Center For American Politics and Public Policy Minority Politics Program Occasional Paper Series (95-1) Crime in Black and White: The Violent, Scary World of Local News Franklin D. Gilliam, Jr., Shanto Iyengar, Adam Simon, and Oliver Wright Department of Political Science University of California, Los Angeles September, 1995 Center for American Politics and Public Policy University of California, Los Angeles 810 GSLIS Building, 405 Hilgard Ave. Los Angeles, California 90024 (810) 206-8109 2801 M Street, NW, 5th Floor Washington, D.C. 20087 (202) 296-8226 Abstract Crime is central to the public debate about the state of American society. Citizens express great concern about crime and are increasingly calling for punitive policies such as "three strikes" and the death penalty. In response, politicians and policy makers have allocated larger and larger shares of their budgets to crime control. This is ironic given that the population adjusted crime rate has declined in recent years. Our paper addresses this paradox by focusing on the role of television news. A content analysis of local television news in a major media market demonstrates that coverage of crime features two important cues -- crime is violent and criminals are nonwhite. We translate these media biases into an experimental design that manipulates the level of violence and the race of the perpetrator to test the relevance of these cues to public thinking about crime. The results indicate that race works independently and in conjunction with racial stereotypes to influence people's concern about crime and their willingness to attribute criminal behavior to breakdowns in the African-American community. We conclude by discussing the implications of our findings for race relations, the practice of journalism, and public policy. Today more Americans name crime as the "most important problem facing the country" than any other issue. Never before (at least since the advent of polling) have we been so preoccupied with issues of public safety (see Warr, 1995). As fears have increased, so have punitive attitudes toward criminals. The death penalty, mandatory jail sentences, adult trial of juvenile offenders, and "three strikes" laws are all embraced by large majorities (Alderman, 1994;Pettinico, 1994; Warr, 1995). In response to the public outcry, policy makers at all levels of government have allocated massive public resources to fortify law enforcement and the criminal justice system. Recent developments in California are revealing: over the past decade, the annual budget of the Department of Corrections has increased at Ave times the rate of increase granted the Departmentof Education. (The annual budget now stands at $4.5 billion.) Surprisingly, this outpouring of collective attention has occurred in an environment which has shown little actual change in the frequency of criminal activity. The statistics simply do not reflect any significant increase in crime. More specifically, two important national databases -- the National Victimization Survey and the Uniform Crime Reports -- both indicate that" the population-adjusted rate of crime (whether defined as total crime or violent crime) has in fact declined over the last two decades. Moreover, the great majority of Americans do not experience crime directly (Beckett, 1993; Alderman, 1994; Gilliam, 1995; California Legislative Analyst's Office, 1995). What, then, is the source of the public's rising preoccupation with crime? If the public's concern is not based on first-hand experience, it must stem from vicarious encounters -- usually through television reports (Alderman, 1994). In general, it has been well documented that the public political agenda is heavily influenced by patterns of news coverage (see Iyengar and Kinder,1987); as the media become preoccupied with particular issues, so too does the public. While the sheer quantity of news coverage is important in determining the level of public concern for crime, we believe that qualitative characteristics of news coverage also have a major impact. Crime has continuously accounted for a heavy share of broadcast news programming over the last decade (Freeman, 1994). A content analysis of national newscasts between 1981 and 1986, for example, found that the ABC, CBS, and NBC television networks broadcast more than 1100 reports on crime, for an average of more than fifteen stories per month Iyengar, 1991). Moreover, news coverage of garden variety crime was surpassed only by coverage of crimes of terrorism.[1] Since the sheer volume of news coverage of crime has not risen noticeably in recent years, the public's current fixation on crime must be attributable to alternative forces. We suspect that two qualitative features of news programming -- violence and race -- are especially important. Television's insatiable demand for "good pictures" and riveting stories means that the most gruesome or notorious episodes of crime receive extensive attention while other forms of crime are virtually ignored. Local television news, the most widely used source of information about crime, is especially prone to dwell on crimes of violence (see Fairness and Accuracy in Reporting, 1994; Dorfman et al., 1995) and the print media is no less taken with scandal and gore (see Elias, 1994). In short, no matter what they view or what they read, Americans are bombarded with information about sensational crimes -- O.J. Simpson, the Menendez brothers, Colin Ferguson, Oklahoma City, the World Trade Center, and so on. Perhaps the public's increased concern with crime is a response to the perception (reinforced on a daily basis) that American society is increasingly anomic and violence-prone (see Wilson, 1975). Not only does news coverage highlight violent crime, it also links the issues of race and crime by overrepresenting minorities in the role of violent criminals and by according them distinctive forms of coverage. Iyengar's study of network news found that crime in black neighborhoods accounted for a substantial share of television news programming Iyengar, 1991). Elias' detailed study of the major news magazines yielded similar results, leading the author to conclude that "criminals are conceptualized as black people, and crime as the violence they do to whites" (Elias, 1994, p. 5; also see Campbell, 1995). Entman's analysis of local news also revealed clear race-based pattern. While black suspects were usually-shown in handcuffs, and in the custody of police officers, white suspects were typically seen with their attorneys (Entman, 1990, 1992). In summary, the typical news story on crime consists of two "scripts": crime is violent, and criminals are non-white. These pervasive themes cry out for further investigation. In this paper, we begin by matching content analysis of local television news stories with aggregate data on the frequency of criminal activity and the ethnicity of criminal perpetrators to evaluate the accuracy of the media's reflection of "real world" criminality. While prior research has failed to assess media coverage of crime against some objective norm, our results demonstrate that local news programs are significantly distorted in two respects; they disproportionately portray crimes of violence, and they overrepresent African-Americans as perpetrators of violent crime. Using an experimental design, we then estimate the effects of each of these biases on viewers' beliefs and opinions about crime. Our experimental results suggest that the first distortion (the news media's focus on violent crime) is not, by and large, responsible for the public's heightened fear of crime. The second distortion, however (concerning the race of the perpetrator) is a much more significant cue: exposure to criminal activity committed by non-whites in and of itself makes viewers more concerned about crime. Most importantly, the media has, in effect, defined crime in racial terms, and this serves to activate widely-shared stereotypes about racial minorities.[2] These stereotypes then become connected to viewers' opinions about crime (for reviews of the role of stereotyping in social judgment, see Taylor, 1981; Krueger and Rothbart, 1988; Pratto and Bargh, 1991). In effect, viewers are "primed" to consider crime through the lens of their racial stereotypes.[3] Violence and Race in Local News Coverage of Crime Our sample of local news reports was drawn from television station KABC the ABC network affiliate in Los Angeles (Channel 7). We examined news coverage over a thirteen month period, from March 1993 through March 1994. We randomly selected two days from each week and in the period viewed KABC's "Action News" program, a thirty minute newscast that airs at 6:00 p.m.[4] Crime was extensively covered during the entire period under examination. The 148 broadcasts sampled contained 436 stories dealing with crime, or an average of three stories per day. During the entire period sampled, there were only five newscasts that failed to include a single story on crime.[5] Since the entire "news" section occupies sixteen minutes of the newscast, the total coverage accorded crime (on average, four minutes per day) accounts for at least twenty-five percent of the daily news. In addition, crime was the focus of the lead story in 51 percent of the sampled newscasts. Based on the content-analytic evidence cited earlier, we suspected that the news would be dominated by crimes of violence, and that the race of the alleged criminals would be disproportionately nonwhite. Accordingly, we tabulated the number of news stories and the duration of coverage accorded violent and nonviolent crime. When the news story provided sufficient information, (which occurred in 40 percent of the stories), we classified the alleged perpetrator as either white, Hispanic, or black.[6] Figure 1 compares the percentage of news reports accorded violent crime with the actual frequency of violent crime in Los Angeles (expressed as a percentage of all crimes).[7] The high level of violence was as expected: the overwhelming majority of news reports were episodic in nature and featured acts of violent crime. But are these reports an accurate reflection of reality? The "mirror image" argument would predict that the bars on the graph should be of equal size. This was clearly not the case. Violent crime made up 30 percent of all crimes in Los Angeles County, but 78 percent of the news reports aired by KABC. The discrepancy was even more extreme when we focused solely on homicide. Murder accounted for only two percent of felony incidents in Los Angeles, but 27 percent of the news coverage was directed at murder. Thus, the news exaggerated the frequency of murder by a ratio of 14:1. While the special seriousness and newsworthiness of murder cannot be denied, the level of the "distortion" is impressive. (Figure 1 here) As mentioned above, the race of the perpetrator was identified in approximately 40 percent (N= 143 stories) of the news stories. While nonwhites made up 55 percent of the violent criminals depicted in the news, 75 percent of news reports about nonviolent perpetrators featured whites. This pattern is hardly the basis for claiming racial bias, however. The crime rate varies considerably across different ethnic groups and must be interpreted according to the groups' relative representation in the population. In Los Angeles County, for example, whites account for a smaller percentage of all violent crime than blacks even though they are the-larger group. To assess these differential rates of criminal activity, we computed population-adjusted real world and media-depicted crime rates for each racial group. An adjusted crime rate of 1.0, for example, would indicate that the group in question commits crime in exact proportion to its share of the population (i.e., a group making up ten percent of the population that accounts for ten percent of the crime). Population-adjusted rates that exceed 1.0 indicate "overrepresentation" of particular groups in criminal behavior -- in the real world and news presentations respectively -- and vice versa. By comparing each group's actual adjusted crime rate with their media- depicted adjusted crime rate, we can assess the extent of racial bias in news coverage. These comparisons are provided in Figure 2. (Figure 2 here) The population-adjusted crime rate (both actual and as represented in the news) is plotted along the vertical axis. In terms of actual rates, blacks commit violent crime at the highest rate -- some 2.8 times their share of the population. The Hispanic crime rate is also higher than expected, but by a much smaller margin -- .20. Finally, whites are involved in violent crime much less frequently (by a factor of .58) than might be expected, given their population. Contrary to our expectations, nonwhites in the aggregate were not substantially overrepresented in news coverage of violent crime. However, the African-American media crime rate was slightly exaggerated -- from 2.8 to 3.2 times their share of the population, while Hispanics were significantly underrepresented in the news; they committed crime at 20 percent above their expected rate, but appeared in the news as violent perpetrators 50 percent less frequently than expected. Also contrary to expectations, white perpetrators were actually overregresented in the news. While the actual rate of white violent crime was considerably lower than the expected rate, the rate of white violent crime in the news equalled the expected rate. The significant underrepresentation of Hispanics can be attributed to organizational factors. Television stations in Los Angeles employ relatively few reporters who are Hispanic or who can speak Spanish (for anecdotal evidence, see Altschull, 1995; Campbell, 1995). Coverage of crime in Hispanic areas is thus more difficult for local stations. The relatively high level of coverage accorded white perpetrators may reflect the producers' desire to appear racially unbiased. Alternatively, since most white violent crime occurred in relatively affluent suburban areas, it may have attracted greater journalistic attention. Regardless of the explanation, coverage of violent crime, by itself, provides an incomplete test of racial biases in the news. What about nonviolent crime? Although news of nonviolent crime is less frequent, the coverage may be allocated disproportionately to particular groups. Figure 3 shows the population-adjusted rates of nonviolent crime for blacks, Hispanics and whites. (Figure 3 here) These slopes reveal considerable racial bias. While the black rate of nonviolent crime is just as disproportionate as their rate of violent crime (slightly more than 3 times the expected rate), the black nonviolent crime rate in the news is almost exactly proportional to their share of the population. In effect, the news dramatically underrepresents black nonviolent crime and slightly overrepresents black violent crime. Whites, on the other hand, are noticeably overrepresented as perpetrators of nonviolent crime (in the news they appear at a frequency that is 200 percent above the expected rate despite the fact that their actual participation is 50 percent below the expected rate) and, also overrepresented as perpetrators of violent crime. Hispanics received less crime coverage than they deserve for both nonviolent and violent crime. In order to summarize the comparisons between violent and nonviolent crime, we calculated a "net violence" index. First, we subtracted each group's relative contribution to nonviolent crime from their contribution to violent crime. Next, we subtracted their share of nonviolent crime in the news from their share of violent crime coverage. Finally, we took the difference of these differences.[8] A score of zero on the net violence index means that the mix of news coverage accorded a particular group exactly matches that group's actual mix of criminal behavior. Positive scores indicate that the group is depicted in the news as more violent than warranted by their actual behavior. Finally, negative scores would indicate the opposite -- that news coverage of the group is less violent than it should be. The net violence index for blacks, Hispanics and whites is presented in Figure 4. (Figure 4 here) The results in Figure 4 indicate that while blacks commit violent and nonviolent crime at about the same rate, the media coverage of black crime is distinctly more violent than nonviolent (by a factor of 22 percent). Conversely, media coverage of white crime is distinctly more nonviolent than violent (by a factor of 31 percent) even though whites are only slightly less likely (by seven percent) to engage in violent rather than nonviolent crime. Finally, while Hispanics are seven percent more likely to engage in violent crime, the news depicts them as 14 percent more likely to be violent. In sum, the media overrepresents black violent crime significantly, overrepresents Hispanic violent crime slightly, and underrepresents white violent crime dramatically. In addition to the amount of coverage allocated to violent crime and particular perpetrators, we also examined the pattern of sources used in the news reports. In particular, we counted the number of interviews-with individuals who represented the victim's or perpetrator's perspective. Stories with an equal number of sources from each side were considered neutral in slant, those featuring a plurality of victim-oriented sources were considered relatively hostile, and those in which the balance of sources favored the perpetrator were considered relatively sympathetic. Finally, we also coded the reporter's comments for evidence of editorial or evaluative content. In a few cases (15 percent of the stories), the reporter saw fit to express outrage or some other affective response over the brutality of criminal act or the plight of the victim. The racial differences in these qualitative indicators are presented in Figure 5. The vertical bars represent the frequency of news stories relying on "hostile" sources and the frequency of critical editorial comments by the reporter. (Figure 5 here) When the news story featured a black perpetrator, the reporter relied on sources hostile to the perpetrator nearly 50 percent of the time. Hispanic perpetrators were accorded similar treatment with 41 percent of their coverage including unsympathetic sources. Whites enjoyed "most favored perpetrator" treatment -- only one in four stories dealing with white crime relied on hostile sources. In this respect, coverage of white crime significantly deviated from coverage of black and Hispanic crime. The same pattern emerged with newscasters' commentary. Editorial comments were most frequent when the perpetrator was black, least frequent when the perpetrator was white, and in-between when the perpetrator was Hispanic. The difference between blacks and whites was statistically significant. To recapitulate, our year-long content analysis of KABC yielded two core findings: first, the local news overrepresents violent crime in general by a factor of more than 2:1, and overrepresents murder by a factor of more than 10:1. Second, the news tends to exaggerate existing racial differences in actual crime rates by disproportionately depicting blacks in the role of violent perpetrators and whites as nonviolent perpetrators. Hispanic perpetrators -- both violent and nonviolent -- remained relatively invisible (for a more general discussion of Hispanics in the news, see Chavez, 1992). In effect, the news depicts crime in "black and white." Experimental Design Both findings from the content analysis -- the exaggeration of violent crime and the tendency of the news to associate blacks with violent crime and whites with nonviolent crime -- were converted into experimental manipulations so as to assess their impact on viewers' beliefs and opinions about crime. Experimentation provides a rigorous as well as realistic experimental paradigm for examining the effects of news coverage of crime. The method developed here consisted of two separate manipulations, (1) the level of criminal violence, and (2) the race of the alleged perpetrator. Study participants were exposed to a news report that described either a violent or non-violent crime in which the perpetrator was either a white or black male. The most innovative aspect of this design concerned its ability to overcome the problem of confounded or correlated variables.[9] A computer- based technique was used to alter specific physical attributes of an individual (particularly skin color and related features). The original "input" was a local news report which included a close-up "mug shot" of the suspected perpetrator of the crime in question. The picture was digitized, then "painted" to alter the perpetrator's complexion, and then re-edited into the news report. Beginning with two different perpetrators (a white male and a black male),[10] the technique produces painted versions of each individual in which their race is reversed, but all other features are identical. Using this method; any differences in the responses of the subjects exposed to the white or black perpetrators can only be attributed to the perpetrator's race.[11] We conducted a study based on this design in August, 1994. Experimental participants were exposed to a fifteen-minute videotaped local newscast (including commercials) that included a report on crime. Participants were told that the newscast they were to watch had been selected at random from news programs broadcast during the past week and that the study concerned "selective perception" of news reports. Depending upon the condition to which they were assigned (at random), they watched a report on a murder at an automated teller machine or a report on a high school baseball coach who embezzled funds from the school's athletic budget. For both violent and non-violent crime, the news report featured a close-up photo of either a black or white suspect (using the method described above). Except for the news story on crime, the newscast was identical in all other respects. None of the remaining stories on the tape concerned crime or matters of race. The experimental treatment was inserted into the middle of the tape following the first commercial break. The experimental "sample" consisted primarily of university administrative and clerical staff (students were excluded) who were recruited through flyers and announcements in newsletters offering $10 for participation in "media research." The age of the participants ranged from 18 to 64. Forty-two percent were white, 24 percent were black, 14 percent were Asian, and 14 percent were Latinos. Fifty-two percent were women. The participants were relatively well educated (40 percent had graduated from college) and, in keeping with the local area, more Democratic than Republican (48 percent versus 28 percent) in their partisan orientation. The study was administered at the UCLA Communications Research Laboratory (which consists of a two-room suite on campus). On arrival, participants were instructed that the study concerned "selective perception" of local newscasts, were given a short pretest questionnaire concerning their social background, party identification and political ideology, level of interest in political affairs, and media habits. They then watched the videotape. Most participants came accompanied with friends or work associates, thus adding an air of informality to the viewing sessions.[12] At the end of the videotape, subjects completed a lengthy questionnaire that included questions about the significance of crime, the causes of crime, their preferred methods for dealing with the problem, and their stereotypes of various social groups including African-Americans. After completing the questionnaire, subjects were debriefed in full (including a full explanation of the experimental procedures) and were paid the sum of $10. Analysis and Results To determine the effects of the experimental manipulations on attitudes toward crime, the questionnaire probed subjects' fear of violent crime, their explanations for rising crime, and, finally, their support for punitive policy measures to remedy crime.[13] We also attempted to assess participants' stereotypic beliefs about blacks. The questionnaire asked subjects to rate blacks (and several other groups) in terms of a wide range of attributes three of which were especially relevant to the issue of crime. These were "tend to be violent," "tend to be law abiding," and "tend to be sexually aggressive." We summed these three ratings to form an index of racial stereotyping.[14] Both experimental manipulations were specified as trichotomies with the control or null group (which saw no news coverage of crime at all) scored as zero. The conditions featuring violent crime were scored as +1 and those presenting coverage of nonviolent crime were scored as -1. Similarly, exposure to a black perpetrator was scored as +1 and exposure to the white perpetrator was scored as -1. The effects of these scores are indicative of the "main effects" of violence and race respectively -- the degree to which viewers' opinions are affected by the type of crime and the race of the perpetrator. In addition to the main effects of the manipulations, we were also interested in the interaction between the measure of racial stereotyping and exposure to black perpetrators. If news coverage of blacks engaging in criminal activity serves to activate longstanding racial stereotypes, this interaction should be positive and significant, revealing that viewers who see news of a black perpetrator are more apt to bring their opinions about crime in line with their beliefs about blacks. We also examined the interaction between the type of crime and the race of the perpetrator, but found it to be nonsignificant. It was thus excluded from the analyses that follow.[15] Finally, we controlled for several factors thought to influence opinions about crime and punishment including gender, party identification, political ideology, and fear of being victimized. Only two of these controls (gender and fear of victimization) were consistently influential and the rest were dropped from consideration. In sum, our analysis of viewers' concern for violent crime, causal attributions, and their support for punitive remedies is based on the following model: Crime Opinion = b1 Violent Crime + b2Black Perpetrator + b3Index of Racial Stereotyping + b4Index of Stereotyping * Black Perpetrator + b5Gender + b6Fear of Victimization If public concern about-crime, attributions of causality, and preferences for more punitive policies are influenced simply by news depictions of violent crime, we should find a significant positive coefficient for the level of crime manipulation. Similarly, if crime attitudes are inflamed by racial imagery in the news, we should find that exposure to a black perpetrator (regardless of the level of violence) evokes greater levels of concern about crime, a willingness to attribute crime to characteristics of the black community, and more support for harsher policies. Our third expectation concerns the interaction between racial stereotypes and exposure to a black perpetrator. If news coverage of black perpetrators serves to prime viewers' with their racial stereotypes, the interaction effect should be strong and positive -- seeing a black perpetrator committing a crime should activate widely-held stereotypes about black criminality, which, in turn, produce more concern, a greater frequency to cite breakdowns in the black community as causes of crime, and higher levels of support for "get tough" policies. Table 1 presents multiple regression analyses of the effects of our experimental manipulations on public attitudes about crime and criminal justice policy. Reading from left to right, the results refer to public concern about crime, causal attributions, and support for punitive measures. (Table l here) ---------------------------------------------------------------- Table 1. Effects of Experimental Manipulations and Racial Stereotypes on Attitudes About Crime Concern Causal Attribution: Support for for Crime Group Characteristics Punitive Measures Predictors: Type of Crime .05 -.16 -.06 (.10) (.34) (.44) Race of Perpetrator .13* .48** -.19 (.05) (.18) (.23) Racial Stereotypes .06 .42* .92** (.06) (.20) (.25) Perpetrator X .11* .43* -.24 Stereotypes (.06) (.21) (.28) Gender .11 .62* -.03 (.09) (.33) (.43) Fear of Victimization .24* -.42 -.40 (.10) (.37) (.47) Constant 2.38 4.93 6.91 (.08) (.33) (.41) Adjusted R2 .07 .11 .07 Number of cases 139 139 139 Table entries are unstandardized regression coefficients with standard errors in parentheses. Asterisks represent one-tailed significance levels: * p < .05, **p < .01. The independent effect of the race of the criminal perpetrator manipulation was positive and significant in two of the three equations. Exposure to a black perpetrator (as opposed to a white perpetrator) yielded an increase of .26 in the concern for crime measure and a .96 increase in the index of causal attribution. The corresponding main effect of the level of violence manipulation failed to generate a robust coefficient in any of the equations. The predicted interaction between the race of the perpetrator manipulation and racial stereotypes emerged clearly in the equations for the level of concern and causal-attributions. There was little difference in the level of concern for crime among low stereotypers exposed to the black or white perpetrator. Among high stereotypers, however, exposure to the black rather than white perpetrator boosted concern about crime by a factor of .30. The priming effect of exposure to the black perpetrator was no less prominent in the case of causal attributions for crime; high stereotypers exposed to a black perpetrator were much more likely than low stereotypers to cite breakdown of the family and religious values in the black community as causes of crime. In addition to influencing viewers' concern for crime and their explanations for rising crime in tandem with the news, racial stereotypes also contributed on their own. Viewers with more negative stereotypes of blacks were significantly more likely to offer group-based attributions of responsibility and to favor punitive policies.[16] The latter effect was especially powerful -- the difference between low and high stereotypers on the index amounted to nearly two points (on a three-point scale)! In sum, our experimental results underscore the importance of news coverage to public opinion about crime. However, it is race and not violence that is the more important element of crime news coverage. The level of violence in news coverage of crime had no discernible effect on viewers' opinions. However, racial imagery in the news triggered fear of crime and a willingness to hold black people as responsible for crime. A mere three second exposure to a picture of a black perpetrator in a local newscast proved sufficient to boost the level of concern for crime and affect attributions of causal responsibility. The presence of racial cues in the news also activated stereotypic beliefs about African-Americans as antecedents of opinions about crime. Conclusion The media has contributed to the current furor over crime. Most Americans do not experience crime directly. They do, however, receive huge doses of crime coverage from the media, especially television news. The market pressures that attract local stations to crime are understandable; our results suggest that these pressures carry a high price tag. Television's fixation on crime means that it cannot provide adequate coverage to a number of other important social and political issues. As a public trustee, television news is obligated to provide the audience with information about issues that affect the quality of their lives. In Los Angeles these pressing issues include a crumbling physical infrastructure, deteriorating public schools, a stagnant economy, and racial divisions. None of these issues received even a fraction of the coverage accorded crime, If local news is to serve the community, it must be balanced in its coverage of public issues. Our results also imply that when the media do address crime, the coverage should be more thematic or contextualized. While particular episodes of violent crime may be newsworthy, violent crime only accounts for some one-third of all crime. The overreliance on "body bag" journalism distorts reality. Television news coverage of crime further distorts reality by exaggerating racial differences in the propensity to commit different types of crime. This distortion is likely to impede thoughtful discussion of racial divisions in contemporary American society. Construing the "crime problem" as a "black problem" plays on the public's worst fears and plays into the hand of politicians who are only too happy to exploit these fears. The image of the "savage Sambo" has no place in contemporary public discourse. The implications for journalistic policy are no less clear. How crime is covered matters! Some media outlets have begun to take steps to reduce the level of violence in their newscasts. Television station WCCO in Minneapolis, for instance, has created a "family sensitive" newscast that eliminates graphic depictions of violent crime. Our findings suggest that journalists would be better served by scaling back on the use of racial imagery as well. Presenting footage of the perpetrator may be a public service if the suspect remains at large, but when the suspect is under arrest, the newsworthiness of his race is questionable and serves only to polarize the audience. To conclude, the research reported here has important implications for public opinion research, the practice of broadcast journalism, and the policy process. Our research paradigm enables systematic investigation of the linkages between news coverage, racial attitudes and attitudes about crime and punishment. In the future, we hope to extend the paradigm to examine other racially charged issues including social welfare, affirmative action, and immigration. ENDNOTES [1]More recent studies reveal similar levels of stability (at high levels) in news coverage of crime. For example, between March 1993 and February 1994, local news coverage of crime provided by one Los Angeles television station remained steady at approximately forty stories per month (Wright, 1994). [2]For similar results with survey evidence, see Hurwitz and Peffley, 1995. [3]For a general discussion of the priming effects of television news coverage, see Iyengar and Kinder, 1987. [4]The videotapes of the newscasts were provided by the UCLA Film and Television Archive and were viewed independently by three student coders who were "blind" to the objectives of the study. [5]In these cases, news coverage of a natural disaster crowded out all other news: [6]Mixed-race individuals and "others" were excluded because of a paucity of cases. We also excluded Asians because of they accounted for a very small share of the news coverage and because of the multiplicity of Asian sub-groups. [7]Where possible, we matched the news stories about crime to the standard categories recorded by criminal justice agencies. The category of violent crime included murder, robbery, assault, carjacking, drive-by shooting, rape, domestic violence, and kidnapping. Nonviolent crime included burglary, narcotics, forgery, fraud, vehicle theft, arson, vandalism, and "hate" crimes. [8]The formula for calculating the net violence index is shown below. Net Violence Index = (% of violent crime committed by Group; - % of nonviolent crime committed by Group;) - (% of violent crime news stories featuring Group; - % of nonviolent crime news stories featuring Group;). [9]The problem is inherent when researchers manipulate race using different stimulus individuals. For example, previous work on racial cues in news coverage has relied on news reports in which the race of some "target" individual was varied. Iyengar (1991) showed his participants news reports of an unemployed black man and an unemployed white man and news stories about crime featuring a white or black perpetrator. Since the individuals featured in the news differed in several respects other than race or ethnicity, Iyengar's studies provided only weak tests of the effects of race. [10]The "originals," obviously, were taken from two different news reports. [11]The validity of this method, of course, can be questioned. We have not taken it for granted. Instead, we pilot tested the original and transformed versions of two different male suspects (one white, one black) with UCLA undergraduates. The students (N=90) were shown the four pictures (on a computer screen) along with a series of other pictures. As part of a "facial memory" test, the students were asked to indicate the ethnicity of each individual presented. In addition to accuracy of racial identification, we measured response latency on the assumption that lower latency would indicate greater confidence in the "target" individual's race. The results of this pretest were encouraging. In both instances, the level of accuracy for the original and painted versions of the target were equivalent (.93 versus .87 and .84 versus .83 for the two cases). Response latency was also uniform across the original and altered faces. Latency was slightly higher in the case of the altered photos, but in neither case was the difference significant. [12]We added further to the informality of the viewing environment by providing coffee and cookies and by furnishing the viewing room to resemble a typical living or family room. [13]The questions were worded as follows: Concern for Violent Crime. Lately, there has been a lot of attention to the problem of random street violence in Los Angeles. How serious of a problem do you think random street violence is in Los Angeles? (The response options included "not at all serious," "somewhat serious," and "very serious.") Causal Attributions for Rising Crime. Please indicate whether you agree or disagree with the following explanations for the high rate of crime -- "breakup of the family" and "declining influence of religion." (The response options ranged from "strongly agree" to "strongly disagree.") Support for Punitive Measures. Respondents were asked to indicate whether they agreed or disagreed with the death penalty and "three strikes" laws. Answers to the two questions (which ranged from "strongly agree" to "strongly disagree") were summed to form an index. [14]Each rating was made along a seven-point scale. After reflecting the item on law-abidingness, we summed the three ratings. The resulting score was trichotomized into low, medium, and high categories. High scorers subscribe to stereotypic beliefs. [15]Because the arguments concerning racial imagery in the news are generally aimed at the predominantly white audience (Entman, 1992; Peffley and Hurwitz, 1993; Hurwitz and Peffley, 1995), our analyses were limited to white participants. Of course, it is possible that news coverage of crime also influences blacks and Hispanics. Our sample, however, included too few nonwhite participants to permit these more complex racial comparisons. We hope to address these comparisons in future work. [16]For a similar result, see Hurwitz and Peffley, 1995. References Alderman, J. 1994. "Leading the Public: The Media's Focus on Crime Shaped Sentiment", The Public Perspective, 5, 26-27. Altschull, J. H. 1995. Agents of Power: The Media and Public Policy. New York: Longmans. Beckett, K. 1993. The Politics of Law and Order: the State and the Wars Against Crime and Drugs. 1964 to the Present. Unpublished Doctoral Dissertation, UCLA. Campbell, C. P. 1995. Race. -Myth-and the.News. Thousand.Oaks, Ca. Sage Publications. California Legislative Analyst's Office. (1994). Crime in California. Sacramento, CA.: Legislative Analyst's Office. Chavez, L. 1992. Out of the Barrio: Towards a New Politics of Hispanic Assimilation. New York: Basic Books. Elias, R. 1994. "Official Stories: Media Coverage of American Crime Policy," Humanist, 54, 3-8. Entman, R. M. 1990. "Racism and local TV News," Critical Studies in Mass Communication, 7, 329-343. Entman, R. M. 1992. "Blacks in the News: Television, Modern Racism and Cultural Change," Journalism Quarterly, 69, 341-361. Dorfman, L., K. Woodruff, V. Chavez, and L. Wallack. 1995. "Youth and Violence on Local Television News," Unpublished Manuscript. Fairness and Accuracy in Reporting, Los Angeles Chapter. 1994. "If It Bleeds It Leads," News Release, June, 1994. Freeman, M. 1994. "Networks Doubled Crime Coverage in '93 Despite Flat Violence Levels in U. S. Society," Mediaweek, 4, 3-4. Gilliam, F. 1995. "The Color of Crime in California: Trends in Arrests, Dispositions, and Victimization", Unpublished Manuscript, UCLA. Hurwitz, J. and M. Peffley. 1995. "Public Perceptions of Race and Crime: The Role of Racial Stereotypes," Paper Delivered at the Annual Meeting of the Western Political Science Association, Portland, Oregon. Iyengar, S. 1991. Is Anyone Responsible? Chicago: University of Chicago Press. Iyengar, S., and D. R. Kinder. 1987. News that Matters. Chicago: University of Chicago Press. Krueger, J., and M. Rothbart.,19.88 .".Use of Categorical and Individuating Information in Making Inferences About Personality," Journal of Personality and Social Psychology, 55, 187-95. Peffley, M., and Hurwitz, J. 1993. "Racial Stereotypes and Their Political Impact," Paper Presented at the Annual Meeting of the American Political Science Association, Washington, D.C. Pettinico, G. 1994. "Crime and Punishment: America Changes Its Mind," The Public Perspective, 5, 29-32. Pratto, F., and J. Bargh. 1991. Stereotyping Based on Apparently Individuating Information: Trait and Global Components of Stereotypes Under Information Overload," Journal of Experimental Social Psychology, 27, 26-47. Taylor, S. 1981. "A categorization approach to; stereotyping," in D. Hamilton ed., Cognitive Processes in Stereotyping and Intergroup Behavior. Hillsdale, NJ: Erlbaum. Warr, M. 1995. "Public Opinion on Crime and Punishment," Public Opinion Quarterly, 59, 296-310. Wilson, J. Q. 1975. Thinking About Crime. New York: Basic Books. Wright, O. 1994. "Race, Crime and Broadcast News: A Content Analysis of Local News Coverage," Unpublished Undergraduate Honors Thesis, UCLA.