESULTS

Preliminary Analyses

Descriptive statistics of the independent and dependent variables are provided in Table 1. In general, the viewing behaviors and sexual experiences for this group were quite diverse.

Table 1. Descriptives of Key Variables

Variable                                Mean    Standard Deviation

Viewing amounts
  Prime-time hours                      24.69         14.95
  Soap opera hours                       7.91         14.93
  Total hours                           60.92         48.23
Viewer involvement
  Active viewing                         3.17          0.77
  View to learn motive                  30.03         10.62
  Perceived realism                      3.14          0.38
  Perceived relevance                    2.49          0.56
  Identification                         2.87          0.48
Sexual outcome measures
  Gender role attitudes                 20.43          5.05
  Recreational sex attitudes             3.23          0.64
  Traditional dating norms               3.00          0.83
  Estimate of female sex rate           57.4          19.4
  Estimate of male sex rate             70.9          16.7
  Estimate of peer sexual experiences   95.9          29.03
  Own sexual experience                  4.6           2.27

Variable                                Minimum   Maximum

Viewing amounts
  Prime-time hours                         0         87.0
  Soap opera hours                         0        113.0
  Total hours                              0        309.0
Viewer involvement
  Active viewing                           1.3        5.3
  View to learn motive                    11         64.0
  Perceived realism                        2          4.0
  Perceived relevance                      1          4.0
  Identification                           1          4.0
Sexual outcome measures
  Gender role attitudes                   12         34.0
  Recreational sex attitudes               1.1        5.2
  Traditional dating norms                 1.3        5.4
  Estimate of female sex rate             10        100
  Estimate of male sex rate               20        100
  Estimate of peer sexual experiences     30        216
  Own sexual experience                    0         10

Participants watched an average of 25 hours of primetime comedies and dramas each month, and from 0 (half the sample) to 113 hours of soap operas each month (Mean = 7.9). Of the 157 participants (UMICH only) who were asked about their own level of sexual experience (Mean = 4.6 of 10), 43% were likely to be virgins (scores of 4 or lower) and 57% were likely nonvirgins.

Preliminary analyses were also conducted examining zero-order correlations between the key independent and dependent variables and the five demographic factors. Because previous research indicates extensive sex and ethnic group differences in both TV viewing behaviors (e.g., Greenberg, 1993; Huston & Wright, 1998) and in sexual outcomes (e.g., Day, 1992; Schwartz & Rutter, 1998; Udry & Billy, 1987), it was anticipated that many of the connections in question would vary by viewers' personal backgrounds. Findings summarized in Table 2 are congruent with these reports. In general, being Black or Latino/a was correlated with heavier total exposure, and being female or Black were each associated with higher estimates of the percentage of sexually active peers. The demographic variables tested here were included as covariates in later analyses, and separate analyses were conducted for each sex.

Table 2. Zero-Order Correlations Between Demographics and TV
Viewing and Sexual Outcome Variables

                              Age       Asian      Black

Viewing amounts
 Prime-time hours                                  .12(*)
 Soap hours                                        .19(**)
 Total hours                                       .18(**)
Viewer involvement
 Active viewing
 View to learn
 Perceived realism
 Identification
 Perceived relevance
Sexual outcomes
 Traditional gender roles
 Recreational sex attitudes                        .13(*)
 Traditional dating norms               .24(***)
 Female sex rate estimate                          .13(*)
 Male sex rate estimate                            .20(*)
 Estimate of peer sex                              .19(**)
 Own sexual experience        .16(*)   -.20(*)

                              Latino/a    Female     SES

Viewing amounts               .20(***)
 Prime-time hours                         .29(***)
 Soap hours                   .22(***)
 Total hours
Viewer involvement
 Active viewing
 View to learn                                        -.12(*)
 Perceived realism            .17(**)     .16(**)     -.13(*)
 Identification
 Perceived relevance          .12(*)
Sexual outcomes
 Traditional gender roles                -.48(***)
 Recreational sex attitudes              -.40(***)
 Traditional dating norms
 Female sex rate estimate                 .14(*)
 Male sex rate estimate                   .27(***)
 Estimate of peer sex         .13(*)      .15(**)
 Own sexual experience

                              School/UCLA

Viewing amounts
 Prime-time hours             .19(**)
 Soap hours
 Total hours                  .19(**)
Viewer involvement
 Active viewing
 View to learn
 Perceived realism
 Identification
 Perceived relevance
Sexual outcomes
 Traditional gender roles
 Recreational sex attitudes
 Traditional dating norms
 Female sex rate estimate
 Male sex rate estimate
 Estimate of peer sex         .20(***)
 Own sexual experience

(*) p < .05.

(**) p < .01.

(***) p < .001.

Testing of the Main Research Questions

The first set of analyses tested the premises of the cultivation model. More specifically, do viewing amounts correlate with participants' sexual attitudes, expectations, and behavior? Correlational analyses were run between the three viewing amounts variables and the seven sexual outcome variables, controlling for contributions of the four demographic factors.

Although support for this model was not overwhelming, several significant results did emerge among the females, and each of these relationships was in the expected direction. First, watching more hours of prime-time television was linked to a stronger endorsement of recreational attitudes about sex, r (164) = .18, p [is less than] .05. Second, heavier viewing of soap operas in particular and of TV comedies and dramas in general were each linked to greater experience with dating and sexual relationships, r (80) = .24, p [is less than] .05, and r (80) = .23, p [is less than] .05, respectively. Finally, greater total exposure to television comedies and dramas was linked to holding more traditional gender role attitudes, r (149) = .18, p [is less than] .05. Therefore, regardless of women's demographic backgrounds, watching more hours of television comedies and dramas was associated with exhibiting attitudes and experiences that closely resemble the image of sexuality TV frequently depicts.

The second set of analyses tested the premises of the more constructivist models. More specifically, do viewers' perceptions of TV's content and their general viewing involvement correlate with their sexual attitudes, expectations, and behavior? Correlational analyses were run between the five viewer involvement variables and the seven sexual outcome variables, controlling for contributions of the four demographic factors. These results are summarized in Table 3. Strong support emerged for the role of viewer involvement, with each of the five involvement factors contributing in some way, and in the direction expected. As results for each sex were strikingly similar yet stronger among the women, our discussion focuses on these findings.

Table 3. Partial Correlations Between Viewer Involvement and
Sexual Outcomes

                Endorse        Endorse      Traditional   Estimates
             Recreational    Traditional    Gender Role   of Female
             Sex Attitudes   Dating Norms    Attitudes    Sex Rate

Females
 Activity       .16(*)
 Learning       .25(**)         .14(+)
 Realism                                                   .20(*)
 Relevance                     -.17(*)                     .14(+)

Identify        .18(*)                                     .25(**)
 Males
 Activity
 Learning
 Realism
 Relevance
 Identify       .17(*)                        .16(+)

             Estimates   Estimate of      Own Level
              of Male    Per Sexual       of Sexual
             Sex Rate     Activity        Experience

Females
 Activity
 Learning
 Realism      .21(**)      .18(*)
 Relevance                                  .21(+)

Identify      .19(*)       .15(+)           .23(*)
 Males
 Activity
 Learning
 Realism                   .17(*)           .23(+)
 Relevance    .16(+)       .20(*)           .45(***)
 Identify                  .24(**)          .23(+)

Note. Correlations are controlled for contributions of age, SES, ethnic group, and school.

(+) p < .075.

(*) p < .05.

(**) p < .01.

(***) p < .001.

Level of identification with the characters produced the strongest connection to the sexual outcome variables for both sexes. Among the female participants, identifying more closely with the TV characters in the sexual situations viewed was associated with stronger endorsement of recreational attitudes toward sex, higher expectations of the level of sexual experience of one's male and female peers, and more extensive experience with sexual relationships. At the same time, perceiving the sexual situations viewed as likely to happen in one's own life was associated with a weaker endorsement of traditional dating norms. In addition, the more realistic participants perceived the portrayals to be, the more sexually experienced they expected their peers to be. Finally, being a more active television viewer and watching comedies and dramas to learn about the world were each associated with a stronger endorsement of a recreational attitude toward sex. Although other notable relationships emerged, as indicated in Table 3, they did not reach conventional levels of statistical significance.

Together these results lend strong support to more constructivist analyses of television's impact. Regardless of their sex, race, or SES, adolescent viewers who experienced greater connection with the sexual situations typically portrayed on network television were also more likely to endorse recreational attitudes toward sex, to expect higher levels of sexual activity among their peers, and to be more sexually experienced.

The final series of analyses investigated which TV factors best predict adolescent sexual outcomes: viewing amounts or viewers' perceptions of and involvement with the sexual content viewed. To address this issue, a series of regression analyses was conducted in which all of the viewing variables were examined simultaneously. For these analyses, the seven sexual outcome variables served as the dependent variables, the four demographic factors were entered first as a block, and the TV viewing variables were entered together as the central predictor variables. Because total viewing hours correlated strongly with hours of prime-time viewing, r (278) = .786, p = .000, the Total Hours variable was not included to reduce the probability of multicollinearity.

The results for the first series of regression analyses, in which only females were included, are summarized in Tables 4 and 5. Here, three of the six regression equations were statistically significant and two others approached significance, with the independent variables explaining from 2.6% to 17.6% of the variance. The block of demographic variables alone accounted for .5% to 13% of the variance. Changes in the Adjusted [R.sup.2]'s brought about by adding in the TV variables were substantial, ranging from an additional 1% to 8.2% of the variance. Among these variables, the contributions of viewing amount were minimal. Instead, the most prominent TV viewing predictors were Viewing to Learn and Identification with the portrayals.

Table 4. Predictors of Sexual Attitudes for Females

                                        Endorse Recreational
                                           Sex Attitudes

                                                B     [Beta]

Adjusted [R.sup.2] for demographic      .01
 variables
Viewing amounts
   Prime-time                                  .01     .13
   Soap operas                                 .00     .06
Viewer involvement
   Active                                      .03     .04
   View to learn                               .01     .17(*)
   Identify                                    .25     .19(+)
   Realism                                    -.06    -.04
   Relevance                                  -.05    -.05
Total Adjusted [R.sup.2]                .08
Change in Adjusted [R.sup.2]         (+).07
F                                      2.11
Signif F                                .02

                                        Endorse Traditional
                                            Dating Norms

                                                B     [Beta]

Adjusted [R.sup.2] for demographic      .13
 variables
Viewing amounts
   Prime-time                                 -.01     .16(+)
   Soap operas                                -.00     .06
Viewer involvement
   Active                                      .11     .01
   View to learn                               .01     .16(*)
   Identify                                   -.19    -.10
   Realism                                     .12     .06
   Relevance                                  -.18    -.12
Total Adjusted [R.sup.2]                .18
Change in Adjusted [R.sup.2]         (+).05
F                                      3.51
Signif F                               <.001

                                         Endorse Traditional
                                        Gender Role Attitudes

                                                 B     [Beta]

Adjusted [R.sup.2] for demographic      .04
 variables
Viewing amounts
   Prime-time                                    .03      .10
   Soap operas                                  -.01     -.05
Viewer involvement
   Active                                        .50      .09
   View to learn                                 .04      .10
   Identify                                    -1.36     -.14
   Realism                                      -.42     -.04
   Relevance                                    -.04     -.01
Total Adjusted [R.sup.2]                .05
Change in Adjusted [R.sup.2]         (+).01
F                                      1.64
Signif F                                .08

Note. N = 164.

(+) p < .075.

(*) p < .05.

(**) p < .01.

(***) p < .001.

Table 5. Predictors of Sexual Expectations for Females

                                Female Sexual Activity

                                          B      [Beta]

Adjusted [R.sup.2] for
 demographic variables            .01
Viewing amounts
  Prime-time                               .00    .03
  Soap operas                              .00    .01

Viewer involvement
  Active                                  -.00   -.00
  View to learn                           -.01   -.07
  Identify                                 .99    .22(*)
  Realism                                  .56    .11
  Relevance                               -.03   -.01

Total Adjusted [R.sup.2]          .03
Change in Adjusted [R.sup.2]   (+).03
F                                1.33
Signif F                          .20

                                 Male Sexual Activity

                                          B      [Beta]

Adjusted [R.sup.2] for
 demographic variables            .01
Viewing amounts
  Prime-time                               .01       .05
  Soap operas                              .02       .16
                                                 p = .08

Viewer involvement
  Active                                   .15       .07
  View to learn                           -.03      -.18(*)
  Identify                                 .87       .23(*)
  Realism                                  .88       .21(*)
  Relevance                               -.53      -.18(+)

Total Adjusted [R.sup.2]          .09
Change in Adjusted [R.sup.2]   (+).08
F                                2.20
Signif F                          .01

                               Estimate of Peer Sexual Activity

                                          B      [Beta]

Adjusted [R.sup.2] for
 demographic variables            .03
Viewing amounts
  Prime-time                               .24    .11
  Soap operas                              .02    .01

Viewer involvement
  Active                                   .17    .00
  View to learn                           -.10   -.04
  Identify                               11.65    .17
  Realism                                13.87    .18(+)
  Relevance                             -10.90   -.20(*)

Total Adjusted [R.sup.2]          .05
Change in Adjusted [R.sup.2]   (+).03
F                                1.67
Signif F                          .07

Note. N = 164.

(+) p < .075.

(*) p < .05.

(**) p < .01.

(***) p < .001.

Looking more closely at the individual regression equations, the results closely resemble those presented by the correlational analyses. The viewing variables that best predicted women's attitudes about sex as recreation were their viewing motive and level of identification. Once again, viewing television to learn about the world and strongly identifying with the portrayals were each associated with a greater acceptance of sex as recreation. Endorsing traditional dating norms was associated with viewing TV to learn about the world and with viewing fewer hours of prime-time television. No TV viewing variables predicted females' gender role attitudes.

The TV viewing variables also appeared as strong predictors of females' expectations about the sexual behavior of their peers. Giving higher estimates of the number of female peers who are sexually active was associated with a stronger identification with the portrayals viewed. Females who believed their male peers to be sexually active and experienced also identified more with the TV portrayals presented and saw these portrayals as more realistic. Ironically, these women also reported less inclination to use TV to learn about the world and perceived the sexual situations as less relevant to their own lives. Findings concerning women's perceptions of general peer sexual experience were similar, yet less extensive. Together these results illustrate varied and complex ways in which TV's sexual content connects with female viewers' developing conceptions of sexuality.

The results for the second series of regression analyses, in which only males were tested, offer a stark contrast to the findings obtained among the female participants. For the males, only two of the six regression equations were statistically significant, and only one viewing variable emerged as a significant predictor. Here, viewing more hours of soap operas was associated with weaker endorsement of a recreational attitude toward sex, Standardized [Beta] = -.206, p [is less than] .05, Adjusted [R.sup.2] = .009. Instead, the majority of the variance in males' sexual attitudes and expectations was explained by the demographic factors, which accounted for 2.7% to 15.8% of the variance.

The last set of regression analyses examined predictors of sexual experience. These analyses differed from the previous ones in two ways. First, because only the UMICH students had been questioned about their own sexual experiences, the sample sizes for the single-sex regression equations were quite small. Therefore, the sample was analyzed separately by gender and then together in order to increase the statistical power. Second, to better predict a construct as complex as sexual behavior, it was decided that all of the TV viewing, demographic, and other sexual outcome variables should be included as potential predictors. However, because the single-sex samples were not large enough to sustain all 19 predictor variables, preliminary analyses were conducted to select only those demographic and sexual outcome variables that correlated with the sexual experience of this sample of males and females. These significant correlates were then included with the seven viewing variables in the single-sex regression equations summarized in Table 6.

Table 6. Predictors of Sexual Experience

                                Total Sample
                                 (N = 152)
                                B      [Beta]

Demographics
  Age                           .06    .22(**)
  SES                           .14    .16(*)
  Sex (female)                 -.35   -.08
  Asian                        -.81   -.11
  Black                        2.32    .22(**)
  Latino/a                      .16    .01
Viewing amounts
  Prime-time                   -.02   -.14
  Soap operas                   .03    .14
Viewer involvement
  Active                       -.09   -.03
  View to learn                -.00   -.01
  Identify                      .51    .10
  Realism                     -1.05   -.18
  Relevance                    1.32    .30(**)
Attitudes
  Gender role                   .10    .20
  Recreational sex             -.54   -.14
  Traditional dating          -1.05    .38(***)
Expectations
  Female sex rate              -.09   -.08
  Male sex rate                 .43    .32(*)
  Total peer sex                .00    .01
Adjusted [R.sup.2]      .33
F                      4.88
Signif F               <.01

                                  Females
                                  (N = 84)
                                B      [Beta]

Demographics
  Age                          N/A
  SES                          N/A
  Sex (female)                 N/A
  Asian                        -1.2    -.17
  Black                        N/A
  Latino/a                     N/A
Viewing amounts
  Prime-time                   -.02    -.13
  Soap operas                   .03     .19
Viewer involvement
  Active                       -.13    -.05
  View to learn                 .00     .01
  Identify                      .96     .18
  Realism                     -1.2     -.23
  Relevance                     .85     .20
Attitudes
  Gender role                  N/A
  Recreational sex             N/A
  Traditional dating           -.65    -.26(*)
Expectations
  Female sex rate              N/A
  Male sex rate                 .40     .29(**)
  Total peer sex               N/A
Adjusted [R.sup.2]      .27
F                      4.08
Signif F               <.01

                                   Males
                                  (N = 68)
                                B      [Beta]

Demographics
  Age                           .11    .36(**)
  SES                          N/A
  Sex (female)                 N/A
  Asian                        N/A
  Black                        3.30    .31(**)
  Latino/a                     N/A
Viewing amounts
  Prime-time                    .00    .01
  Soap operas                  -.12   -.12
Viewer involvement
  Active                       -.02   -.01
  View to learn                -.00   -.01
  Identify                      .18    .04
  Realism                     -1.04   -.16
  Relevance                    2.00    .43(**)
Attitudes
  Gender role                  N/A
  Recreational sex             -.18   -.04
  Traditional dating           -.82   -.26(*)
Expectations
  Female sex rate              N/A
  Male sex rate                 .30    .20
  Total peer sex               N/A
Adjusted [R.sup.2]      .36
F                      4.07
Signif F               <.01

Note. Sexual experience

(+) p < .075.

(*) p < .05.

(**) p < .01.

(***) p < .001.

For the sample as a whole, the strongest predictors of participants' sexual experience were their rejection of traditional dating norms and their high expectations of the sexual experience of their male peers. The only TV predictor of participants' level of sexual experience was their Perceived Relevance of TV's sexual content. Those perceiving the sexual situations viewed as more likely to happen in their own lives also reported greater levels of experience with sexual relationships. As expected, holding more traditional attitudes toward gender roles was also associated with greater levels of sexual experience. Adjusted [R.sup.2]s for this equation indicate that the five demographic factors accounted for 8.2% of the variance; the seven viewing variables contributed an additional 9.2%; and participants' own sexual attitudes and expectations accounted for an additional 15.6%.

When the results were analyzed by gender, it became clear which factors served more strongly as predictors for females than for males. For the female participants, holding higher expectations about the level of sexual experience of their male peers and giving less support to traditional dating norms emerged as the only significant predictors of their level of sexual experience; neither the demographic factors nor the viewing variables figured in. For the males, the one viewing variable predicting level of sexual experience was Perceived Relevance. Males perceiving the sexual situations viewed as more likely to happen in their own lives also reported higher levels of experience with sexual relationships. Additional factors contributing to greater sexual experience were being older, being Black, and being less accepting of traditional dating norms. Together, these predictors accounted for approximately 36% of the variance in males' level of sexual experience and 27% of the variance for females.

DISCUSSION

The goal of this project was to begin to sort out how television viewing might relate to adolescents' sexual attitudes and expectations. Results indicate that TV viewing is associated with sexual outcomes in important and diverse ways, and that these associations typically differ by sex. While not every connection tested produced a statistically significant outcome, nearly all of those that did were in predicted directions. Both viewing amount and viewer involvement were significant correlates of participants' sexual attitudes, expectations, and behavior. Yet when all of the viewing factors were tested together in the regression analyses, viewer involvement emerged as a more consistent predictor of the sexual outcomes examined. Thus, it appears as if it is not the amount viewed that matters most, but viewers' connections with the content encountered.

Within these results, two interesting trends emerged. First, TV viewing appears to be a stronger predictor of viewers' expectations of their peers' behavior than of their own sexual attitudes or experiences. Whereas Viewing Motive was the only viewing variable that significantly predicted females' sexual attitudes, several viewing variables predicted their expectations about the prevalence of sexual activity among their peers. The general trend was that women expected men to be quite sexually active, estimating that 75% of 18-year-old males are sexually experienced (compared to 66% estimated by men). Television viewing appears to contribute to this perception in several ways. More specifically, identifying with the portrayals, perceiving them to be realistic, perceiving the actions as less likely to happen in one's life, and less frequent use of TV to learn about the world were each associated with giving higher estimates of male (and often female) sexual experience. Females watching soaps more frequently gave notably higher estimates as well. Thus, our data indicate that females, in general, believe that their male peers are quite sexually experienced, and that several types of interactions with TV contribute to this perception.

It is no surprise that TV exposure is associated more strongly with females' conceptions of male sexuality than it is of female sexuality. Television's impact is assumed to be greatest in areas in which viewers have the least personal experience (Dom 1986). While women are likely to have more knowledge about female sexual behavior from their own lives and from direct contact with their female peers, they are likely to know less about how men feel and to have less direct knowledge of male sexual behavior. Therefore, women may be more inclined to look to and accept outside images of men. Indeed, a key sexual theme on TV is that males are always thinking about sex and are perpetually ready and willing (Ward, 1995). Our data suggest that whether women are heavily involved viewers or are less purposeful, they may not be able to avoid picking up information about this ubiquitous stereotype of male sexuality.

The second provocative trend in the results is the sex difference: more specifically, the finding that for males, TV viewing associates only weakly with their sexual attitudes and expectations. One of the few associations uncovered was that heavier viewing of soap operas predicts weaker endorsement of recreational attitudes toward sex. This finding runs counter to both our initial expectations and to trends found among females. Perhaps this finding indicates something significant about the types of males drawn to soap operas. Only 27% of all males sampled reported watching any soaps at all (versus 66% of the females), and nearly half of this group watched only 2 hours a month. It is possible that the types of males who watch soap operas regularly may be more relationship-focused. Indeed, Alexander (1985) reported that among adolescents aged 10 to 14, males who had watched soaps longer believed more strongly in the importance of talk in relationships, and that males who currently watched more hours of soaps perceived relationships as more fragile. Neither outcome emerged for females in that study. However, because heavier viewing of soap operas among undergraduate males has been linked to having more sexual partners (Strouse & Buerkel-Rothfuss, 1987), further research is needed to help sort out the contributions for males of soap opera viewing.

Why did the viewing variables not make a stronger contribution among the males? Some argue that men appear less likely than women to use media to learn about interpersonal relationships (Huston et al, 1998). Moreover, for those who do, the genres examined here--dramas, situation comedies, and soaps--may not be their most preferred "instructors." Instead, it is possible that males may rely more on music and music videos, sports programming, action films, and male-oriented magazines for both intentional and inadvertent sexual education.

A second explanation is that TV viewing may, in fact, be associated with males' sexual attitudes and expectations, but at earlier points in their lives. Perhaps by this age (i.e., 19 and 20), adolescent males have already picked up a great deal of the culture's messages about male sexuality, and are now less affected by additional depictions and images. Indeed, it is difficult to assess the cumulative effect on males of years of exposure to a mass culture that sells the message that masculinity is equated with being sexual, and that "being a player" is something to aspire to. Longitudinal research that follows and tests a group of males through adolescence is needed.

Also worthy of further examination are other predicted outcomes that did not materialize. For example, whereas previous reports have linked heavier viewing with holding more traditional gender role attitudes (e.g., Morgan, 1982; Signorielli, 1989), there was little association here between any TV variables and participants' gender role attitudes. It is possible that the one measure of gender role attitudes employed here was too short or too crude to adequately tap the complex, evolving, and nuanced attitudes of college students. Indeed, most of the students disagreed with the traditional attitudes represented in the scale, as indicated by the truncated range of scores. Whereas scores for this measure potentially range from 12 to 48, the majority of the scores for both sexes fell in the lower third of this range, with no scores in the upper third. Therefore, it may be necessary to use a different measure in the future, or to use multiple measures.

A second unfulfilled expectation was the minor role played by the level of Active Viewing. Whereas all of the other Viewer Involvement variables contributed repeatedly, both in the initial correlational analyses and in the regression equations, this construct did not. Additional research is needed to investigate the utility and range of this construct. It is possible that the three dimensions of viewer activity incorporated in our measure--Selectivity, Involvement, and Use--are differentially predictive. Post hoc analyses exploring this possibility support this notion, with significant connections between the sexual outcome variables and the three dimensions of active viewing emerging for Active Use only. For females, Use correlated positively with traditional gender role attitudes, r (162) = .15, p = .05, and with endorsement of recreational attitudes about sex, r (163) = .22, p = .005. For males, Use correlated positively with endorsement of recreational attitudes about sex, r (132) = .17, p = .05, and with expectations of peer sexual experience, r (129) = .20, p = .02. These results indicate that tests of active viewing may need to focus specifically on viewers' intended use of the content, and less on the other dimensions.

Finally, results indicated that TV viewing was not a consistent predictor of adolescents' sexual behavior. Instead, sexual behavior was best predicted by demographics (males only), by students' attitudes about sexuality, and by their expectations about normative sexual behavior. The only contribution of TV viewing was that for males, perceiving TV's sexual situations as likely to happen in one's own life predicted more extensive sexual experience. However, TV exposure and involvement did significantly predict students' sexual attitudes and expectations. Our data, therefore, suggest that TV viewing may help shape young viewers' attitudes and expectations about sexual relationships, which, in turn, are some of the strongest predictors of their sexual behavior. It appears, then, that TV's depictions may be influential in framing a view of reality and in providing bases for social comparison. Television appears to set an agenda about what is important to relationships and to young people--sex--and gives the impression that "everyone is doing it." Those failing to live up to these standards may feel inadequate, when, in fact, it is the expectations that are distorted. Indeed, Baran (1976a) reported that undergraduates perceiving TV portrayals as real or perceiving TV characters as having more pleasure were less satisfied with their own status as virgins.

Although this study breaks new ground in exploring television's multi-faceted role in the sexual socialization process, it also has limitations which future work will need to address. First, the results presented here are correlational only, and do not allow causal inferences to be drawn. Consequently, experimental follow-up work is needed that examines the direct impact of viewing specific types of sexual messages. Also needed are studies that focus on these associations from the other direction, exploring in more detail the many factors--demographic, attitudinal, and experiential--that affect how television's sexual content is perceived. Existing work in this area has been provocative (e.g., Brown, White, & Nikopoulou, 1993), and additional projects are under way (Ward, Gorvine, & Cytron, in press). Clear answers will come only by examining these questions from both directions, employing both correlational and experimental formats.

Second, this study's assessments of some complex constructs were somewhat limited. For example, because sexual behavior was not the focus of the study, only one self-report measure was employed to assess it. Yet to adequately cover such a broad and diverse construct, multiple measures would be more appropriate, and, with such, stronger results may have been obtained. In addition, assessments of Perceived Realism, Relevance, and Identification were based on students' responses to four particular scenes. Whereas it was assumed that these clips were representative of the type of content adolescents frequently watch, this fact was not tested directly with this sample. Indeed, because this content was chosen for the participants, it is possible that we may have underestimated the strength and impact of viewers' personal connections with portrayals, for it is likely that participants may feel even more connected to portrayals they select for themselves.

Finally, we acknowledge the limitations of the sample, both of the participants and of their television diets. Because only older adolescents were tested here, future work needs to extend this paradigm to a younger sample, who may possibly be more vulnerable to media messages. Also, this study's assessments of TV exposure focused on network comedies and dramas. Because adolescents are exposed to a variety of genres, including music videos and talk shows, future studies may want to include these formats in their analyses.

In conclusion, this study offers up a unique approach to examining television's contribution to sexual socialization. By including dimensions of TV exposure other than viewing amount, we were able to uncover significant and diverse relationships between viewing and sexual outcomes. The findings reported here illustrate that TV's countless verbal and visual references to dating and sexual relationships are, indeed, associated with adolescents' own sexual attitudes and expectations. In many respects, it appears as if TV's sexual portrayals may help to shape adolescents' sense of what is normative and expected. Yet this contribution is problematic because TV's portrayals, while exciting and attractive, also tend to be unrealistic. The extent to which holding such expectations may lead to dissatisfaction with one's sexual experiences and to irresponsible sexual decision-making are valid concerns.

REFERENCES

Alexander, A. (1985). Adolescents' soap opera viewing and relational perceptions. Journal of Broadcasting and Electronic Media, 29, 295-308.

Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice Hall.

Bandura, A. (1994). Social cognitive theory of mass communication. In J. Bryant & D. Zillman (Eds.), Media effects: Advances in theory and research (pp. 61-90). Hillsdale, NJ: Erlbaum.

Baran, S. J. (1976a). How TV and film portrayals affect sexual satisfaction in college students. Journalism Quarterly, 53, 468--473.

Baran, S. J. (1976b). Sex on TV and adolescent self-image. Journal of Broadcasting, 20, 61-68.

Bauder, D. (1998, July 8). Cable ratings top networks. The Washington Post, p. D7.

Brown, J. D., & Newcomer, S. F. (1991). Television viewing and adolescents' sexual behavior. Journal of Homosexuality, 21, 77-91.

Brown, J. D., & Steele, J. R. (1995). Sex and the mass media. Menlo Park, CA: Kaiser Family Foundation.

Brown, J. D., White, A. B., & Nikopoulou, L. (1993). Disinterest, intrigue, resistance: Early adolescent girls' use of sexual media content. In B. S. Greenberg, J. D. Brown, & N. Buerkel-Rothfuss (Eds.), Media, sex, and the adolescent (pp. 177-195). Cresskill, NJ: Hampton Press.

Bryant, J., & Rockwell, S. C. (1994). Effects of massive exposure to sexually oriented prime-time television programming on adolescents' moral judgment. In D. Zillman, J. Bryant, & A. C. Huston (Eds.), Media, children, and the family: Social scientific, psychodynamic, and clinical perspectives (pp. 183-195). Hillsdale, NJ: Erlbaum.

Bryant, J., & Zillman, D. (1994). Media effects: Advances in theory and research. Hillsdale, NJ: Erlbaum.

Buerkel-Rothfuss, N. L., & Mayes, S. (1981). Soap opera viewing: The cultivation effect. Journal of Communication, 31, 108-115.

Buerkel-Rothfuss, N. L., & Strouse, J. S. (1993). Media exposure and perceptions of sexual behaviors: The cultivation hypothesis moves to the bedroom. In B. S. Greenberg, J. D. Brown, & N. L. Buerkel-Rothfuss (Eds.), Media, sex, and the adolescent (pp. 225-247). Cresskill, NJ: Hampton Press.

Calfin, M. S., Carroll, J. L., & Shmidt, J. (1993). Viewing music-videotapes before taking a test of premarital sexual attitudes. Psychological Reports, 72, 475-481.

Carveth, R., & Alexander, A. (1985). Soap opera viewing motivation and the cultivation process. Journal of Broadcasting and Electronic Media, 29, 259-273.

Courtright, J. A., & Baran, S. J. (1980). The acquisition of sexual information by young people. Journalism Quarterly, 1, 107-114.

Day, R. D. (1992). The transition to first intercourse among racially and culturally diverse youth. Journal of Marriage and the Family, 32, 749-762.

Dorr, A. (1986). Television and children: A special medium for a special audience. Beverly Hills, CA: Sage.

Fabes, R. A., & Strouse, J. (1987). Perceptions of responsible and irresponsible models of sexuality: A correlational study. The Journal of Sex Research, 23, 70-84.

Fox, G. L. (1977). Sex-role attitudes as predictors of contraceptive use among unmarried university students. Sex Roles, 3, 265-283.

Galambos, N. L., Peterson, A. C., Richards, M., & Gitelson, I. B. (1985). The Attitudes Toward Women Scale for Adolescents (AWSA): A study of reliability and validity. Sex Roles, 13, 343-354.

Gerbner, G., & Gross, L. (1976). Living with television: The violence profile. Journal of Communication, 26, 173-199.

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. 17-41). Hillsdale, NJ: Erlbaum.

Greenberg, B. S. (1988). Some uncommon television images and the drench hypothesis. In S. Oskamp (Ed.), Television as a social issue (pp. 88-102). Newbury Park, CA: Sage.

Greenberg, B. S. (1993). Race differences in television and movie behaviors. In B. S. Greenberg, J. D. Brown, & N. L. Buerkel-Rothfuss (Eds.), Media, sex, and the adolescent (pp. 145-152). Creskill, NJ: Hampton Press.

Greeson, L. E., & Williams, R. A. (1987). Social implications of music videos for youth: An analysis of the content and effects of MTV. Youth and Society, 18, 177-189.

Hawkins, R. P., & Pingree, S. (1982). Television's influence on social reality. In D. Pearl, L. Bouthilet, & J. Lazar (Eds.), Television and behavior: Ten years of scientific progress and implications for the eighties (Vol. 2, pp. 224-247). Washington, DC: National Institute of Mental Health.

Huston, A. C., Wartella, E., & Donnerstein, E. (1998). Measuring the effects of sexual content in the media. Menlo Park, CA: Kaiser Family Foundation.

Huston, A. C., & Wright, J. C. (1988). Mass media and children's development. In N. Damon (Ed.), The handbook of child psychology. (5th ed., Vol. 4, pp. 999-1058). New York: Wiley.

Kim, J., & Rubin, A. M. (1997). The variable influence of audience activity on media effects. Communication Research, 24, 107-135.

Levy, M. R. (1983). Conceptualizing and measuring aspects of audience `activity'. Journalism Quarterly, 60, 109-115.

Levy, M. R., & Windahl, S. (1984). Audience activity and gratifications: A conceptual clarification and exploration. Communication Research, 11, 51-78.

Metzler, C. W., Noell, J., & Biglan, A. (1992). The validation of a construct of high-risk sexual behavior in heterosexual adolescents. Journal of Adolescent Research, 7, 233-249.

Morgan, M. (1982). Television and adolescents' sex-role stereotypes: A longitudinal study. Journal of Personality and Social Psychology, 43, 947-955.

Morgan, M., & Shanahan, J. (1991). Television and the cultivation of political attitudes in Argentina. Journal of Communication, 41, 88-103.

Nakao, K., & Treas, J. (1992). The 1989 Socioeconomic Index of Occupations: Construction from the 1989 Occupational Prestige Scores (General Social Survey Methodological Report No. 74). Chicago: University of Chicago, National Opinion Research Center.

Perse, E. M. (1986). Soap opera viewing patterns of college students and cultivation. Journal of Broadcasting and Electronic Media, 30, 175-193.

Peterson, J. L., Moore, K. A., & Furstenberg, F. F., Jr. (1991). Television viewing and early initiation of sexual intercourse: Is there a link? Journal of Homosexuality, 21, 93-118.

Pleck, J. H., Sonenstein, F. L., & Ku, L. C. (1993). Masculinity ideology: Its impact on adolescent males' heterosexual relationships. Journal of Social Issues, 49, 11-29.

Potter, W. J. (1986). Perceived reality and the cultivation hypothesis. Journal of Broadcasting and Electronic Media, 30, 159-174.

Roberts, E. J. (1980). Television and sexual learning in childhood. In D. Pearl, L. Bouthilet, & J. Lazar (Eds.), Television and behavior: Ten years of scientific progress and implications for the eighties (pp. 209-230). Rockville, MD: U. S. Department of Health and Human Services.

Rouner, D. (1984). Active television viewing and the cultivation hypothesis. Journalism Quarterly, 61, 168-174.

Rubin, A. M. (1984). Ritualized and instrumental television viewing. Journal of Communication, 34, 67-77.

Rubin, A. M. (1994). Media uses and effects: A uses-and-gratifications perspective. In J. Bryant & D. Zillman (Eds.), Media effects: Advances in theory and research (pp. 417-436). Hillsdale, NJ: Erlbaum.

Rubin, A. M., & Perse, E. M. (1987). Audience activity and television news gratifications. Communication Research, 14, 58-84.

Schwartz, P., & Rutter, V. (1998). The gender of sexuality. Thousand Oaks, CA: Pine Forge Press.

Signorielli, N. (1989). Television and conceptions about sex roles: Maintaining conventionality and the status quo. Sex Roles, 21, 341-360.

Signorielli, N. (1991). Adolescents and ambivalence toward marriage: A cultivation analysis. Youth and Society, 23, 121-149.

Strouse, J. S., & Buerkel-Rothfuss, N. L. (1987). Media exposure and the sexual attitudes and behaviors of college students. Journal of Sex Education and Therapy, 13, 43-51.

Strouse, J. S., Buerkel-Rothfuss, N. L., & Long, E. C. (1995). Gender and family as moderators of the relationship between music video exposure and adolescent sexual permissiveness. Adolescence, 30, 505-521.

Sun, S.-W., & Lull, J. (1986). The adolescent audience for music videos and why they watch. Journal of Communication, 36, 115-125.

Truglio, K. T. (1990). The role of prime time television viewing in the sexual socialization of adolescents. Unpublished doctoral dissertation, University of Kansas.

Udry, J. R., & Billy, J. O. (1987). Initiation of coitus in early adolescence. American Sociological Review, 52, 841-855.

Ward, L. M. (1995). Talking about sex: Common themes about sexuality in the prime-time television programs children and adolescents view most. Journal of Youth and Adolescence, 24, 595-615.

Ward, L. M., Gorvine, B., & Cytron, A. (in press). Would that really happen? Adolescents' perceptions of sexual relationships according to prime-time television. In J. Brown, K. Walsh-Childers, & J. Steele (Eds.), Sexual teens, sexual media. Hillsdale, NJ: Erlbaum.

Whitley, B. E., Jr. (1988). The relation of gender-role orientation to sexual experience among college students. Sex Roles, 19, 619-637.

Manuscript accepted April 14, 1999

This research was supported by a Postdoctoral Research Fellowship from the National Science Foundation. An earlier version of this paper was presented at the 1998 biennial meeting of the Society for Research in Adolescence, San Diego, California, and at the 1998 meeting of the Society for the Scientific Study of Sexuality, Los Angeles, California. The authors would like to thank Adena Cytron, Dr. Aimee Dorr, Katie Eschwege, Maya Gordon, and Ben Gorvine for their assistance with various aspects of this project, as well as Dr. Jane Brown for her insightful comments.

Address correspondence to L. Monique Ward, Psychology Department, University of Michigan, 525 E. University Avenue, Ann Arbor, MI 48109-1109; email: [e-mail].