Content-Type: text/html ABSTRACT Riding the Hoopla: An Analysis of Mass Media Coverage of GMOs in Britain and the United States: 1997-2000 Eric A. Abbott Tracy Lucht Jeffrey P. Jensen Zajira Jordan-Conde Greenlee School of Journalism and Communication Iowa State University Three models - social amplification of risk, hoopla, and triggering effects - were used to develop and test predictions about coverage of genetically modified organisms in the New York Times, London Times and London Daily mail from 1997-2000. A content analysis showed scientists have declined significantly as sources over time, while citizens' groups have remained constant. Themes, or frames, for articles shifted in response to triggering events. Positive themes declined over time while negative ones remained relatively constant. Abbott is a professor, Lucht is a graduate student, and Jensen and Jordan-Conde are undergraduate research assistants. Contact is Eric Abbott, 204 B Hamilton Hall, Greenlee School of Journalism and Communication, Iowa State University, Ames IA 50011. Tel: 515-294-0492. email: [log in to unmask] Paper presented to the Science Communication Interest Group, Association for Education in Journalism and Mass Communication, Washington DC, August 5-8, 2001. Riding the Hoopla: An analysis of mass media coverage of GMOs in Britain and the United States Riding the Hoopla: An Analysis of Mass Media Coverage of GMOs in Britain and the United States: 1997-2000 Eric A. Abbott Tracy Lucht Jeffrey P. Jensen Zajira Jordan-Conde Greenlee School of Journalism and Communication Iowa State University Abbott is a professor, Lucht is a graduate student, and Jensen and Jordan-Conde are undergraduate research assistants. Contact is Eric Abbott, 204 B Hamilton Hall, Greenlee School of Journalism and Communication, Iowa State University, Ames IA 50011. Tel: 515-294-0492. email: [log in to unmask] Paper presented to the Science Communication Interest Group, Association for Education in Journalism and Mass Communication, Washington DC, August 5-8, 2001. Introduction One well-documented fact about mass media coverage of scientific risk topics is its unevenness across time. A large number of agenda-setting studies (McCombs and Shaw, 1972; Gonzenbach and McGavin, 1997) have shown that mass media coverage in general tends to be very episodic, with heavy coverage of topics across media for relatively short periods of time. Studies of individual science risk topics also have documented either cyclical or one-time peak coverage of such topics as flouridation of water (Mazur, 1996), nuclear power (Sandman, 1994), pesticides (Sandman, 1994; Mazur, 1996; ESCOP Survey, 1994), AIDS (Rogers, Dearing and Chung, 1991), salmonella infections (Miller and Reilly, 1995), radon (Mazur, 1996), alar in apples (Anderson, 1995), silicon breast implants (Nelkin, 1994; Entman, 1997), and bovine spongiform encephalopathy (BSE) (Miller and Reilly, 1995). A few longitudinal studies, such as Rogers, Dearing and Chung (1991), have documented "triggering events" leading to multiple peaks in coverage over time. Neuman (1990) studied peak coverage of a number of events and topics to try to document a "threshold" level at which the public would come to regard the event or topic as very important. The fact that coverage is uneven raises important questions about exactly what happens before, during and after these peaks in coverage. While many studies have examined the nature of coverage either at a peak when a crisis occurs, or in general over time, there has been little examination of the dynamics in media coverage of risk topics that occur within these media cycles. For example, when coverage suddenly "takes off," are there important changes in who gets quoted, in the types or numbers of themes that are covered, or in the positive or negative valence of the coverage? This paper synthesizes three conceptual models that shed light on what leads to these peaks in coverage, and then examines a current issue - coverage of genetically modified organisms (GMOs) - to study patterns of coverage across a four-year period. Three Conceptual Models Explaining Differential and Cyclical Coverage of Risk Issues by the Mass Media Many scientific issues that pose potential risks never receive extensive media coverage. What are the factors that lead some of these issues to be heavily covered while others are not? Three complementary conceptual models are offered as a way of explaining both why some issues are covered and others are not, and why one might expect cycles of coverage of topics that do get covered by media. The three are: (1) Social amplification of risk; (2) A "Hoopla" model of mass media coverage of science and technology; and (3) A model of "triggering" events that resonate with mass media reporters and the public. 1. Social Amplification of Risk Theory The idea of social amplification of risk, developed in 1988, is that the coverage of risk topics is determined in part by the interests, reactions, and agendas of individuals and the social organizations to which they belong. Kasperson (1992: pp. 157-158) notes that "the concept of social amplification of risk is based on the thesis that events pertaining to hazards interact with psychological, social, institutional and cultural processes in ways that can heighten or attentuate perceptions of risk and shape risk behavior." Groups in society function as potential "amplification stations," each potentially able to generate activity that may lead to media coverage and public awareness or response. "Social group mobilization appears to be highly intertwined with both media coverage and eventual social impacts of the hazard events. Heavy media reporting appears to stimulate social mobilization, and vice versa, and that activism is a 'downstream' variable in shaping eventual economi c and social impacts of hazard events (Kasperson, 1992, pp. 169-170). Burns (1990) subsequently applied social amplification theory to an analysis of 128 hazard events including pesticide use, mercury and benzene release, lead emissions, chemical releases, asbestos use, CO emissions, and acid rain. Results supported the idea that actions by social groups played an important role in public attention to issues, media coverage, and subsequent "ripple" effects. Hoban (1995), in an analysis specifically focusing on applications to biotechnology issues, noted that when groups and organizations become involved, an issue moves from being a "scientific" issue of interest mainly to researchers to a "social" issue in which many groups in society actively push their views. The result is that when something becomes a social issue, both the range of groups involved and the breadth of their concerns broaden. Scientists are often most concerned with the scientific efficacy of the innovation, while social groups often are concerned about political, economic, mor al, and other aspects. Hoban (1995) emphasized that involvement by social groups in the public debate about scientific risks often involves more than just a perception that something might be dangerous. These groups may be less concerned about the technical aspects of the issue than they are about how they can use it effectively to gain media coverage that will enhance group and fund recruitment. This view is also strongly emphasized in a recent congressional report on the issue (Subcommittee on Basic Research, 2000, 59-62). From this perspective, one might expect several results when something moves from becoming a scientific issue to a social issue: (1) an increase in mass media coverage as more and more groups focus attention on a topic; (2) an increase in the variety of sources that are used in mass media coverage as more and more groups seek to put forth their views; (3) a shift toward moral, political and economic issues rather than scientific issues. 2. The Hoopla Effect The hoopla effect was conceived as a way of characterizing a surge of mass media coverage and interest in scientific and technological innovations. Abbott and Yarbrough (1989) noted that mass media coverage concerning farm computers in the 1980s tended to build rapidly to a peak, and then decline over time. This peak in coverage was not directly related to adoption; rather, it was associated with a pattern of hype or promotion of new technologies. Roger Fidler, who helped Knight-Ridder develop and launch the Viewtron videotex experiment in the 1980s, notes that: "The announcement of practically every major 'discovery' or 'breakthrough' in the past two centuries has been immediately followed by a deluge of wild speculation and hyperbole proclaiming the birth of a new era or revolution in the making. So hungry are the media and their audiences for stories about the promise of things 'new' that in recent years the hype has begun to precede and overshadow the formal announcements" (Fidler, 1997, p. 253). Brody (1991) explained this pattern of hype by pointing out it takes a great deal of capital to bring innovations to market, that that in turn requires convincing an audience of bankers that the innovation is bound to succeed. Thus, promoters and inventors pour their efforts into getting media coverage of their innovation. Klopfenstein (1985) and others have noted how industry and promoter self-interest, coupled with an enthusiastic and optimistic media, consistently create "vaporware," products that either will never actually exist or are touted as being far superior to what the actual product will be. Bryant, Gonzenbach and McCord (1994) found a take-off in media coverage about "the information highway" in leading U.S. newspapers in 1993-94. This increase, they found, "did not spring from a rapid adoption and use of these technologies by the majority of people." Abbott and Eichmeier (1998) examined the hoopla effect by applying it to both a technological innovation (videotex/teletext) and a scientific health claim about oat bran. In addition to tracking the frequency, volume and tone of articles, they tracked sources quoted in the articles to see if they changed. It was expected that in the hoopla period (a takeoff in media coverage), producers would be major sources since they would have a great interest in promoting their innovations. Over time, the tone of the dialogue would be expected to become less positive as opposition or competing groups emerge and some of the optimistic forecasts are tempered. They found that the tone of articles for oat bran was significantly more optimistic in the pre-hoopla period, suggesting the creation of a positive hoopla for the new product, but this was not the case for videotex/teletext. In terms of sources, they found that producers were significantly more likely to be used as sources during the ho opla period for both oat bran and videotex/teletext. The hoopla research suggests that the hype associated with new products and technologies is due to several factors: (1) The self-interest of promoters; (2) the tendency of media to look for new products and present over-optimistic claims about them. It also suggests that the number of sources varies over time for several reasons. First, in the early stages, relatively few people are aware of a new technology, and these people are often promoters. Second, as more and more people adopt or there is more publicity, others learn about the innovation and begin to be used as sources. Thus, over time the number and diversity of sources might be expected to increase. 3. Triggering Events The third model has to do with media and public response to certain events associated with risk topics. Why, for example, do reports of the danger of salmonella infections - which experts agree is a serious health problem in the United States - arouse relatively little interest by either media or the public, while mad cow disease - which currently poses no major health threat in the United States - gains front page coverage day after day? In psychometric studies carried out at Stanford and elsewhere, Slovic (1992) found that public response to scientific risk topics could not be predicted simply by knowing the number who might die, the economic costs, etc. Of key importance was the "dread" of the consequences, and the degree to which the probability of their occurrence is unknown or delayed. For example, nuclear reactor accidents and DNA technology both aroused perceptions of "dread" and "unknown" danger in the public. Slovic also found that there were cross-cultural differences in how publics reacted to potential dangers. Genetically modified foods in Britain have been characterized as "Frankenfoods" in the media and by citizens. In the United States, that term has not been used much by mass media and it has in general not resonated with U.S. audiences. Meanwhile, the danger posed to the monarch butterfly by bt corn in the United States has resulted in significant media coverage and public reaction in the United States but less in Britain (Abbott and Lucht, 2000). Rogers, Dearing and Chang (1991) found that neither the media nor the public seemed to be very interested in AIDS when it was reported as a serious disease in Africa. Nor did it get much attention when presented as a disease of the gay community. However, when a medical journal article suggested that AIDS might be spread within the heterosexual community by "casual contact," attention soared. A second triggering event documented by Rogers, Dearing and Chang was the death of film star Rock Hudson of AIDS. It is difficult to predict what will trigger coverage and what will not. However, the cycles of media coverage will be uneven due to the fact that some events trigger much more attention than others. Framing of stories occurs when a particular way of looking at a complex topic narrows discussion. As Dunwoody (1992) has noted, because journalists lack knowledge about scientific risk issues, they are particularly susceptible to being influenced by either sources or events to frame their stories in a particular way. Once a story is framed, the themes used in the story and the sources sought out are narrowed to match the frame. Triggering events have an important influence on framing, because they call attention to certain aspects of a scientific risk topic. One other aspect of triggering is the extent to which it might offer opportunities for opposing views to be expressed. Sandman (1994) and Nelkin (1994) point out that the journalistic tendency to report "both" sides of an issue causes them to look for opposing points of view when they perceive that an issue is controversial. Sandman noted that "reporters typically start with a government official, the swing vote. If the government says 'dangerous,' they look for an industry source or possibly an expert to say 'safe.' If the government says 'safe,' they look for a citizen or possibly an activist to say 'dangerous.' However, this tendency to look for both sides of an issue does not occur when an issue is not seen as controversial. Priest (1995) and Gaskell, Bauer, Durant, and Allum (1999) found that most early coverage of biotechnology and genetically modified organisms tended to be overwhelmingly positive. This suggests that in the early stages of a new scientific developm ent, a "benign science" approach might be used by journalists either because they don't see possible negative implications or groups putting forward negative implications have not yet emerged. A triggering event raises public and media awareness of an issue, and also might function to open the doors to opposition views. Gaskell et al. (1999) in fact suggest that heavy coverage of an issue, regardless of its tone, is associated with negative public perceptions (and possible subsequent amplification by social groups). From this, one might expect that a triggering event would: (1) increase coverage of a scientific risk; (2) frame the discussion in a particular way; (3) cause journalists to shift from a "benign science" model to a "controversy" model of coverage that involves reporting "both" sides of the debate, which leads to an increase in the number of sources (and especially sources that might be opposed to the issue). Synthesis of the Three Models Synthesizing the three models leads to five specific predictions: 1. Wide fluctuation in coverage. Because of variances in group "amplification" and the hoopla cycle, coverage of science risk topics such as genetically modified organisms would be expected to fluctuate widely over time. Given that certain events or developments "trigger" media coverage and public response, one would predict a series of peaks in coverage of GMOs over time, each organized thematically around the "triggering" topic, which then frames media discussion. 2. Increase in variety of sources quoted. As an issue unfolds, one would predict an increase in the number and variety of sources used by journalists. This occurs both because more people and groups have knowledge and see how they might gain from becoming sources, and because as coverage increases, journalists become more likely to see the topic as sensitive or controversial, thus increasing their tendency to "report both sides." 3. More negative coverage over time. Over time, the tone of articles would be expected to become more negative. Initially, a positive tone is expected, both because the only knowledgeable sources tend to be those who have a stake in the innovation's success, and because the issue is seen as benign - a routine story about a scientific achievement. As triggering events occur, the issue would become framed as "controversial," leading to the reporting of positive and negative aspects. 4. Shift in themes. Story themes would be expected to shift over time as triggering events reshape frames used by journalists to cover their stories. It is unclear whether older themes would continue to be reported as new ones emerge. If this occurred, the number of total themes in an average story would increase over time. 5. Cross-cultural differences in reporting on scientific risk topics. Because social groups, journalistic conventions, and cultural histories vary, the way in which a scientific risk topic is covered will vary from country to country. That is, although the "scientific issue" may be exactly the same, the "social issue" may be quite different. Media Coverage of Genetically Modified Foods Genetically modified foods (or Genetically Modified Organisms) is an appropriate issue to study for several reasons. First, it represents a watershed in biotechnological research that has rich promise for human ability to shape and modify foods in the future to meet human needs. Second, it is a topic that, as Slovic (1992) found, arouses public concern. Third, it is a topic that will continue to be important over time, and thus lends itself to longitudinal study. Globally, 44.2 million hectares have been planted to genetically modified crops, primarily to corn, soybeans and cotton. The United States has 68% of all genetically modified crops, while Argentina has another 23%. Canada has 7%, leaving the rest of the world with 2%. A total of 36% of soybeans and 7% of corn globally are now genetically modified (Brown, 2001). However, since these two crops find their way into a majority of grocery store products, food industry experts estimate that 60-80% of items in a typical U.S. grocery store could now potentially contain at least one component that is genetically modified (Hopkin, 2001; Holander, (2001). Early studies of biotechnology found that much of the coverage was positive and most of the sources were either researchers or industry representatives. In 1990-91, for example, Priest (1995) found that 81 percent of the U.S. coverage of biotechnology was positive; only 7% of the "arguments" presented in news texts concerned risks of biotechnology. In addition, Priest found that 75% of sources were either industry or university (university sources were positive by a ratio of 11 to 1). Gaskell et al. (1999) compared coverage of genetically modified foods in 12 European national newspapers compared to the Washington Post in the United States. They also found that early coverage (1984 to 1991) tended to be positive on both continents and occurring at a fairly low level. Following 1992, the pattern of coverage diverged, with increased coverage in Europe (coverage doubled in European papers between 1992 and 1996), while it remained constant in the United States. They also noted that during this time, coverage in Europe was mainly about "progress" associated with genetically modified organisms, while in the United States it moved from "progress" to "economic prospect." United States stories tended to report on both risks and benefits, while European papers mainly focused on benefits. Two other factors affecting mass media coverage concern activity by industry and efforts by citizens' activist groups. The biotechnology industry spans both Europe and the United States. However, most of the production of genetically modified foods thus far has come from the United States. Monsanto Corporation (now a subsidiary of Pharmacia) was the most prominent company openly promoting genetically modified crops, and thus became a lightning rod for protests. In some ways, the controversy over GMOs has been linked to multinational corporations and their efforts to sell GMO crops and seed globally. Citizens' groups, including Greenpeace, Environmental Defense Fund, and the Union of Concerned Scientists have become actively involved in opposition to GMOs. A coalition of more than 50 non-profit organizations came together in 1999 to launch a series of full-page advertisements in the New York Times designed to "educate" the public about transgenic research. Under headlines su ch as "Who plays God in the 21st Century?" "Unlabeled, untested_a you're eating it," "Genetic roulette," "Where will the next plague come from?" and "Biotechnology=Hunger," the coalition presented negative arguments concerning not only the monarch butterfly but a series of issues related to concentration of economic production, lack of diversity in seed, and globalization (the ads can be found at http://www.turnpoint.org/geneng.html). Later, protesters generated heavy media coverage at both the World Trade Organization meetings in Seattle and Washington, D.C. Method A content analysis was conducted of agriculturally related transgenic topics, GMOs and related key words in three newspapers - the New York Times in the United States, and the London Times and London Daily Mail in Britain - from Jan. 1, 1997 through Dec. 31, 2000. The first two papers were selected because they represent the elite press in each country. The London Daily Mail was selected because it has given special emphasis to transgenic/GMO issues. No comparable paper was found for the United States. A Lexis/Nexis database search was conducted to select all relevant articles in the three papers during the time period, and these were graphed. Based upon the results, articles were divided into four time periods: (1) Pre-hoopla - coverage tended to be at a rather low level; (2) Hoopla - there was a sharp increase in coverage up to a peak; (3) Post-hoopla - after a peak at the hoopla, coverage declined; (4) Second Hoopla - due to another triggering event, coverage increased substantially again. Table 1 shows the distribution of articles into the four time periods for the three newspapers. A total of approximately 50 articles were selected from each newspaper for each of four time periods. In the pre-hoopla and post-hoopla periods, there were fewer than 50 articles total in the New York Times. In the second hoopla period, two different sampling groupings resulted in more than 50 articles being analyzed for each newspaper. The time periods and numbers of articles are shown in Table 1. Three different coders analyzed the articles. Intercoder reliability was .85. Variables coded included: ù Date, publication, and length in words ù Period (pre-hoopla, hoopla, post-hoopla, second hoopla) ù Sources (a list of potential source types was checked for each story, including government scientists, university scientists, other scientists, journal editors or the journal itself, industry, industry associations, citizens, citizens' activist groups, international organizations, doctors, religious leaders, politicians and government employees, farmers, wholesalers, etc.). The coding counted only the presence or absence of each source type, not how many times in a story a particular source might have been used. Some source categories were then collapsed: "scientists" includes all forms of scientists except those who worked for industry or a particular activist group. "Industry" includes both industry and industry associations. ù Themes. The presence or absence of the following themes in each story was coded: human health, environment, business, farming, moral, regulatory, labeling (a special subset of regulatory that occurred frequently). ù Tone. Each theme was coded for the presence of "positive" or "negative" tone. An article could have both positive and negative tones as well as having neither positive nor negative tones. This permitted an analysis of positive and negative material separately rather than relying on an overall summary coding of positive or negative. For example, it could be that both positive and negative thematic material increase during a triggering event. Results Fluctuation in Coverage Figure 1 shows the wide fluctuation in coverage over the 1997-2000 period. While coverage for all three newspapers was relatively low from 1997 through October of 1998, the number of articles in the British press shot up to more than 100 per month (an average of 3.5 per day) in the London Times and 80 per month in the Daily Mail during December 1998 and January, 1999. The trigger was a scientific report that rats fed raw genetically modified potatoes developed serious health problems (many died). Although the scientific report was later criticized sharply by other scientists, coverage of the issue continued, as it turned out that some government officials and industry representatives had not been entirely candid about GMO research or the presence of genetically modified foods in stores. In the United States, the first triggering event in May, 1999, was another scientific publication, this one suggesting that the monarch butterfly might be harmed by eating the pollen from corn plants that had been genetically modified to contain a small amount of a naturally occurring insecticide known as bt. This led to a peak of 23 stories per month in the New York Times, far above the previous level, but significantly below British coverage levels. Coverage on both sides of the Atlantic then declined substantially until May 2000 when a second triggering event hit Britain. The second event concerned a revelation that canola seed brought in from Canada and planted at more than 75 locations throughout Britain and Europe had been accidentally contaminated by a bt variety growing in a nearby Canadian field. This led to a second peak of more than two stories per day in the London Times and a doubling of stories in the Daily Mail. The second United States triggering event was the discovery in October 2000 of StarLink, a bt corn variety not approved for human consumption, in taco shells (and eventually it was found to have spread throughout the U.S. corn supply). These results support research expectations, and suggest that very different triggers were operating at different times in the two countries. Increase in Number and Variety of Sources Quoted Table 2 shows the total number of sources across the four time periods for each of the three newspapers. These results do not support the idea of an overall increase over time in the number of sources quoted in stories. However, they do support the idea that at times of triggering events, the number of sources increases. This occurred despite the fact that the average story length did not vary significantly over time. A one-way ANOVA (with a scheffe test) for each newspaper and all three collectively showed that the post-hoopla period used significantly fewer sources of all kinds than the other periods. Table 3 examines the use of public officials, scientists, citizens' groups, farmers and industry as sources. One clear trend shown is the decline in the use of scientists as sources over time, especially in the British press. From a pre-hoopla level of being quoted in 28 to 33% of stories, they declined to less than half that level by the second hoopla period. Meanwhile, citizens' groups and industry remained the same. A gradual decline in the proportional use of scientists was expected, since in the beginning they constituted one of the few sources that knew about this topic. However, the steep decline noted raises important questions about what happened to this group as a source. This may reflect a loss of confidence in scientists by British journalists. At any rate, it means that in Europe, where the public has strong doubts about GMOs, scientists are unlikely to be quoted in articles on this topic, while opposition groups are quoted in from 35 to 45% of all articles. The overall expectation was that the use of both scientist and industry sources would decline over time as more sources of other types competed for media attention. A pooled grouping of both scientists and industry sources did show a significant reduction in use over time (Table 4). (Although not significant, industry sources alone also showed a slight reduction in use over time). The other expectation was that other sources, such as clergy, citizens, citizens' groups, doctors, international organizations, etc. would increase as sources over time. Individually, none of these changed significantly, nor did a pooled grouping of all other sources show any significant change over time. Since scientists declined as a source over time while other sources remained constant, the expectation that other voices would be heard proportionately more is supported by these results. The results also show that in both countries, farmers are used in stories during hoopla periods, but not otherwise. Public officials constituted the main source for all three newspapers across all four time periods. The majority of all stories tended to quote a public official. Positive or Negative Tone of Themes The total average number of themes mentioned in GMO stories over time varied significantly, with the fewest themes mentioned in the post-hoopla period and the greatest number in the pre-hoopla and second hoopla periods (See Table 5). Rather than a triggering event serving to open up the number of themes about GMOs, it seemed to lead to a reduction in themes, perhaps by focusing coverage more narrowly on some aspect of the topic. The trend was significant overall for all three newspapers, but was especially pronounced in the Daily Mail. It had been expected that the number of negative themes would increase over time, as triggering events emerged and social groups elected to become sources to speak out against them. However, this did not occur. In fact, Table 6 shows the mean number of negative themes in a story did not vary significantly over time. However, the number of positive themes declined significantly during the post-hoopla period. The New York Times shows a gradual decline in positive themes over time (although short of being statistically significant). The British papers showed a slight increase in positive themes in the second hoopla. It should be noted that at all time periods, the Daily Mail ran approximately 2.5 negative themes for every positive one, and the London Times ran 1.5 negative themes for every positive one. The New York Times tended to run equal numbers of negative and positive themes. Shift in Themes Table 7 shows a shift in themes over the four periods, and also indicates whether there was more positive or negative coverage of each theme. The three most frequently mentioned themes are listed for each time period, along with the percentage of stories from the period that mentioned that theme. A "+" indicates that there were more positive than negative ratings of the theme, while a "-" indicates there were more negative than positive ratings for that theme. (The regulatory theme was not rated positive or negative since it was very difficult to judge the intent or effect of some regulations). Results show that there was a shift in themes over time. For the Daily Mail human health and regulatory issues were most prominent in the pre-hoopla period. Human health issues then declined in importance, while business concerns rose to be the most frequent theme. During the second hoopla, when the canola scare occurred, environmental concerns became the top issue. Moral concerns were within the top three themes in every time period. For the Daily Mail, every category was negative in tone, with the sole exception of human health themes in the second hoopla (when Golden Rice, rice fortified with vitamin A and other benefits of GMOs were discussed). The London Times also featured human health and regulatory concerns in the pre-hoopla and hoopla periods, but it also emphasized business themes. Interestingly, human health was consistently treated as a positive issue in the Times, in contrast to the Daily Mail. On the other hand, business issues were consistently treated negat ively by both British newspapers throughout all time periods. Like the Daily Mail, the Times also gave more emphasis to environmental themes at the time of the canola scare, and treated it negatively. Finally, moral themes, which were less emphasized during the first three periods, took the top spot with 33% of articles having this theme during the second hoopla. The New York Times took a quite different approach to coverage of GMOs. It strongly emphasized business concerns in the first two periods, with 71.4% of stories mentioning business themes during the pre-hoopla period. The focus on business declined over time (although it has always been in the top three themes), while environmental and regulatory concerns rose following the monarch report. In addition, the business coverage was strongly positive in the first two periods, but has been negative ever since. Coverage was overwhelmingly positive in the early periods, emphasizing the positive possibilities for farmers and human health. The monarch triggering event caused negative environmental coverage to rise in importance during the hoopla period, and this continued through the post-hoopla period. The overall expectation was that themes would shift from discussion of scientific aspects (such as human health and the environment) or business/farming aspects to issues such as moral or regulatory/labeling. To test this, four thematic groups were created (human health/environment; business/farming; moral; regulatory/labeling). Mean scores for each group were then compared across the four time periods. Results (Table 8) showed that the human health/environment theme did decline across the first three time periods, but then sharply increased. Since it is the sharp increase that resulted in statistical significance, the notion of a linear decrease over time is rejected. None of the other groups showed any significant change over time. Moral and regulatory issues in fact decreased slightly over time - opposite what was expected. Discussion The three synthesized models used to guide this paper suggest that themes (or frames) within newspaper stories would be expected to change in response to triggering events. Four such events occurred during the period of study - the rat/potato study and the canola scare in Britain, and the monarch butterfly study and StarLink corn incident in the United States. These events did seem to guide both the thematic and positive/negative dimensions. In the United States, the monarch study resulted in a shift from a business frame to an environmental frame, and introduced the first systematic negative coverage of GMOs. In addition to environmental concerns, the event triggered an examination of regulatory capacity to cope with environmental issues. More recently, the finding of traces of StarLink corn in taco shells resulted in human health becoming the dominant theme in the New York Times, although GMOs overall were still portrayed positively. In Britain, the initial triggering event began with the rat/potato study, but quickly shifted to moral and regulatory concerns after it became apparent that public officials had covered up information about GMOs. Although the rat/potato study focused attention on human health concerns, these became overshadowed by business, moral, and regulatory issues. The canola scare caused a clearer effect, with environmental concerns moving suddenly to the top of the thematic agenda for the Daily Mail and the number two position for the London Times. Both were negative in tone. The triggering events clearly did have an effect in shaping this coverage, but they could not provide a total explanation. Underlying business concerns guided the New York Times coverage throughout the periods, and moral concerns were pronounced through all periods for the Daily Mail. The cycle of use of sources during the four periods clearly documents the disappearance of scientists from the dialogue, at least in Britain as well as a pooled decline in both scientist and industry sources. This was predicted by the initial models. However, the models suggested that citizens, citizens' groups and other sources would increase and they did not. Perhaps, by 1997, these groups had already established themselves as sources and were able to maintain this over time. Results show that the average total number of sources cited in a story declined over time, with the fewest sources in the post-hoopla period. Thus, as a proportion of all stories, citizens' groups and other sources did increase. Thematic coverage was expected to move away from scientific and producer (industry) themes and to moral, regulatory and labeling concerns. However, no significant linear pattern was found for any of the grouped themes. The scientific themes -- human health and environment -- peaked significantly in the second hoopla. None of the other groupings experienced significant change over the four periods. One possible explanation for this is that other factors, such as the moral orientation of the Daily Mail, or the business orientation of the New York Times, were stronger influences. A second possibility is that the topics of human health and environment are not the exclusive domain of scientists. The peak in human health/environment themes in the second hoopla period, when scientists were disappearing as sources, strongly suggests that citizens' groups and other sources effectively used these same themes to oppose GMOs rather than relying on moral or regulatory appeals. Simply said, they argued that GMOs were not safe for humans or the environment, rather than arguing that they were not morally acceptable. The tone of the themes indicated that, contrary to expectations, negative themes did not increase over time; however, positive themes declined significantly in the post-hoopla period, giving an overall negative tilt to coverage. The reduction of positive themes correlates with the departure of scientists and industry representatives as sources. This study is somewhat exploratory, utilizing the social amplification of risk, hoopla and triggering models to develop predictions for coverage of GMOs in Britain and the United States. Although the results in some cases did not match the models well, they do suggest there is value in searching for underlying patterns of coverage within media cycles for scientific risk topics. Studies including additional risk topics would help refine these predictions further. References Abbott, Eric A. and Eichmeier, April A. (1998, August 7). The hoopla effect: Toward a theory of regular patterns of mass media coverage of innovations. Paper presented to Theory and Methodology Division, Associate for Education in Journalism and Mass Communication, Baltimore. Abbott, Eric A. and Lucht, Tracy. (2000, July). How triggering events affect mass media coverage and source use concerning genetically modified organisms (GMOs) in Britain and the United States. Paper presented to USACC 2000 Congress, Washington DC. Abbott, Eric A. and Yarbrough, J. Paul. (1989, April). Seminar on the role of information in the diffusion process. Department of Communication, Cornell University, Ithaca, NY. Anderson, Clifton. (1995). The food information war: Consumer rights and industry prerogatives. In Mauer, Donna and Sobal, Jeffrey, Eating Agendas: Food and Nutrition as Social Problems, (pp. 167-187). New York: Aldine de Gruyter. Brody, Herb. (1993). Great expectations: Why technology predictions go awry. pp. 150-159 in Albert H. Teich (ed.), Technology and the Future (6th Edition), New York, St. Martin's Press. Brown, Kathryn. (2001, April). Seeds of concern. Scientific American 284:4, 52-57. Bryant, Jennings; Gonzenbach, William J. and McCord, Lola. (1994, May). Press coverage of and public attitudes toward the information highway. Paper presented to the World Association of Public Opinion Research Conference, Danvers, MA. Burns, W. (1990). Introducing structural models and influence diagrams into risk perception research: Their value for theory construction and decision making. Unpublished doctoral dissertation, Department of Decision Sciences, University of Oregon. Dunwoody, Sharon. (1992). The media and public perception of risk: How journalists frame risk stories. In Bromley, D. and Segerson, K. The Social Response to Environmental Risk. Boston: Kluwer Academic. Entman, Robert M. (1997, June 20). The media and public health scores. Report to the Advancement of Sound Science Coalition. ESCOP/Gallup Surveys of Public Attitudes Toward Pesticides. (1994, October). Unpublished survey results presented to NCR-90 Agricultural Research Committee, Michigan State University. Fidler, Roger. (1997). Mediamorphosis: Understanding the New Media. Pine Forge Press. Gaskell, George; Bauer, Martin W.; Durant, John, and Allum, Nicholas. (1999, July 16). Worlds apart? The reception of genetically modified foods in Europe and the U.S. Science 285: 384-387. Gonzenbach, William J. and McGavin, Lee. (1997). A brief history of time: A methdological analysis of agenda setting; pp. 115-136 in McCombs, Maxwell; Shaw, Donald L. & Weaver, David. Communication and Democracy. Mahwah, NJ: Lawrence Erlbaum Associates. Hoban, Thomas J.(1995) The construction of food biotechnology as a social issue. In Maurer, Donna and Sobal, Jeffrey, Eating agendas: Food and nutrition as social problems. (pp. 189-212). New York: Aldine de Gruyter. Holander, Susan. (2001, Jan. 9). Special presentation: the food industry view on GMOs, Illinois Crop Technology Conference, University of Illinois, Urbana. Hopkin, Karen. (2001, April). The risks on the table. Scientific American 284:4, 60-61. Kasperson, R.E. (1992). The social amplification of risk: Progress in developing an integrative framework. In Krimsky, S. and Golding, D. Social Theories of Risk (pp.153-178). Westport CT: Praeger. Klopfenstein, Bruce Carl. (1985). Forecasting the market for home video players: A retrospective analysis. Unpublished Ph.D. dissertation. The Ohio State University. Mazur, Allan. (1996). Why do we worry about trace poisons? Risk, Health, Safety and Environment Vol 7. pp. 35+. McCombs, M.E. & Shaw, D. (1972). The agenda-setting function of the mass media. Public Opinion Quarterly 36: 176-185. Miller, David and Reilly, Jacquie. (1995). Making an issue of food safety: The media, pressure groups, and the public sphere. In Maurer, Donna and Sobal, Jeffrey, Eating agendas: Food and nutrition as social problems. (pp. 305-336). New York: Aldine de Gruyter. Nelkin, Dorothy. (1994). Reporting risk: The case of silicone breast implants. Risk, Health, Safety and Environment. Vol 5. pp. 233+ Neuman, W. Russell. (1990). The threshold of public attention. Public Opinion Quarterly 54:159-176. Priest, Susana Hornig. (1995, Winter). Information equity, public understanding of science and the biotechnology debate. Journal of Communication 45:1 39-54. Rogers, Everett M.; Dearing, James W.; Chang, Soonbum. (1991, April). AIDS in the 1980s: The agenda-setting process for a public issue. Journalism Monographs No. 126. Sandman, Peter M. (1994). Mass media and environmental risk: Seven principles. Risk, Health, Safety and Environment Vol. 5. pp. 251+. Slovic, Paul. (1992). Perception of risk: Reflections on the Psychometric Paradigm; pp. 117-152 in Sheldon Krimsky and Dominic Golding (eds.) Social Theories of Risk. Westport, CT.: Praeger. Subcommittee on Basic Research. (2000, April 13). Seeds of opportunity: An assessment of the benefits, safety and oversight of plant genomics and agricultural biotechnology. Committee print 106-B, One-hundred sixth U.S. Congress. Washington, D.C. Table 1: Sampling Periods and Numbers for Each Paper Pre-hoopla Hoopla Post-Hoopla Second Hoopla NY Times 1/97 - 4/30/99 n=28 5/1/99 - 12/31/99 n=44 1/1/00 - 3/31/00 n=27 4/1/00 - 12/31/00 n=109 London Times 1/97 - 12/98 n=60 1/99 - 10/99 n=47 11/99 - 3/00 n=48 4/00 - 12/00 n=94 Daily Mail 1/97 - 12/98 n=49 1/99 - 10/99 n=50 11/99 - 3/00 n=49 4/00 - 12/00 n=84 Table 2: Mean score: Total number of sources used by hoopla period Daily Mail London Times NY Times Overall Pre-hoopla 2.22 2.10 2.79 2.28 Hoopla 1.98 1.94 3.30 2.38 Post-hoopla 1.73 1.42* 2.74 1.83* Second hoopla 2.19 1.86 2.52 2.21 Overall Mean 2.06 1.85 2.75 2.19 Significance .215 .034 .115 .020 * indicates the mean is significantly different from the high value by a scheffe test. Table 3: Percentage of Articles Using Sources Across Periods by Newspaper Daily Mail London Times NY Times Overall Public Officials Pre-hoopla 81.6% 46.7% 46.4% 59.1% Hoopla 78.0% 51.1% 52.3% 61.0% Post-hoopla 61.2% 37.5% 66.7% 53.2% Second Hoopla 57.1% 50.0% 61.5% 56.45 Significance .007 .495 .320 .594 Scientists Pre-hoopla 28.6% 33.3% 21.4% 29.2% Hoopla 22.0% 19.1% 31.8% 24.1% Post-hoopla 6.1% 10.4% 25.9% 12.1% Second Hoopla 13.1% 14.9% 21.1% 16.7% Significance .014 .011 .546 .001 Industry Pre-hoopla 24.5% 31.7% 64.3% 35.8% Hoopla 16.0% 36.2% 56.8% 35.5% Post-hoopla 26.5% 27.1% 40.7% 29.8% Second hoopla 23.8% 20.2% 50.5% 32.8% Significance .609 .186 .312 .711 Citizens' Groups Pre-hoopla 30.6% 30.0% 21.4% 28.5% Hoopla 38.0% 25.5% 45.5% 36.2% Post-hoopla 34.7% 20.8% 37.0% 29.8% Second Hoopla 45.2% 35.1% 36.7% 38.7% Significance .361 .317 .234 .123 Farmers Pre-hoopla 6.1% 6.7% 21.4% 6.6% Hoopla 8.0% 10.6% 31.8% 14.2% Post-hoopla 4.1% 4.2% 25.9% 3.2% Second Hoopla 16.7% 11.7% 21.1% 13.9% Significance .007 .424 .546 .002 Table 4: Pooled Mean Scores of Source Use by Period Pooled source use: all scientists and all industry Pooled source use: all other sources Pre-hoopla .65 1.38 Hoopla .60 1.48 Post-hoopla .42 1.15 Second Hoopla .49 1.38 Significance F=3.9 p=.009 F=2.27 p=.079 Table 5: Total Number of Themes by Period by Newspaper Daily Mail London Times NY Times Overall Pre-hoopla 2.08 1.42 2.46 1.87 Hoopla 1.50 1.19 2.27 1.64 Post-hoopla 1.08* 1.23 2.00 1.34* Second hoopla 1.89 1.38 2.40 1.92 Mean Score 1.67 1.33 2.33 1.75 Significance p<.000 .603 .647 .001 * indicates significantly different than highest value by scheffe test. Table 6: Number of Negative and Positive Themes by Period by Newspaper A. Negative Themes Daily Mail London Times NY Times Overall Pre-hoopla 1.10 .68 1.00 .90 Hoopla .82 .60 1.30 .89 Post-hoopla .75* .71 .89 .77 Second Hoopla 1.24 .79 1.00 1.00 Mean Score 1.02 .71 1.05 .92 Significance .014 .541 .357 .133 B. Positive Themes Daily Mail London Times NY Times Overall Pre-hoopla .43 .42 1.32 .61 Hoopla .18 .49 .89 .50 Post-hoopla .16* .35 .67 .34* Second Hoopla .54 .42 .88 .63 Mean Score .36 .42 .91 .55 Significance .008 .809 .075 .011 * indicates significantly different than highest value by scheffe test. Table 7: Shifts in Theme Topics and Positive or Negative Valence Over Time Theme Daily Mail London Times NY Times Pre-hoopla 1st Rank Human Health - 42.9% Regulatory 33.3% Business + 71.4% 2nd Rank Regulatory 40.8% Human Health + 26.7% Human Health + 39.3% 3rd Rank Moral - 38.8% Business - 25.0% Farming + 35.7% Hoopla 1st Rank Regulatory 42.0% Business - 40.4% Business + 54.5% 2nd Rank Moral - 32.0% Regulatory 25.5% Environ/ Farming - + 40.9% 40.9% 3rd Rank Business - 28.0% Human Health/ Environ/ Farmer/ Moral + + + - 12.8% 12.8% 12.8% 12.8% Post-hoopla 1st Rank Business - 24.5% Business - 27.1% Regulatory 48.1% 2nd Rank Moral - 22.4% Regulatory 20.8% Environ/ Farming/ Business - + - 29.6% 29.6% 29.6% 3rd Rank Regulatory 20.4% Environ/ Farming - + 20.8% Second Hoopla 1st Rank Environ - 46.4% Moral - 33.0% Human Health + 56.0% 2nd Rank Moral - 44.0% Environ - 30.9% Regulatory 48.6% 3rd Rank Human Health + 32.0% Human Health tie 22.3% Business - 37.6% Note: Table indicates the rank order of themes by the percentage of stories they appear in. Valence indicates whether the majority of articles on that theme were negative or positive. The percentage indicates the percent of articles containing that theme. Table 8: Pooled Theme Mean Scores by Time Period Human Health/ Environment Business/ Farming Regulatory/ Labeling Moral Pre-Hoopla .54 .55 .55 .23 Hoopla .39 .61 .43 .21 Post-hoopla .35 .44 .36 .18 Second Hoopla .75 .47 .41 .29 Significance p<.000 .101 .080 .07 Figure 1: Total number of articles concerning GM Foods and similar topics by month in three newspapers: The New York Times, The Times of London, and The Daily Mail, 1997-2000.