The Association between Television Viewing and College Students’ Perceptions, Attitudes, and Behaviors Regarding Alcohol Use
Linda Godbold Kean and Kelly Fudge Albada
[WJMCR 5:2 March 2002]
Abstract|Introduction|Research Study|Method|Results|Conclusion
Abstract
Past content analyses indicate that alcohol consumption on television is frequent and glamorized. Moreover, previous research confirms a negative influence on adolescent viewers, though the evidence for college students is mixed. This study involved a survey designed to investigate college students’ television viewing habits and their relationship with alcohol. Total television viewing failed to predict the outcome variables, yet viewing dramas (particularly life dramas) was associated with lower estimations of alcohol use by peers, more negative attitudes, less frequent drinking behavior, and reduced intentions to drink. Viewing female-oriented programs was linked with higher estimations of peer use, more positive attitudes, more frequent drinking, greater intentions to drink, and more alcohol-related health consequences.
Introduction
While college administrators and faculty seek to improve the education of the students attending their schools each year, one particular problem stands in the way of providing many students with the best experience possible. That problem is alcohol use. The Chronicle of Higher Education recently reported that alcohol arrests and violations increased by 24% from 1998-1999 on 481 college campuses with more than 5,000 students.1 This high rate of alcohol consumption during the college years has powerful effects on the health and social reality of the students. One of the major health risks when young people use alcohol is death. According to the CDC (2000), alcohol and other drug use constitute one of the primary types of behaviors that contributes to morbidity and mortality of individuals 5-24 years old. Presley, Meilman, and Lyerla found that due to the influence of alcohol, (a) 15% of students reported participation in acts of sexual misconduct, (b) 23% indicated poor performance on a major test or project, (c) 14% had trouble with the police or campus authorities, and (d) one-third admitted to arguing or fighting.2
Given the severity of this problem, researchers, educators, and parents have queried the causes behind college students’ excessive alcohol use. Although peer pressure is a well-documented influence on students’ alcohol use, televised portrayals of alcohol use, and its effects may greatly inform students’ beliefs and attitudes about the substance. Hence, the goal of this research project was to consider the association between exposure to these television portrayals and young people’s relationship with alcohol. Most content analyses and effects studies classify content genres according to production categories (e.g., comedies, dramas, news), yet program selection may cluster along lines other than those typically considered (e.g., youth comedies, legal dramas). This study asked students what they watched, analyzed the data for clusters, and looked for a relationship between clusters and the dependent variables.
Theoretical Perspective
Behind much of the research on television images is the assumption that the pervasive and persistent portrayal of certain images leads to greater learning and acceptance of those images by viewers.3 This is especially troubling when the portrayals are undesired images and are accompanied with anti-social behaviors (e.g., binge drinking, teenage sex). Cultivation describes the “independent contributions television viewing makes to viewer conceptions of social reality.”3Rather than positing a direct, immediate effects model, the cultivation perspective accounts for a slow, cumulative model. Put simply, if viewers are repeatedly exposed to certain images, they come to consider these images normative and thus, are more likely to behave in line with these norms at some point. Gerbner argues that heavy television viewers perceive the real world to be more like the television world than light viewers do.4 Cultivation theory emphasizes that it is the amount of television that one is exposed to that determines effects. For instance, in investigating television violence, cultivation theorists maintain that there is little variety in television content, so the viewing of specific content is unimportant. As Gerbner and Gross emphasized, “The system as a whole plays a major role in setting the agenda of issues to agree or disagree about…only system wide analysis can reveal the symbolic world which structures common assumptions.”5 Gerbner and Gross further argued that the relationship of characters and events cannot be investigated by examining a single program or genre.5
Although there is significant evidence for cultivation, one of its basic assumptions has been questioned. Despite Gerbner’s initial arguments about the uniformity of messages on television and the equal treatment of content in research, a great deal of cultivation research actually supports content specific effects. Hawkins and Pingree investigated whether different program genres have a varying influence on beliefs about violence in our society.6 Their analysis of television diaries and questionnaire responses from children revealed that genres were differentially related to beliefs about social reality. While viewing of crime shows, cartoons, and game shows were significantly correlated with beliefs about violence in society, viewing news, sports and drama were not related to the children’s beliefs. Furthermore, the effects for separate genres were stronger than the effect for overall viewing. Content-specific effects were also found with a college population. Armstrong, Neuendorf and Brentar determined that the viewing of specific content affected white students’ racial beliefs.7 Students who viewed a great deal of TV news estimated the socio-economic status of blacks lower than students who watched little TV news did. Moreover, students who frequently viewed entertainment television perceived the socio-economic status of blacks as higher than students who viewed entertainment television infrequently. Finally, Strouse and Buerkel-Rothfuss gathered questionnaire data from almost 500 college students regarding sexual attitudes and behaviors.8 For females, the only viewing measure significantly related to the attitudinal and behavioral measures was amount of exposure to MTV. Those females who watched MTV more often had more liberal sexual attitudes and reported a greater number of sexual partners. No other viewing measures (soap operas, comedy, drama, action/sports, and overall viewing) were significantly related to their attitudes and behaviors. For males, no viewing measures were correlated with sexual attitudes, but the number of partners was positively related to the amount of soap opera viewing. Total viewing time was not associated with the dependent variables.
Hence, it appears that content specific effects occur and may be an equal, if not more valid, way of studying effects over across-the-board television viewing. The follow-up questions become how does television portray alcohol use and what are the effects of these portrayals.
Television and Alcohol Use
TV Portrayals of Alcohol. Breed, DeFoe and Wallack examined alcohol in the mass media from 1976-82.9 The investigators examined over 600 hours of the top 15 (as rated by Nielsen) television situation comedies and one-hour dramas. They found approximately seven acts of alcohol consumption or preparation per hour. In fact, alcohol was the most consumed beverage on television, followed by tea and coffee, water and then soft drinks. According to the authors, the actual consumption of beverages in the U.S. occurs in the reverse order.
In a later study, an increase in portrayals of alcohol consumption was observed. Wallack, Grube, Madden and Breed’s analysis of 1986 fictional prime time television revealed that 64% of program episodes contained a reference to alcohol and actual ingestion was portrayed in 50% of all episodes.10 In total, the authors found over eight alcohol acts per hour. Consumption occurred most frequently during television dramas and made-for-television movies. Within the category of dramas, twice as many incidents occurred in evening soap operas over police dramas. When Cheers was included in the analysis, situation comedies had more than nine acts per hour.
Recent content analyses confirm these earlier findings. Mathios, Avery, Bisogni, and Shanahan looked at 224 hours of programming from the 1994-1995 television season.11 Of all food and drink categories, alcohol episodes were depicted most frequently. Christenson, Henriksen and Roberts looked at 168 episodes of top rated prime time programs from the 1998-1999 television season.12 The authors found that alcohol was mentioned or used in 71% of all the episodes coded. Godbold and Albada looked at acts of alcohol activity in prime time entertainment programming in the spring of 1999. While alcohol activity was abundant in itself, the authors also coded particular types of alcohol activity. The most frequently coded alcohol related act was holding an alcoholic drink. The second most frequent activity was alcohol being near a character, followed by consumption as the third most frequent activity.13 Out of nine hours of comedies coded in this study, the researchers found 287 acts of alcohol activity; 58% of all acts. Dramas accounted for 30% of the alcohol acts. Fantasy shows (e.g. X-Files) had 11% and Crime shows had 2% of the alcohol acts.
In addition to alcohol portrayal frequencies, the characteristics of those who drink are important. Viewers are more likely to see a male character drink alcohol as compared to a female character.9, 11 However, Wallack et al. noted that out of all characters presented, the percentage of female characters that drank was slightly higher than that for males.10 Approximately 19% of the female characters consumed alcohol, while 15% of the male characters drank alcohol. In other words, overall there are more male characters on television, so a viewer will witness more males drinking. However, out of the total number of characters, a higher percentage of females drink than do males.
Regarding the occupation or socio-economic status of characters associated with alcohol activity, drinkers tended to be professionals. Breed et al. reported that very few poor people drank; rather, those who drank tended to be “doctors, lawyers, executives, detectives and middle-class housewives”.9 Likewise, the Wallack et al. study found that 22% of professionals and 40% of upper class characters, as compared to 10% of service workers and 8% of poor characters, drank alcohol.10Mathios et al. verified the pervasiveness of the high-income drinker (48%) and added frequency information on the middle-income drinker (34%).11 When Godbold and Albada looked at occupation, most of those individuals associated with alcohol were white-collar workers.13
Age is another characteristic that is often investigated in content analyses of alcohol use on television. Researchers have consistently found that few young characters (under 21) drink, and when they do, they often regret the activity or are otherwise portrayed in a negative light.9, 10, 11, 12, 13 The age group most likely to drink in the Wallack et al. study was 30-39 year olds, with 22% of those characters drinking.10 Godbold and Albada again mirrored these outcomes. Only 2.2% of the acts in that study were coded as involving those under 21. Characters coded as 21-50 represented 81% of the acts and those over 50 were responsible for 13% of the acts.13
Overall valence of the depictions was investigated by Wallack et al who had coders indicate whether acts were attractive, unattractive or neutral.10 For all drinkers, 62% of portrayals were neutral; 26% were attractive; and only 12% of the alcohol portrayals were unattractive. In other words, 84% of the incidents were portrayed as free from problems. Christenson, Henriksen and Roberts also looked at overall tone of the portrayal and found that 40% of the episodes made alcohol use seem to be a positive experience while only 105 of the episodes gave use a negative tone.12 Further, nearly half of all episodes that depicted alcohol use did so in a humorous context. Godbold and Albada also coded for tone and had similar findings. Tone was coded as positive, negative and neutral. In total, 131 acts, or 26% were coded as positive. Thirty-three acts,7% were coded as negative, and 334 acts or 67%, were coded as neutral.13 There was additional information in the Godbold & Albada study regarding gender and valence. Women were significantly more likely to be involved in scenarios with a positive or neutral tone. 57% of the alcohol acts involving female characters were coded as neutral and 37% were coded as positive code. Only 6% were negative.13 Finally, Godbold and Albada also found an interesting pattern regarding valence and program genre. There was a significant relationship between show type and tone. Dramas accounted for 94 % of the negative acts and fantasy shows accounted for the other 6%. However, there were no negative portrayals of alcohol in comedies. Comedies accounted for 77% of all positive acts, dramas, 21% and fantasy shows were at 2%.13
In general, these studies suggest the continued glamorization of alcohol. First, use is frequent and tends to be engaged in without negative or serious consequence. Moreover, characters who drink tend to be professionals, of high socio-economic status, and of legal drinking age. While alcohol is portrayed in both comedies and dramas, comedy is often the backdrop for alcohol use. The implicit assumption of content analysis studies is that viewers attend to, learn, and imitate the frequently appearing and rewarded depictions.14 Previous effects research supports this assumption.
Perceptions, Attitudes and Behaviors. Much of the research on alcohol portrayals and their associations with perceptions, attitudes and behaviors focuses on an adolescent sample. For pre-college aged students, the evidence consistently supports an association. In a survey of young males, Tucker found that, even when controlling for demographic variables, heavy television viewers consumed significantly more alcohol more often that light viewers did.15 Connolly, Casswell, Zhang, and Silva conducted a longitudinal study relating exposure to entertainment portrayals of alcohol use at ages 13 and 15 with actual consumption at age 18.16For females, heavy television viewing at ages 13 and 15 was related to higher average and maximum amounts of beer, wine and spirits that they drank at age 18. For males, no significant relationship between viewing at ages 13 and 15 and consumption at 18 was revealed. Godbold found a correlation between television viewing and perceptions, attitudes, and behavioral intentions among sixth graders.17 Students who watched more TV believed that more of their peers thought that drinking at their age was acceptable. In addition, students who watched more television held more positive attitudes toward alcohol use and had greater intentions to drink in the future than students who watched less television. These patterns were further supported for the viewing of television dramas, after controlling for overall viewing.
When considering college students and the relationship between viewing alcohol-related content and outcome measures of interest, the evidence is mixed. Bahk investigated the presence versus absence of negative consequences of alcohol use in a movie.18 Students were assigned to one of three viewing conditions. In the first condition, participants viewed a segment of a movie that included negative and positive portrayals of alcohol use. For the second condition, the negative scenes were edited out, so the version depicted only enjoyment of alcohol. The final group saw the segment with all alcohol scenes removed (neutral). Responses to questionnaire items after viewing indicated that those who saw the segment with the negative consequences held the least favorable attitudes towards alcohol use. Participants in the neutral condition held slightly favorable attitudes, and participants in the positive condition held the most favorable attitudes. These results suggest that college students are, in fact, affected by alcohol portrayals.
Nevertheless, other studies refute this conclusion. In an experimental design, Sobell, et al. presented 96 college males with televised advertisements that either promoted a brand of beer, a non-alcoholic beverage, or food.19 These ads were embedded within an episode of the program Dallas. The program also contained seven alcohol-related cues. In half of the conditions, the alcohol-related scenes and ads were removed from the program. After viewing, the participants were asked to do a second, supposedly unrelated study involving beer tasting. The authors found no correlation between the amount of beer consumed in the taste test study and the viewing condition. Though the artificial setting of the laboratory may have influenced the students’ drinking, this study does not support an effect of one time exposure to alcohol content.
While the results for college students are mixed, some support for the association between television viewing and alcohol attitudes and behaviors exists. One focus of the current investigation was to further investigate the link between viewing and alcohol-related issues in a college sample and to consider habitual viewing versus a one-time exposure to the material. As argued above, cultivation theory provides the justification for examining the effects of repeat exposure to televised images.
Summary. The research to date leads to several conclusions. Regarding content, alcohol use on television is frequent. This use is more prevalent among high status individuals and is also relatively free of negative consequences. Though effects studies have found an association between viewing and outcome measures for adolescents, the evidence regarding college students is unclear but suggests influences on attitudes. Cultivation theory indicates potential relationships between exposure to TV images and perceptions, attitudes, and ultimately, behaviors. Tests of this theory have added the caveat that specific content must be considered along with overall TV viewing.
Research Study
Based on the above study, and previously reviewed content analyses from the last two decades, the following hypotheses were posited: Due to the overall abundance of portrayals, and the problem-free nature of the portrayals, our first hypothesis involves general television viewing.H1: Overall television viewing will be positively associated with estimations of peer alcohol use (perceptions), attitudes toward use, present use, and likelihood of future use.
Past content analyses have found that dramas tend to portray more negative consequences as a result of alcohol use. Based on those findings and the nature of the genre some content-specific associations are predicted.H2: Dramatic television viewing will be negatively associated with estimations of peer alcohol use, attitudes toward use, present use, and likelihood of future use.
Comedies tend to be light hearted in nature as a rule. Additionally, alcohol use in these programs tends to be more humorous. Due to this pattern, comedies are expected to show a pattern opposite that of dramas.H3: Comedic television viewing will be positively associated with estimations of peer alcohol use, attitudes toward use, present use, and likelihood of future use.
Finally, because past analyses have found that a greater percentage of females characters drink and these portrayals tend to be more positive in tone, viewing programs with major female characters should be associated with the outcome variables of interest.H4: Viewing programs with major female characters will be positively associated with estimations of peer alcohol use, attitudes toward use, present use, and likelihood of future use.
Method
Participants. Two hundred college students from a mid-sized southeastern university participated in this study. The students were recruited from several communication courses and received extra credit for their participation. The sample was comprised of 136 females and 62 males (two participants did not list their sex), with an average age of 21 years old (range 18-36). In terms of racial composition, the sample was 87% White/Caucasian, 8% Black/African American, and 5% Hispanic/Mexican-American, Asian/Asian-American or other.
Instrumentation. Measures included television viewing patterns, estimates of peer alcohol use, the participant’s attitudes toward use, past/current behaviors and behavioral intentions.
Television viewing. A variety of television viewing measures were used in this study. Students indicated the number of hours that they watched television on an average weekday, during an average week, and on the previous day. One variable was computed for overall television by summing these three items. The alpha reliability of this measure was .84.
Participants were also given a list of specific programs and were asked to indicate whether they watched these shows never, sometimes, often, or very often. The top 20 network prime time entertainment programs among teens and adults as ranked by Nielsen from 2-1-99 to 2-15-99 were included in this study (See Table 1). Factor analysis was conducted to determine if the 20 television programs were associated with each other in a particular way. Principal components analysis with varimax rotation yielded a six-factor solution (See Table 2 for factors and factor loadings). Initially, in keeping with other mass media research studies, all factors had an eigenvalue of 1.0 or higher, and items were considered part of a factor if they loaded at .50 or higher. Additional decisions were made when looking at the face validity of the factors produced. Factor one, labeled Comedies included Dharma and Greg, Just Shoot Me, Frasier, Drew Carey, and Veronica’s Closet. Factor two, female-oriented programs, included Beverly Hills 90210, Ally McBeal, Party of Five, Jesse and Friends. These programs were considered female-oriented because they include a number of strong, successful female lead characters. Jessealso loaded high on the comedy factor. It loaded slightly higher on the female-oriented factor, and due to the similarity with the other programs in this factor, it was determined that Jesse should be considered a part of this program type. Factor three, Police dramas, included: NYPD Blue and Law and Order. The fourth factor appeared to consist of male-oriented programs, with Everybody Loves Raymond, Becker, and JAG. The lead characters in these shows are adult males in their 30s, 40s and 50s. The fifth factor was named Life dramas for their focus on serious life situations. This factor included ER, Providence, and Touched by an Angel. X-Filesdid not load high on any factor. Its highest loading was on the final factor at .39. It was dropped from the analysis. Home Improvement was also eliminated because it only loaded highly on the final factor and did not seem to represent a meaningful factor as a single item. Additional analyses included the first five factors. Alpha reliabilities for the factors were.76 for female-oriented programs, .68 for police dramas, .59 for male-oriented programs, .70 for comedies and .52 for life dramas. For the program viewing factors (1=never, 2=sometimes, 3=often, and 4=very often), means were computed. Means were 1.59 for comedies, 2.03 for female-oriented programs, 1.40 for police dramas, 1.37 for male-oriented programs, and 1.54 for life dramas. This six-factor solution explained 60% of the variance in these items. The five factors used in the analyses explained 55% of the variance.
Perceptions of alcohol use. Participants indicated what percentage of their friends and what percentage of their peers drank alcohol (0-100%). Participants estimated that an average of 76% of their friends and 80% of their peers drink alcohol.
Attitudes towards alcohol use. Attitudes were measured using semantic differentials on a seven-point scale. Participants responded to the statement, “Drinking alcohol at my age is�”. Word pairs included harmful/not harmful, unsafe/safe, bad/good, wrong/right, immoral/moral, risky/not risky, and stupid/not stupid. The attitudes toward alcohol use index yielded an alpha coefficient of .90. The mean for the attitude measure was 3.80 on the seven-point scale, suggesting moderate attitudes towards alcohol.
Past alcohol behaviors. There were four behavioral measures used in this study. Participants were asked how many days within the past month that they drank alcohol, how many times in the past month that they had been drunk, how many times in the past month that they had more than five drinks in one night, and how many times in the last month that they suffered health consequences from drinking. Examples given for this final question were vomiting, hangover, or injury. Response options for the above items ranged from 0, 1-2, 3-4, 5-6, 7-8, 9-10, or 11 or more. The reliability for the behavior measure was computed based on the first three items, and an alpha of .92 was produced. The mean for the behavior items was 3.13, indicating 3-4 instances of drinking, being drunk, and having more than five drinks in one night during the previous month. The health item was considered separately because this item seemed to tap a separate, more severe behavior. The mean for suffering health consequences in the past month was .66. This mean suggested that participants suffered health consequences from alcohol use an average of 1-2 times within the past month.
Behavioral intentions for alcohol use. For behavioral intentions, participants were asked to rate their likelihood of drinking alcohol this weekend, drinking alcohol this week (Sunday-Thursday), getting drunk this weekend, getting drunk this week (Sunday-Thursday), drinking alcohol this month, and getting drunk this month on a five-point scale (highly unlikely to highly likely). An alpha of .91 emerged for this six-item index. The mean for behavioral intentions on a five-point scale was 3.40, suggesting that participants were fairly likely to drink in the near future.
Results
Because there were more females than males in the sample, a comparison of means was computed to see if the two groups differed significantly on any of the variables. The two groups did not differ significantly on any of the dependent variables, but they did differ on three viewing measures. These measures were the amount of viewing for the life dramas, female-oriented programs, and male-oriented programs, with females watching significantly more of the first two genres and less of the latter genre. These differences lent support to the factor interpretation; however, they also necessitated the inclusion of sex as a covariate in subsequent analyses.
Regression analyses were used to test the four hypotheses. Sex was entered on step one, and the independent variables were entered on step two of the analyses. Sex did not emerge as significant predictors for any of the dependent variables.
Overall Viewing. H1 posited that overall viewing would be positively associated with perceptions, attitudes, behaviors, and behavioral intentions. The three item viewing measure was entered in step two of the regression equation. Overall viewing did not emerge as a significant predictor for any of dependent measures; therefore, H1 was not supported.
Content Specific Viewing. To test H2, H3 and H4, the program variables from the factor analysis were entered as a block in step two of the regression equation. Because the factor analysis broke programs down a bit differently than the created hypotheses, the comedy factor was used to test the comedy viewing hypothesis (H3), while viewing of life dramas and police dramas were both used to test the drama viewing hypothesis (H2). Viewing of female-oriented programs was used to test H4. Male-oriented program viewing was entered in the block as well, although it was not considered a direct test of any of the hypotheses. The R2 change for the block is reported below, as well as the standardized coefficients for each of the program factors. Table 4 reports all significant analyses.
The first dependent variable tested was estimations of other students’ alcohol use. For this variable, three of the five viewing variables were significant. Viewing of life dramas, police dramas and female-oriented programming was associated with the dependent variable and accounted for an R2 change of .15 (F=6.06, df = 5, 182, p<.001). For viewing of life dramas, the relationship was negative; as viewing increased, estimations of peer use decreased ( = -.27, p<.001). Police dramas were also negatively associated with estimations of peer use ( = -.20; p <.05). These results support H2 in which we hypothesized that viewing of dramas would be negatively associated with estimations of peer alcohol use. Viewing of female-oriented programs was positively associated with estimations of peer use; as viewing increased so did estimations ( = .30, p<.001). Hence, H4 was confirmed for the perception outcome variable.
The next dependent variable tested was estimations of friends’ alcohol use. For this variable, only two of the viewing factors emerged as significant predictors. Viewing life dramas and female-oriented programs accounted for an R2 change of .07 (F=2.57, df = 5, 182, p<.05). Again, the results lent support to both H2 and H4. For life dramas, the relationship was negative ( = -.25, p<.01); as viewing increased, estimations decreased. For viewing of female-oriented programming, the relationship was positive, as viewing increased so did estimations of friends’ alcohol use ( =.22, p=.01). Because comedies did not emerge as a predictor, H3 was not confirmed for the perception variables.
In terms of attitudes toward alcohol use, only viewing of female-oriented programs surfaced as a significant predictor in the analyses, accounting for 8.5% of the variance (F= 2.84, df = 5,182, p<.01). Viewing female-oriented programming was positively related to attitudes towards alcohol use. As viewing increased, attitudes toward alcohol use became more positive ( = .18, p<.05). These findings confirmed H4.
The viewing of female-oriented programming and life dramas were also significantly related to the behavioral index. These two genres accounted for 10% of the variance, (F=3.74, df = 5, 182, p<.005). Viewing of female-oriented programs was again positively associated with the outcome variable ( = .31; p<.001), such that participants who watched more female-oriented programming were more likely to consume alcohol. On the other hand, the viewing of life dramas was negatively associated with alcohol use ( = -.24, p<.005). These results further support H2 and H4. An additional behavioral measure in this study was the frequency of health consequences experienced due to drinking alcohol in the prior month. For this variable, only viewing female-oriented programs emerged as a significant predictor. As viewing of this genre increased so did the number of alcohol-related health consequences ( = .32, p<.001), accounting for 9% of the variance (F= 2.967, df = 5,182, p<.01). This result supports H4.
The final variable of interest concerned intentions to drink in the future. Two viewing variables surfaced as predictors for this dependent variable, viewing of female-oriented programs and life dramas (F= 3.57, df = 5, 182, p<.005) which accounted for 9% of the variance. Greater female-oriented program viewing was related to stronger intentions to drink in the future ( = .31, p<.001), while greater life drama viewing was associated with weaker intentions to drink in the future ( = -.22, p<.05). Hence, H2 and H4 are supported.
Conclusion
This study was concerned with the association between exposure to television and college students’ perceptions, attitudes and behaviors regarding alcohol use. An across-the-board association between total amount of time spent watching TV (regardless of program type) and the dependent variables of interest was not evidenced. However, in terms of the specific programs in the Nielsen top 20, there was fairly strong support for hypotheses regarding life dramas and female-oriented programming.
For viewing of life dramas, the results consistently supported a relationship between viewing and the dependent variables in the hypothesized direction. The more that participants watched life dramas, the less that they believed their peers and friends drank, and the less likely they were to either be drinkers or intend to drink alcohol in the future. The programs in the life drama category would certainly lead to this conclusion. Touched by an Angel, Providence and ER tend to present the negative side of alcohol use. In fact, they tend to focus on overuse. So, the viewer does not get a picture of everyday use among a variety of individuals but is told a story of some people who use to excess with negative consequences. So, although there is alcohol activity in these dramas, it is negative in tone when it is present.
Regarding a different type of dramatic programming, police dramas, the pattern was less consistent. Viewing of police dramas was only associated with lower estimations of peer alcohol use. This pattern among dramas is not too surprising based on past research. As reviewed earlier, Wallack, Grube, Madden, and Breed found a great deal more alcohol use in evening soap operas than in police dramas.10 Godbold and Albada found that both dramas and fantasy shows contained more alcohol activity than police dramas, although both dramas and police programs tended to have negative images of use.19 Overall, results suggest that all dramas may not behave in the same way, and thus, researchers may want to consider the overall message and tone of the programming, as well as their population’s frequency of viewing, before employing unilaterally the term “dramas” in their investigations.
For the female-oriented shows, the relationship between viewing and the dependent variables was remarkably consistent. Viewing was associated with higher estimations of peers’ and friends’ use, more positive attitudes towards alcohol, more frequent alcohol consumption and alcohol-related health consequences, and stronger intentions to use alcohol in the future. When bright, capable women are portrayed as using alcohol frequently, with few negative consequences, it seems to be associated with the idea that many people in society drink to this level with few negative consequences. In contrast to the female characters, however, the female viewers appeared to experience negative alcohol-related heath consequences. When we consider the characters on the programs, we see that the characters tend to be strong, professional women. On Ally McBeal, there are five characters appearing regularly who are lawyers and one character who is a judge. The women on Party of Five, Friends, Jesse and Beverly Hills 90210 have occupations ranging from fashion designer to head chef to high achieving college students. If these young, professional women are portrayed consuming alcohol, they may be sending a message of alcohol use as normative, and thus be associated with higher estimates of use and more positive attitudes towards use. Yet, these programs may distort reality somewhat because the viewers experience more negative consequences than do the characters. Instead of these real-life health consequences steering female viewers away from unhealthy drinking behaviors and lowering their attitudes towards use, it may be that the viewers’ desire to be like the attractive, successful, high status female characters alters the viewers impression of the consequences of use. Of course, because this data is from a survey design we do not know in which direction casuality flows. It may be that those who have positive attitudes about drinking and who drink themselves, enjoy and tend to watch programming that reinforces this behavior. Alternatively, it may be that exposure to these messages creates a sense of use as normative and motivation to behave in a way similar to others.
In addition to confirming that particular content is differentially associated with our dependent variables, this study revealed that additional methods of classifying programs are necessary to support content specific claims. By using participant responses, we were able to create meaningful content categories beyond the typical comedy versus drama dichotomy. Some programs fell into a factor that was strictly comedic (e.g., Dharma and Greg). However, other comedies, such as Ally McBealand Friends, clustered more closely with female-oriented dramas, such as Party of Five and Beverly Hills 90210. Similarly, some male-oriented comedies and dramas did not associate with other comedies and dramas. Becker, Everybody Loves Raymond and JAG stood as their own factor. Dramas were divided among the female-oriented programs, police dramas, and life dramas factors. The latter tended to focus on the harsh realities that people may face and performed differently in the analyses than the female-oriented programs and police dramas.
Limitations
There are some areas in which this study could be improved to provide even stronger results. The most consistent association between television viewing and alcohol use perceptions, attitudes and behaviors was produced for the female-oriented programs. As stated in the method section, we had a disproportionate sample in terms of male and female participants. Just over two-thirds of the sample was female. While sex was entered as a covariate and failed to significantly predict the variance in the dependent variables, a more balanced sample would provide a stronger test of our hypotheses. This imbalance may also have contributed to the somewhat low reliabilities for the life drama and male-oriented viewing factors (.52 and .59, respectively). If more males had participated in the study, we might have seen a greater role, or a different pattern emerge for, the male-oriented programs. Additionally, this data comes from one campus. A larger study, which gathered participants from many regions of the country could provide for more clear patterns of results.
Implications
Portrayals of alcohol use on television are abundant and sometimes inaccurate. Those who drink on television do so often, but in moderation. Men are seen drinking more but when women drink it tends to be more positive. Only those of legal drinking age do so on a regular basis. The problem with these portrayals is that they do not accurately represent the reality of alcohol use. In reality young people drink frequently, often with serious, even fatal, consequences. Adults also can drink beyond a reasonable extent and in an inappropriate manner. When the picture on television does not match that of reality, individual perceptions are affected. Young people may see the large numbers of portrayals of alcohol on television that rarely result in problems and may adopt a particular attitude or even behavior with regards to alcohol use.
The investigation reviewed here contributes to the existing literature in two ways. First, the data support the hypotheses that viewing is related to perceptions, attitudes, behaviors, and behavioral intentions for a college-aged population. While media scholars and educators tend to focus their efforts on increasing the media literacy of younger populations, continued efforts to enhance the media literacy levels of college students are needed. Communication and public health professionals in the university setting could aid college students by focusing on the often-unrealistic nature of certain portrayals regarding alcohol consumption. It benefits all students to become critical consumers of media messages, whatever the topic of those messages may be.
In addition, this study contributes to cultivation theory research by providing additional support for content specific associations with variables of interest. The type of programming, characteristics of the alcohol consumers, and tone of the drinking situation emerged as critical factors. This research offers further evidence that we cannot consider the issue of cultivation without acknowledging specific content. Viewing of life dramas was negatively associated with the dependent measures, while female-oriented programming had a positive association. In more dramatic, even medical fare, the portrayal of alcohol use is often negative, while female-oriented programs, with their strong female lead characters, presented a completely different picture of alcohol use. By using participants’ responses regarding viewing to create program factors, we were able to gather some data about programming choices and effects that might have been revealed if the traditional categories were employed. This technique moves us beyond the typical content groupings to ones that resonate with the viewers themselves.
Table 1: Nielsen ratings for network prime time entertainment shows
| 1. | 1 | ER | NBC |
| 2. | 2/3 | Frasier | NBC |
| 3. | 3/2 | Friends | NBC |
| 4. | 6 | Touched by an Angel | CBS |
| 5. | 7/5 | Veronica�s Closet | NBC |
| 6. | 8/4 | Jesse | NBC |
| 7. | 10/12 | NYPD Blue | ABC |
| 8. | 11 | Everybody loves Raymond | CBS |
| 9. | 12 | X-Files | FOX |
| 10. | 13 | Dharma and Greg | ABC |
| 11. | 15 | Home Improvement | ABC |
| 13. | 18 | Just shoot me | NBC |
| 14. | 19 | Drew Carey | ABC |
| 15. | 20 | Providence | NBC |
| 16. | 20 | JAG | CBS |
| 17. | 19 | Ally McBeal | FOX |
| 18. | Law and Order | CBS | |
| 19. | Party of Five | FOX | |
| 20. | BH 90210 | FOX |
(19. and 20. added due to ranking on Teen List)
Table 2: Factor Loadings for Television Programs Used in Regression Analyses
| Factor | Eigenvalue | factor 1 | factor 2 | factor 3 | factor 4 | factor 5 | factor 6 |
| Comedies | 4.186 | ||||||
| Dharma and Greg | .54 | .10 | .03 | -.10 | .36 | .36 | |
| Frasier | .69 | .14 | .29 | .09 | -.04 | .07 | |
| Drew Carey | .51 | -.38 | .13 | .29 | -.07 | .40 | |
| Just shoot me | .70 | .09 | .03 | .21 | .01 | -.19 | |
| Veronica�s Closet | .66 | .17 | -.08 | .07 | .35 | -.03 | |
| Female-Oriented | 2.47 | ||||||
| Beverly Hills 90210 | .02 | .70 | .10 | .07 | .24 | .03 | |
| Ally McBeal | .20 | .73 | .09 | -.07 | -.02 | -.06 | |
| Party of Five | -.04 | .78 | .16 | .08 | .17 | .01 | |
| Friends | .37 | .58 | -.07 | -.13 | .03 | .29 | |
| Jesse | .52 | .52 | -.05 | .06 | -.05 | .13 | |
| Police Dramas | 1.71 | ||||||
| NYPD Blue | .08 | .08 | .80 | .18 | .11 | -.01 | |
| Law and Order | .02 | .08 | .84 | .06 | .06 | .02 | |
| Male-Oriented | 1.37 | ||||||
| Becker | .23 | .02 | .04 | .78 | -.05 | .10 | |
| JAG | -.18 | -.02 | .25 | .61 | .16 | -.09 | |
| Everybody loves | .35 | -.07 | .02 | .68 | -.06 | .23 | |
| Raymond | |||||||
| Life Drama | 1.18 | ||||||
| ER | .30 | .22 | .27 | -.13 | .60 | -.08 | |
| Touched by an | .04 | -.03 | -.03 | .16 | .77 | -.06 | |
| Angel | |||||||
| Providence | -.04 | .26 | .07 | -.07 | .51 | .18 | |
| Factor 6-unlabeled | 1.06 | ||||||
| X-Files | .24 | -.05 | .36 | -.04 | -.29 | .39 | |
| Home Improvement | -.11 | .16 | -.03 | .22 | .20 | .79 |
Table 3: Means for dependent variables and viewing variables.
| Total sample | Males | Females | ||
| Overall viewing (1-11 scale) | 2.75 | 2.98 | 2.64 | |
| Life Dramas (1-4 scale) | 1.54 | 1.30* | 1.65* | |
| Comedies (1-4 scale) | 1.62 | 1.63 | 1.61 | |
| Male-Oriented Programs (1-4 scale) | 1.37 | 1.58* | 1.28* | |
| Female-Oriented Programs (1-4 scale) | 2.02 | 1.56* | 2.23* | |
| Student use (1-100 scale) | 76.42 | 73.07 | 77.65 | |
| Friend use (1-100 scale) | 79.60 | 76.29 | 80.10 | |
| Attitudes (1-7 scale) | 3.80 | 3.85 | 3.80 | |
| Behaviors | 3.13 | 2.90 | 3.23 | |
| Health Consequences | .66 | .57 | .70 | |
| Behavioral Intentions | 3.40 | 3.16 | 3.49 |
* Means differ at the .05 level.
Table 4: Significant associations between television exposure and outcome variables.
| Viewing Variable | Dependent Variable | Beta |
| Life Dramas | ||
| Estimates of student use | -.27*** | |
| Estimates of friend use | -.25** | |
| Drinking behavior | -.24* | |
| Intentions to drink | -.22** | |
| Police Dramas | ||
| Estimates of student use | -.20** | |
| Female-Oriented | ||
| Estimates of student use | .30*** | |
| Estimates of friend use | .22** | |
| Attitudes toward use | .18+ | |
| Drinking behavior | .31*** | |
| Intentions to drink | .31*** | |
| Health consequences | .32*** |
*** p< or = .001
** p < or = .005
* p < or = .01
+ p < or = .05
About the Authors:Linda Godbold Kean is a faculty member in the Department of Communication and Broadcasting at East Carolina University. Kelly Fudge Albada is a faculty member in the Department of Communication at North Caorlina State University. A version of this manuscript was presented at the National Communication Association’s annual convention, Seattle, WA, 2000.