Despite a 30-year-long debate about conceptual and methodological foundations of the cultivation hypothesis, the core idea, that people who are avid media consumers tend to adopt worldviews similar to those presented in mass media content (Gerbner, 1972), continues to provide impetus for mass communication research. One of the most prominent topics of cultivation research is crime. Theory in this area has evolved around two general dimensions: audience characteristics and message-specific influences on the cultivation process. The study reported here focuses on message characteristics, which have arguably been less influential than audience-focused studies in shaping cultivation research. Figure 1 represents a summary map of these developments. It also guides the conceptual and methodological positioning of the study reported here and presents suggestions for future work in this area. Reference will be made to it throughout the article.
[FIGURE 1 OMITTED]
Audience characteristics have played a role in the refinement of the original cultivation idea since shortly after the hypothesis was first posed in the 1970s. Gerbner, Gross, Morgan, and Signorielli (1980, 1982) added the concepts of mainstreaming and resonance to accommodate the criticism (Hirsch, 1980; Hughes, 1980) that, when demographic variables are controlled for, TV exposure is not statistically associated with fear of crime. Since then, control of demographic influences has become a standard practice in cultivation research. A number of scholars have also investigated the specific influences of demographic variables such as race, gender, area of residence, and firsthand experience of criminal victimization on cultivation (Chiricos, Eschholz, & Gertz, 1997; Gross & Aday, 2003; Morgan & Shanahan, 1997; Shrum & Bischak, 2001 ; Tyler, 1980). (1)
Perceived realism of the message is another audience-related dimension that has been reported as influencing cultivation. For example, Potter (1986, 1988) showed that a high level of perceived realism of crime drama is associated with higher fear of crime and O'Keefe (1984) found that perceived credibility of crime drama is a better predictor of fear of crime than the level of exposure to this genre. Oliver and Armstrong (1995) reported higher levels of perceived realism for reality-based TV shows such as Cops than for crime drama shows such as Law and Order, which led them to recommend that future studies test cultivation outcomes for the two genres. Perceived realism of messages might seem misplaced under audience characteristics in Figure 1. Yet, it is important to point out that it emerges as much from audience perceptions of content as dichotomous researcher-imposed classification of content as fiction (less real) or nonfiction (more real). Busselle and Greenberg (2000) conceptualized the measurement of media realism judgments along six dimensions and Busselle, Ryabovolova, and Wilson (2004) discussed the importance and particulars of how perceived realism should be measured in cultivation studies.
Finally, information processing theories have recently surfaced in research efforts to parse out the relations between TV viewing and crime orientations (e.g., Busselle, 2001 ; Busselle & Shrum, 2003; Mares, 1996; Shrum, 1996, 2001; Shrum & O'Guinn, 1993; Shrum, Wyer, & O'Guinn, 1998). Taken together, this body of mostly experimental work explains audience-specific information processing influences on the cultivation process.
The four areas of audience-centered cultivation research are listed in Figure 1. More will be added later. Moving to message characteristics in the model, three general areas of focus dominate research into message-specific influences on cultivation. (2) Studies have shown that particulars of crime message content influence cultivation. For example, Liska and Baccaglini (1990) reported that newspaper coverage of local homicide is associated with increased fear, whereas nonlocal homicide coverage is not (see also Heath, 1984). They concluded, "crimes in other cities make people feel safe by comparison" (p. 367). Although the original formulation of the cultivation hypothesis and much of the subsequent research focused heavily on TV fiction, a number of message-specific studies investigated other media genres (news, reality TV) and channels (radio, film, newspapers) for cultivation outcomes. Some scholars focused on one genre or channel and others conducted comparative investigations. Because of methodological inconsistencies across investigations and the relatively small collection of work in this area, these studies have been less influential in shaping cultivation theory. Yet, they represent a shift away from the original cultivation hypothesis that among media channels, TV is most centrally positioned to cultivate perceptions of crime and that similarities among TV genres outweigh their differences in cultivating crime perceptions.
In view of the changing media landscape in which producers are targeting specific demographic segments of the population through genre-specific content, it is important to revisit the notion of uniform cultivation effects across TV programming. Moreover, because five corporations (News Corporation, Viacom, AOL/Time Warner, Bertelsmann, and Disney) now control 90% of U.S. media content, it is reasonable to suspect that the market specifications of TV fiction content might be extended to other genres and channels (Bagdikian, 2004). Characteristics particular to a genre or channel might make some more robust than others in cultivating worldviews. Both channel and media genre comparisons have been conducted within the cultivation framework, but not within the parameters of one study. The goal is to pull these strands of research together in one study, using a relatively large statewide sample of adults. While controlling for several demographic variables, this study tests the idea that media genres and channels have varying potential to cultivate perceptions, fears, and potential behavior related to crime.
The Original Cultivation View on Channels and Genres
The original cultivation research team focused mostly on one media channel (TV) and a specific genre within that channel (fiction; Gerbner 1972; Gerbner & Gross, 1976; Gerbner et al., 1977; Gerbner et al., 1980; Gerbner, Gross, Signorielli, Morgan, & Jackson-Beeck, 1979). TV was singled out because of its prominent role in society, its distinct commercial character, and the ritualistic rather than selective consumption patterns of the audience (Gerbner et al., 1979). Indeed, they viewed TV as unlike any other medium, in that the content of commercial TV is organically composed of interrelated stories to meet the same market specifications (Gerbner, Gross, Morgan, & Signorielli, 1994). They defended their focus by arguing that because TV genres (e.g., news, advertising, drama) share packaging features and appeal it forms a coherent message system. Different genres reinforce, rather than contradict each other. TV fiction was therefore treated as representative of all TV genres' potential to uniformly cultivate worldviews (Gerbner, 1998; Gerbner et al., 1994).
Testing these views forms the basis for comparing genres and channels and leads to the first three hypotheses. The dependent variables of interest are linked to three areas of conceptual work, represented in Figure 1. The idea to distinguish between first- and second-order estimates originates from theoretical discussions by Hawkins and Pingree (1982) and further explication and testing by several scholars (e.g., Gross & Aday, 2003; Potter, 1991; Shrum, 2004; Shrum & Bischak, 2001). First-order estimates are measured by asking respondents to make estimates of the frequency or probability of events that occur on the societal level. The answers to these questions can be learned directly from TV content. In this study, estimates related to the nature of crime (e.g., how much is committed with a weapon, by non-Caucasians) correspond with first-order effects. Perceived crime risk, which is a staple measure in cultivation research, is conceptualized as a second-order estimate. (3) Second-order judgments are described as general judgments or attitudes about the world that need extrapolation from what viewers might learn from media content (e.g., personal risk estimates of violent and property crime; see Hawkins & Pingree, 1982; Potter, 1991; Shrum, 2004; Shrum & Bischak, 2001). (4) The third group of dependent measures of interest to this study relates to possible protective behavior. Gerbner et al. (1980; Gerbner et al., 1979) suggested that heavy viewers might adopt conservative attitudes about law and order and exercise these opinions in voting behavior. Similarly, Nabi and Sullivan (2001) and Holbert, Shah, and Kwak (2004) argued for a move beyond estimates of fear and beliefs about crime and for the inclusion of behavioral measures in cultivation research. The authors found evidence that TV exposure is related to intentions to take protective measures. Therefore behavior is included as a potential outcome of cultivation in Figure 1.
[H.sub.1]: The higher the exposure levels to TV crime drama, the higher the first-order estimates of crime in society.
[H.sub.2]: The higher the exposure levels to TV crime drama, the higher the second-order estimates of crime in society.
[H.sub.3]: The higher the exposure levels to TV crime drama, the more likely that audience members will report possible defensive behavior.
Variance Across TV Genres
Hawkins and Pingree (1981) were among the first to argue that TV genres do not uniformly cultivate worldviews. Based on their study of Australian school children, they recommended that cultivation researchers distinguish between TV genres and study the possible effects of differences between types of content rather than lumping them together and treating TV viewing as a uniform experience. They reported that cultivation...