Children on the island of St. Helena were surveyed on two occasions, in 1993 at age 3 to 4 years and again in 1998 at age 7 to 8 years, before and after the introduction of broadcast television in 1995. Pre-TV measures of social behavior were provided by teachers using the Preschool Behavior Checklist (in the first survey) and post-TV social behavior was measured using the Rutter Behavior Questionnaire for the second survey wave. During the second survey wave, the children also provided viewing data via a three-day TV diary. For those same three days, the content of all broadcast television output was analyzed for the appearance of violence in programs. The television content analysis data were merged with children's TV viewing data to yield measures of exposure to televised violence. Just over half (55%) the children who kept viewing dairies watched any television during the three days of viewing measurement. These children watched an average of 3 hours and 10 minutes of television. These children were foun d to have seen an average of 95 violent acts or 7.28 minutes of violence across this same period. Boys (10.23 minutes) watched more violence than girls (5.43 minutes). Viewers did not differ significantly from non-viewers on either their pre-TV or post-TV antisocial behavior scores. Boys displayed significantly more antisocial behavior than girls in wave two following the introduction of broadcast television, whereas the two sexes had exhibited no inter-gender difference in antisocial tendencies prior to broadcast television. Overall TV viewing was not correlated with antisocial behavior scores at any point. Cartoon viewing was significantly correlated to post-TV antisocial scores over all children, but within the sexes for girls only, and this relationship survived statistical controls for pre-TV antisocial behavior. Total number of violent acts seen was correlated with post-TV antisocial behavior scores for girls only, while number of male-perpetrated violent acts was correlated with antisocial behavior amo ng boys and girls. All these relationships disappeared when pre-TV antisocial behavior scores were controlled. Cartoon viewing was significantly related to pre-TV antisocial behavior and post-TV neuroticism levels even after the effect of number of violent acts seen was controlled, suggesting that nonviolent ingredients in these programs appealed to already unruly children and that cartoon viewing was linked to lower anxiety among children.
One of the most controversial topics of debate about the impact of television has focused on the effects that televised depictions of violence might have on children. For many years, the debate remained open with disputes about the validity and conclusiveness of the scientific evidence being widespread. In the United States, major commissions of inquiry into the causes of social violence initially were equivocal in their conclusions about the role of television in this context (National Commission on the Causes and Prevention of Violence, 1969; Surgeon General, 1972), while later reviews of the evidence concluded that most of the research does suggest a link between violence on television and aggressiveness in children and teenagers (Huesmann & Eron, 1986; Pearl, Bouthilet, & Lazar, 1982). Major studies funded by the American television networks during the 1970s produced conflicting conclusions about the effects of television-mediated violence or antisocial conduct on the antisocial and delinquent tendencies of viewers (Belson, 1978; Milavsky, Kessler, Stipp, & Rubens, 1982; Milgram & Shotland, 1973).
Cause-effect analyses conducted in laboratory experiments have been viewed by some critics as lacking sufficient ecological or external validity to be of any use (Barker & Petley, 1997; Cook, Kendzierski, & Thomas, 1983; Cumberbatch & Howitt, 1989; Gauntlett, 1995; Stipp & Milavsky, 1988). Critiques of surveybased evidence have found the scientific evidence problematic for methodological reasons associated again with the strength of the measures used to establish television viewing and personal aggressiveness or because of inappropriate interpretation of statistical relationships between these variables (Freedman, 1984, 1986).
During the 1980s and 1990s, more sophisticated conceptual models emerged about viewers' involvement with media violence that revealed modified thinking about the psychological mechanisms that underpin audiences' reactions to such material (e.g., Berkowitz, 1984; Berkowitz & Heimer, 1989). The viewer was regarded as playing a cognitively more active role in the context of media content interpretation and choice of response to media stimuli. Even so, comprehensive reviews of the scientific evidence have emerged from the U.S. that have reached the broad conclusion that a cause-effect relationship between violence on television and child or teenage aggressiveness has been demonstrated (Comstock & Paik, 1991; Wilson et al., 1996). This largely political position on a still debatable social scientific conclusion does not entirely square with arguments that more attention needs to be given to the context in which violence occurs on television and to the attributes of the violence itself (Wilson et al., 1996), as we ll as the context within which viewing occurs (Potts, Huston, & Wright, 1986). Specifying the relative significance of these features when applied individually or in combination with one another to the overall strength of impact of media violence portrayals is a matter on which further research is needed. In addition, the reactions of audience members to television content and choices of what to watch may be further mediated by their individual personalities (see Conway & Rubin, 1991; Kremar & Greene, 1999).
More sophisticated psychological models of the processing by viewers of television content represent important developments that enhance the conceptual and methodological ability of media researchers to understand the nature of the medium's potential effects in complex media environments. The ideal environment to study the impact of television is one in which the population has had little or no prior exposure to the medium. This phenomenon is a rarity today. Few reachable communities exist in which television is being introduced for the first time. Early research of this sort in virginal television communities took place while the medium was in its infancy and indicated effects upon children's allocation of time to tasks and the displacement of certain activities by television viewing (Himmelweit, Oppenheim, & Vince, 1958; Schramm, Lyle & Parker, 1960). As television spread, later studies were pushed out to study more remote communities left behind in the television era by quirks of geography. In these commu nities, leisure activity displacement was observed in the short term (Coldevin & Wilson, 1985; Williams, 1986). However, evidence emerged in at least one case that such displacement effects weakened in the longer term (Coldevin & Wilson, 1985). More worrying were findings that indicated upward shifts in the prevalence of antisocial conduct among youngsters following the onset of television in a community that had previously had no television reception (Williams, 1986). The extent to which children in remote communities react with increased aggression upon experiencing television entertainment for the first time may depend upon whether they perceive the Western lifestyle and its role models as attractive or whether they regard such things as a threat to their local heritage (Granzberg, 1985). What these previous studies lack is any detailed measurement of the nature of individual children's viewing experience. Behavior changes are measured before and, at some point, after the introduction of television where t he only treatment variable in such quasi-experiments is the presence of television (e.g., Williams, 1986). What is really needed are data at the level of the individual child about his or her particular viewing habits and about the kinds of program content that have been featured prominently in their viewing diet. This paper reports findings from a study of the impact of new television services on a broadcast television-naive community where such individualized data were obtained.
The St. Helena Study
St. Helena is a British colony in the South Atlantic Ocean with a population of around 6,000. It is among the world's most isolated inhabited islands. Access to the Island is restricted to infrequent visits by the RMS St. Helena. The island is without daily newspapers, a cinema, a regular bus service, an airport and, until recently, broadcast television. Satellite television was beamed to St. Helena for the first time in March 1995. Initially only a single news service from CNN was offered. The television service expanded from 1 November 1996, with the addition of The Cartoon Network, Hallmark (films), SuperSport and Discovery (documentaries). Since June 1998, the program line-up has been MONet Brochure (KTV, Movie Magic and BBC), Discovery and SuperSport. Transmissions can be received 24 hours a day.
Educational provision is well developed on the island and in 1992 (when the research project started) the 1,300 pupils were provided for in a three-tier education system incorporating six first schools (for 5- to 8-years-olds),  three middle schools (for 9-to 12-year-olds), and a secondary school (for 13- to 16-year-olds). There is also nursery provision at each first school for infants aged 3 to 5 years. Children enter this provision when they reach the age of 3.5 years. A falling school roll led to the closure in 1996 of one of the first schools (with only five first schools remaining).