Section
The first analysis examines measures of the impact of news stories by the type of crime detailed in the articles. The results of this analysis are presented in Table 1.
The results indicate substantial differences among the measures of potential impact of news stories by the type of crime covered. The crime that was covered most in the articles examined in this research was robbery. A total of 58 articles, over 46% of those analyzed, reported on a criminal event which included robbery in a retail setting. The second highest level of coverage was for murder or homicide in a retail setting, with a total of 29 articles. Thus, about 23% of the articles in the study reported on a criminal event that included murder or homicide as part of the crime. There were 18 stories each that included the crimes of theft and kidnapping. The lowest number of stories was for larcenies, with only 5 of the 125 articles detailing a crime classified as larceny. There were 42 stories for “other” type crimes, which primarily included “carjacking” and drug and weapon offenses, among other crimes.
The average length of the stories varied from 450 to 554 words. Five of the types of crime had articles with average lengths of greater than 500 words. These crimes were theft, burglary, murder, kidnapping, and other crimes. The average story location scores varied a great deal, ranging from 2.4 for larceny to 6.8 for murder. So, as we might expect, unusual violent crimes like murder, robbery, rape, and kidnapping seem to be accorded more prominence in terms of article placement, length, and coverage.
The story impact score “Percentage of Total Articles” is particularly interesting when compared to actual crime rates (see Table 2). The most common Index crime in the U.S. is larceny/theft – nearly 60% of all Index crime in America can be classified as such (FBI 2000). Only 18.4% of the articles on crime in retail settings, however, related to larcenies or thefts. Even extrapolating to retail settings, we can make some generalizations about crime rates. Only 12% of violent crime occurs in commercial establishments, and only 34% of burglaries are of non-residences. Thus, crime in these settings is comparatively rare. As predicted by the literature, newspaper coverage of crimes in retail settings does not seem to mirror reality.
Kidnappings and abductions (particularly of adults) are so rare that they are generally not even indexed with other violent crimes (only 220 federal arrests for kidnapping were made in 1999), but over 14% of the articles under study focused on a kidnapping or abduction. Likewise, murder constitutes only 0.1% of all crime in the U.S. (and if 12% of this crime occurs in retail settings, then less than 0.01% of overall crime is murder in retail settings). Over 23% of the newspaper articles analyzed in this research reported on a homicide in a retail setting.
Actual crime rates for each city from which articles were analyzed were also compared to levels of crime reporting in the articles. Pearson correlations between reporting on a given crime and actual levels of that crime (for the specified city) further confirmed that there was no relationship between actual and reported crime. These analyses revealed only one significant correlation – there was a weak, significant negative correlation between actual burglary rates and reports of burglary, indicating an inverse relationship between reporting and actual occurrences.
An important goal of this research is to develop a multivariate model that identifies factors related to the dependent measures of the potential impact of a news article about crime in retail settings. We selected story length and story location score as our dependent variables for OLS regression modeling.
In implementing the multivariate analysis we examined a large number of independent variables. The bulk of the independent variables that will be related to the impact measures described above were constructed as dichotomous variables based on the categories that appeared on the coding sheets. Variables assessing the “event” (classification of the reported crime, whether a description of the event was relayed, the level of violence described, and whether similar events were noted); the “offenders” (whether there were multiple offenders, whether a description was included, what personal elements of description were used, whether a picture was included and whether the offender’s prior crimes were noted); the “victims” (whether there was a victim or if it was a property crime, whether there were multiple victims, whether a description was included, what personal elements of description were used, and whether a picture was included); the “location of the event” (the city and state of the crime, whether this is local to the newspaper, whether the business name was given, if an individual store or shopping area/mall was involved, and whether the crime scene was inside or outside the business); the “quotes” that appeared (what types of people were quoted – police, businesspeople, experts, victims, bystanders, others, whether statements were made confirming that crime is a problem, confirming that fear is present/appropriate, criticizing popular conceptions of crime, criticizing official statistics and trends, conveying alarmist reactions and whether statistics were cited); and, how many times certain “key words” appeared in the text (armed, dangerous, violent, fear, threat, random, bloody, gang, inner-city, crime, evil, scared, business leaders, community leaders, happen to anyone, gun, safe/safety, secure/security).
In addition, two variables were constructed based on the secondary data collected for each city. The cities represented by the newspapers included in the research were divided into size categories based on their populations. Cities with populations of 90,000 and below were classified as “small” cities, those with populations between 90,001 and 299,999 were classified as “medium” cities, and those above 300,000 were classified as “large” cities. The cities represented by the newspapers included in the research were also divided based on their crime rates. Utilizing FBI data on crimes known to police for the nation as a whole, for the South, and for the particular cities, the cities were classified as having a crime rate that was “low to average,” “above average,” or “high” in relation to the rest of the region. [This regional comparison is especially important given that the South has higher crime rates than other regions of the country.]
The first multivariate analysis examines the impact of the various independent variables on the dependent variable that measures the length of the articles on crime in retail settings. The results of this analysis are presented in Table 3.
The results of the regression analysis are easy to interpret. The “parameter estimate” corresponds to the number of word increase (or decrease) in the dependent variable for a one-unit increase in the independent variable. Many of the variables are dichotomous variables. For example, the variables that indicate whether a picture of the victim was included in the article, whether quotes from the police were included, whether statements were included which confirm crime is a problem, and whether statistics were cited are all coded as either “0” or “1.” A zero indicates, for example, that there are no quotes from the police while a “1” signifies that a statement by a law enforcement officer is cited in the article. Thus, in interpreting the model, we can say that including a statement by a police officer increases the number of words in the article, on average, by 72.6 words.
Other variables, such as keywords, are coded as continuous variables – they are given a numeric value that corresponds to the number of times they appear in the article. The keyword variables “violent,” “scared,” “business leader,” “gun,” and “secure/security” are examples for these types of variables. If the word “gun” is used once in the article it has a coded value of “1,” if “gun” is used twice in the article it has a coded value of “2.” The interpretation of the parameter estimates for these variables is similarly straightforward. Each time, for example, the term “gun” is used in an article, the number of words increases, on average, by 31. The adjusted R-square value shows that these variables, taken together, account for 33% of the variation in story length for specific articles about crime in retail settings.
The results of the multivariate OLS regression analysis for the dependent variable “story location score” are presented in Table 4.
Again, the “parameter estimate” corresponds to the amount of increase in the story location score for every one-unit increase in the independent variable. Therefore, we can say that including a picture of the offender increases the story location story by nearly two points. Similarly, including statements that indicates that fear of crime is justified or appropriate increases the story location score by three points; including quotes from bystanders increases the score by nearly two and a half points; and, including quotes from people other than police, experts, businesspeople, victims, or bystanders increases the score by three points. Referencing a criminal event that includes a murder increases the story location score by over two points. Again, we see that the variables most predictive of prominent placement within the newspaper are those that relate to unusual, violent crimes and fear. When taken together, these five independent variables explain 20% of the variance in story location scores.
Media Representations of Crime in Retail Settings
Tracy L. Johns, University of Florida
Michael J. Scicchitano, University of Florida
Read Hayes, University of Florida
While crime rates have dropped or held steady for decades, extensive media coverage of crime persists. Disproportionate media coverage of violent crimes may lead readers to conclude that relatively rare incidents of crime occur frequently. Given that perceptions of crime in retail settings affect shopping behavior and have a substantial impact on businesses and local economies, research focusing on how the media represent crimes in these settings is important. This research examines newspaper articles about crime in retail settings. The data for the study were collected from an extensive content analysis of newspaper articles in the Southeast, and both descriptive and multivariate analyses were performed. The results indicate that violent crimes in retail settings, such as murder and kidnappings, receive more attention in the media than less violent, more typical crimes such as burglary or larceny. Moreover, violent crimes such as murder or kidnapping receive more attention in the media relative to their actual Index crime rate.
Section
Despite media reports that often focus on the “epidemic” of crime in America, overall rates of crime are declining. The Crime Index is an official measure used by the FBI to gauge fluctuations in the volume and rate of crime reported to law enforcement in the U.S. It is composed of selected offenses that include the violent crimes of murder and non-negligent manslaughter, forcible rape, robbery and aggravated assault, and the property crimes of burglary, larceny-theft, motor vehicle theft, and arson (FBI 2000). In 1999 the Crime Index total was estimated at 11.6 million offenses, the lowest total since 1978. This number represented an 8% decrease over the preceding year, and marked the eighth consecutive decline in the Crime Index. In fact, this number was 16% lower than in 1995 and 20% lower than in 1990 (FBI 2000).
The highest volume of offenses reported to police is, by far, larceny-thefts and the lowest is murder. More specifically, statistics show that most thefts are from motor vehicles. Shoplifting, however, constitutes about 14% of thefts (Pastore and Maguire 2000). Aggregated by location, 34% of burglaries known to police in 1999 occurred at non-residences (stores, offices, etc.), mostly at night. The vast majority of robberies reported in 1999, however, occurred on a street or highway (48%). The remaining robberies occurred at commercial houses (14%), gas or service stations (2%), convenience stores (6%), residences (12%), banks (2%), and miscellaneous other locations (16%) (Pastore and Maguire 2000). In addition, the Bureau of Labor Statistics notes that 80% of the 566 workplace homicides reported in 1998 occurred during the commission of a robbery or other crime (Pastore and Maguire 2000).
The content of crime news diverges widely from the patterns shown in the official statistics. Most studies show that crime in the media bears little resemblance to actual known crime (Heath and Gibert 1996; Katz 1987; Stacco 1995; Tunnell 1992). More importantly, the differences between official statistics and media reports of crime do not appear to be random (Chermak 1994; Katz 1987; Sorenson, et al. 1998). Katz (1987: 57) notes that, “in many respects, the picture one obtains about crime from reading the newspapers inverts the picture about crime one gets from reading police statistics.” The news media consistently feature particular violent crime rather than property crime (Best 1999; Boulahanis and Heltsley 2004; Heath and Gilbert 1996; Katz 1987; Tunnel 1992). “Street crime” (like the Index offenses noted above) is disproportionately featured over “white collar” crime despite the fact that fraud is the most common crime in America (Welch et al. 1997).
While the statistics show that actual crime has dropped or remained relatively consistent in recent years, coverage of crime stories on television news has gone up dramatically. During the 1990’s crime became the number one story covered by the network nightly news, and, in local markets, crime stories consume one-quarter to one-third of total news time (Doi 1998). From 1990 to 1995, for example, homicide arrests dropped by 13% while coverage of homicide on NBC, ABC, and CBS nightly news increased by 336% (Doi 1998).