The Gang Problem

Constructionist and Realist Perspectives

Any review of literature on the gang problem over time will encounter claims of one form or another that surges in gang activity have many of the properties of a panic, stressing discrepancies between media and police responses to the problem and any evidence of a real surge in the problem. For example, McCorkle and Miethe use a constructionist rubric in their book, Panic: The Social Construction of the Gang Problem (2002) where they emphasize the fabrication of many aspects of the problem. Similarly, Shelden, Tracy, and Brown (2001, p. 3) state that data on gang-related articles in newspapers and magazines “demonstrates that media reporting of events does not always conform to reality.” Of course, it is far from clear how that “reality” would be measured since data used to highlight various discrepancies can and have been challenged as valid measures as well. The gang problem could be interpreted as declining in one source of subjective imagery while remaining constant, or even increasing, using data pertinent to an alternative subjective reality.

One of the most common types of evidence used to highlight the constructionist position is any indication of a tenuous or fabricated “reality” as the impetus for defining a social problem such as gang violence. In fact, the most impressive “deconstruction” of a problem from such a perspective is a demonstration that the alleged realty is a fabrication or a dramatic distortion of some real world events. Socially constructed myths can be exposed when information on the actual events that prompt the moral enterprise defining a problem can be shown to be an overdramatization or an overreaction to those events. For example, in the 1980s, news about cocaine use among athletes was depicted by social constructionists as a problem fabricated through media hype with little or no relation to actual drug use (Reinarman & Levine, 1989). Similar arguments have been proposed for problems ranging from the search for Satanists and witches in preschools (Richardson, Best, & Bromley, 1991) to child abuse data (Best, 1989) to the “crack-baby panic” (Logan, 1999). Sudden declines can be depicted as fabrications serving the interests of police as well.

Constructionists’ positions can range from interpretations where the problem is a virtual fabrication to milder forms where variations in some “measurable” form of “real world” phenomena are allowed to enter into the social construction of reality. ¹ For example, in Communication Campaigns about Drugs (1989), Pamela Shoemaker argues that surges in news about drugs were not a product of actual changes in drug use. Rather, such surges were an example of a media campaign. She includes a chapter by Lloyd Johnson who provides quantitative evidence that surges were real, but Shoemaker deems Johnston’s view to be an “unusual position” because Johnston does not view the depiction of the drug problem as merely a media campaign, a panic, or a fabrication. He proposes that real trends in drug use can be measured and that surges depicted as fabricated panics are surges in actual use.

A milder form of constructionism is Goode’s and Ben-Yehuda’s (2009) “contextual constructionism” as a compromise between objectivist and constructionist theories. Contextual constructionists hold that the concern over a problem is subjective with the validity of the concern over a problem to be assessed and evaluated by experts on the basis of some form of objective or scientific evidence. This milder form of constructionism encourages attention to scientific or “objective” data as well as moral and media panics.

Although McCorkle and Miethe (1998, p. 58) emphasize the fabrication of aspects of the gang problem, they also note that there was, in fact, an increase in gang violence during the span of time they studied in Nevada. They argue that the contribution of gangs to violence and drug sales was greatly exaggerated, but “Even so, the figures presented here suggest that gang-related crime increased during the period; thus, perhaps, the response by authorities was proportionate to the objective threat” (1998, p. 58). In short, while they emphasize the fabricated construction of gang panics, their data on gang violence suggest a correlation between gang events and police attention to the problem. Indeed, in the quotation above they acknowledge that “perhaps the response by authorities was proportionate to the objective threat.”

In contrast to the stress on media or police fabricated panics, most quantitative criminologists operate on the assumption that patterns of real criminal events, including gang violence, can be discerned and measured. The central research question from a “realist” perspective is whether valid and reliable measures of the gang problem can be found. For example, a recent example of this realist perspective is Decker and Pyrooz’s (2010) analysis of different measures of gang homicide as reported in the journal, Homicide Studies. Analyzing different sources of data, they conclude that there are strong correlations among existing measures, tapping an underlying reality of gang violence. They assume there is an objective phenomenon and that patterns over time are real. Similarly, Katz and Webb (2006, p. 6) argue that “Although many of these images and perceptions were the product of media generalization and sensationalism, most researchers agree that gang behavior had in fact changed over the past two decades, particularly with regard to violence.”

While Goode and Ben-Yehuda (2009) acknowledge that there may be a scientifically assessable foundation for surges in a problem, they accord primacy to the subjective approach. Most quantitative criminologists are likely to take a position that can be deemed “neorealist” in that they start with the assumption that there are real events and measurable temporal variations but acknowledge that there may be discrepancies to be explained. For example, Huff (1990) proposes that the relationship between a measureable gang problem and police response can take the form of denial, overreaction, or misidentification of the problem, a position that assumes a degree of independence between a “real” problem and response to it. Decker and Kempf-Leonard (1991) have analyzed variations among police, task force, and detained youth as sources of information on gangs, assuming that there is an underlying reality. Moreover, it should be noted that Shelden uses police statistics to defend the claim that media news on gangs does not always coincide with “reality” and cites police data as evidence of major discrepancies. Of course, the implicit assumption in such a statement is that the police data tap an underlying reality that can be used to highlight media fabrications. Yet, other constructionists emphasize the role of police in generating fabricated panics that do not coincide with that reality.

A puzzling characteristic of these different points of view is the lack of actual detailed research relating measures of media coverage over time to the magnitude of the gang problems, especially at a national scale. For example, Shelden et al. (2001) present data on news coverage of youth gang violence and state that surveys of law enforcement show steady growth in estimates of number of gangs and gang members. The National Youth Gang Surveys (NYGS) use data compiled from law enforcement agencies and do, as they argue, generate discrepancies. However, there has been little empirical attention to other possible measures nor to explaining disparities.

Although there has been little quantitative research on media correlations with measures of gang membership or gang violence by constructionists, those proposing a real, objective, and measureable gang problem have not related their measures to measures of media coverage of the problem. Moreover, the measure of news and magazine coverage of gang violence used to claim a gang panic have not been explored in relation to quantitative measures of the gang problem.

This article initiates tests of certain features of the constructionist and neorealist arguments. The first issue to be examined is the relation between media reports of the gang problem over time, rates of youth gang homicide, and national data on rates of gang membership.

The Current Investigation

This article will focus on two measures of media news in relation to two measures used in discussions of the magnitude of the gang problem. There are limitations to each of these measures, but rather than casually dismissing them as sources of data for quantitative analysis, this study explores shared and disparate patterns over time and proposes possible explanations.

LexisNexis data on published printed media and Vanderbilt News Archive data on topics discussed in television national news are used to estimate gang news on a national level. LexisNexis data are appropriate for addressing media coverage in this study because they are the data used by Shelden and others to emphasize media dramatization. The LexisNexis database has been used increasingly to measure variables for quantitative research. Examples of its use include Lau and Pomper (2004) in political science, Weaver and Bimber (2008) in journalism and communications and in the study of binge drinking (Weshsler & Nelson, 2001). Moreover, the LexisNexis database is widely used in marketing research.

Using the LexisNexis search engine, Shelden et al. (2001, p. 3) coded the annual number of major world publications (limited to the United States) and magazine reports containing the words “youth” and versions of violence appearing within five words from the word “gang.” They reported media coverage for the years 1983 to 1999, and Shelden (2006, p. 108) extended the coverage through 2002. The spans used in prior research will be examined first since the relationships may diverge over time. To be fair to constructionist claims, the initial analysis should encompass the span that they were discussing. After that assessment, this study will extend the span of time analyzed, examining the years 1980 through 2010. ²

The Vanderbilt News Archive web site lists a large number of studies using its database, but it has not been used to study trends over time. Its use in the current study is the first of its kind. However, any discovery of shared patterns across alternative measures of specific types of news coverage should strengthen the findings. These data are archived at Vanderbilt University and the words used can be searched and counted online. Different searches were undertaken, including youth gangs, gang violence, and the single term “gangs.” There were too few instances for the combinations of terms to be used but they were reasonably correlated with the data for the general term, “gangs.” ³

One measure of the magnitude of gang violence over time is the delineation of gang homicides in the Federal Bureau of Investigation’s Supplementary Homicide Reports (SHR). Decker and Pyrooz (2010, p. 371) argue that for the period 2002 to 2006 “Both the SHR and NGC measures of gang homicide appear to be robust measures, displaying sufficient levels of internal and external validity for use in cross-sectional and time series analysis.” Based on their findings, they argue that “Those who argue that gang measures reported by the police are political, haphazard, or unreliable, now face the burden of documenting that such claims are true.” The current analysis covers a 30-year span of time and adjusts for variations in the size of the youthful population using rates relative to the population of youths 15 through 24. ⁴

Another measure of the gang problem is the NYGS data estimating the number of gangs and gang members. These estimates have been compiled for a considerable span of time, but jurisdictional reports specifically on gang homicide data are limited to 2002 through 2006. However, estimates of number of gangs and gang members are available for over 30 years and are used to create a measure of estimated gang members per 100,000 youth.

The NYGS data have been criticized for several reasons with the foremost criticisms being the ambiguities and inconsistencies involved in defining the meaning of “gang” among jurisdictions and over time. Yet, police estimates of the number of gangs and gang members have been used to illustrate the independence of media coverage from the reality of the gang problem (Shelden, 2006, p. 108). However, these illustrations have not been accompanied by examination of the relation between media gang coverage and the NYGS gang data over long spans of time. This article attempts such an analysis, recognizing that there may be shared patterns as well as disparities. ⁵

Analysis

The first analysis is limited to the years reported on in Shelden’s update (2006) of data from 1983 through 2002. Claims stressing discrepancies were made based on that span of years and should apply to that specific span of time. Shelden (2006, p. 108) updated the measure of media attention through 2002 and stated that “Although media coverage has declined recently, the number of gangs, gang members, and gang-related crimes has actually increased . . .” Not only is it unclear what is “recent,” but no actual data on the magnitude of the gang problem over time in relation to the measure of youth gang news was presented. That task can be carried out here, using the LexisNexis news data as reported by Shelden together with FBI data on youth gang killings and the NYGS gang data.

Examining the plot for the three measures yields plausible arguments about meaningful temporal patterns. First, there is a very strong relationship between the FBI youth gang homicide rate and LexisNexis gang news (r = + .85). Second, although the NYGS measure is significantly related to LexisNexis gang news (r = + .70), it is not significantly correlated with youth gang homicide rates over time and lags behind these other measures. Third, the NYGS measure is relatively flat after the peak in the mid-1990s as compared to the marked decline in youth gang homicides and gang news. Fourth, NYGS data, FBI youth gang homicide, and Shelden’s gang news all peak between 1993 and 1996. While the specific year may vary, it is safe to state that all ways of assessing the gang problem analyzed in this analysis support a view of the peak in the mid-1990s as “real.” Moreover, all measures begin to decline until approximately the year 2000 with some indication of an upturn in the last year studied.

Figure 1 suggests that national trends in the LexisNexis gang news follow national trends in FBI youth gang homicide rates and that the National Youth Gang police measure of rates of gang membership follows the LexisNexis gang news. In fact, the correlation between the youth gang homicide rate and the LexisNexis news measure is +.94 when the news is lagged one year. ⁷ Thus it can be argued that the trends in Figure 1 support a view where the news follows youth gang homicide until the peak in the mid-1990s with the youth gang murder rate declining before the national gang news began to decline.

OPEN IN VIEWER

Figure 1.

LexisNexis gang news, youth gang homicide, and National Youth Gang Survey data (standardized measures).^a

Examining the National Youth Gang data on estimated rates of gang membership over time, it can be proposed that the organizational investment required to generate new measures of the gang problem would lag behind homicidal events and media attention to the problem. Rather than accord law enforcement an active role in generating panic about the gang problem, it can be proposed that such investment follows the news rather than generates it. In addition, once the investments have been made they will not decline in lockstep with other measures of the problem. The NYGS developed into an institutionalized mechanism for reporting on the gang problem and organizational inertia alone may sustain its temporal patterns. Moreover, while the relationship between gang news and youth gang homicide is not perfect, the media news appears to be either a response to gang killings or a strong covariate.

Additional Analysis of Gang News, 1980-2010

The analysis reported above was limited to the span of time discussed as relevant to the concept of a media or police induced panic. A longer time span was used to encompass the full set of years between 1980 and 2010. As was the case for the span studied by Shelden, the strongest relationship is between the LexisNexis gang news measure and the FBI youth gang murder rate (r = + .89). The correlation for the Vanderbilt News Archive measure and the FBI youth gang murder rate is significant as well but weaker than that found for LexisNexis news, yielding a correlation of +.60. The weakest relationships involve the NYGS data from law enforcement agencies. The NYGS measure was unrelated to the Vanderbilt Archive measure (r = + .06) and weakly related to the LexisNexis measure (r = + .33). In sum, both measures of media news are related to the FBI homicide data with an especially striking relationship between the LexisNexis measure and FBI youth gang homicide rate. The NYGS police data tend to follow rather than stimulate gang news.

Shared and Disparate Patterns

While patterns can be discerned by “eyeballing” plots of the measures, a more complex approach is to discern the best-fitting least squares equations. In Figure 2 each measure has been regressed on Year 1 through year 8. For example, the best-fitting equation for FBI youth gang homicide rates included significant relationships for Years 1 through 7 and yielded an R² of .87. The most striking feature of the analysis is the overlap between the FBI equation and the LexisNexis news measure (with Years 1 through 7 significant and an R² of .87). The television news archive data follow a similar pattern (Years 1 through 7 significant with an R² of .87). The archive equation exhibits a more distinct downward movement through the year 2000 and a more distinct upward trend through 2007. ⁶

OPEN IN VIEWER

Figure 2.

Plot of best-fitting equations for the four measures.^a

The National Youth Gang data on gang members yields a similar equation to the other measures (Years 1 through 7 were significant with R² equal to .96), but follows a far different trajectory. A common argument by constructionists has been that the gang problem may be exaggerated by police to increase resources and expand authority (Ben-Yehuda, 1986; Glaser, 1978; Jackson & Rudman, 1993; McCorkle & Miethe, 1998; Zatz, 1987). For example, McCorkle and Miethe (1998, p. 34) argue that

easy access to legislative and executive leaders, the potential for creating social disorder, the monopoly of information about crime grants enforcement agencies a great deal of political power, which they have wielded successfully to protect and advance their own interests.

If that argument were correct, the data based on law enforcement agency reports should precede other measures.

Police departments are public institutions that do not have organizational incentives to grow. It can be argued that police resist innovation and expansion of their work load although they do respond to real events, publicity, and political pressures (e.g., Silverman, 2001). The creation of the NYGS increased the amount of paper work as did the proliferation of gang units charged with identifying and responding to gang-related events. In fact, based on an ethnographic study of police gang units, Katz and Webb (2003, p. 485) conclude that “Unlike previous research examining gang data, the findings of this study did not reveal that the production of gang statistics was influenced, or manipulated, for the benefit of the police organization.”

While there are instances where police propose and lobby for legislation, the autonomous power of the police over the distribution of resources seems to be more anecdotal than systematic. Police, sometimes reluctantly, respond to the environment around them, including but not limited to actual criminal events. Indeed, discussions of police resistance to change are quite common, and it can be argued that as strong a case for inertia can be made as for the view that police actively pursue resources for new tasks (see Skogan, Why Reforms Fail, 2008). There is more evidence that increases in gang news are followed by increases in police attention to the gang problem than the opposite. This pattern is consistent with Huff’s delineation of “denial” as a common response to the gang problem with the qualification that such denial will be less common if the underlying problem escalates.

Patterns for Major and Nonmajor Newspapers

As noted in the introduction to this study, little or no attention has been paid to ways of measuring media attention to the gang problem. The LexisNexis measure was based on major publications and magazines restricted to the United States with LexisNexis defining major publications. ⁷ There has been some discussion of differentiating between local and national newspapers in the study of collective behavior and social movements (see Earl, Martin, McCarthy, & Soule, 2004), but there has been no comparable attention to diverse forms of print media in relation to the gang problem.

In Figure 3, FBI youth gang homicides are plotted together with the measure of LexisNexis news in major newspapers with a separate measure for all other newspapers. ⁸ There are two notable differences in the plots for major and nonmajor newspapers: The gang news in the nonmajor newspapers lag behind the major newspapers, and the nonmajor newspapers exhibit a major surge in the 2000s. There is a surge for major newspaper gang news in the 2000s as well although it was not as prominent as that found for the nonmajor newspapers. In contrast, youth gang homicide rates were declining from 2001 through 2006 during the surge in gang news.

OPEN IN VIEWER

Figure 3.

FBI youth gang homicide, LexisNexis major and nonmajor newspapers.^a

There are several possible explanations for these patterns. The lag and the surge of gang news for lesser newspapers may reflect the widely noted expansion of gangs to smaller cities, suburbs, and rural areas. Gang news in these areas would be more local as well as lag behind news in major newspapers. Another possibility is the proliferation of new types of gangs in the areas covered by nonmajor newspapers. Celinda Franco (2008, pp. 2-3) proposes a major contrast between street gangs and more recent Latino gangs:

Although street gangs are largely an urban phenomenon, MS-13 and M-18 cliques have emerged among disaffected Latino youth in suburban and rural areas across the country. According to a national gang survey, the MS-13 and M-18 gangs have an established presence in Washington DC; Northern Virginia; certain cities in Maryland; Nashville, Tennessee; New York, New York; Houston, Texas; and other rural and urban areas.

A search for La Mara Salvatrucha (MS-13) using LexisNexis shows that this type of gang was first referred to in LexisNexis in 1991 with a total of seven references in the entire decade of the 1990s. By 2003 there were 24 references with a peak at 93 references in 2004. There were 11 references in 2009 and only two in 2010. Such gangs would not generate the same patterns of violence as the FBI supplementary data on youth gang killings.

A constructionist might point to the disparity between gang news and FBI gang killings in the 2000s as support for the notion of a media induced panic. However, the two media measures of the prevalence of gangs or gang violence do not play a role in generating that specific surge. The basic argument has focused on the role of law enforcement and police in generating media, political, and public panic. Rather than deem the surge in the 2000s as a panic, plausible arguments about changes in the nature of gangs and the location of different types of newspapers seem most plausible.

Conclusions

This research initiates a systematic analysis of the relation between media news coverage of youth gangs, measures of gang membership, and youth gang homicide over time and at a national level. Since one of the measures of the gang problem (NYGS) is compiled by law enforcement personnel, the possible role of police in the generation of panics and/or fabricated trends was explored as well. The research challenges claims made by eyeballing quantitative data coupled with general, broad statements about gang membership and activities. Contrary to emphasis by some constructionists on discrepancies between media gang news and measures of the gang problem, the data show gang killings to be tightly related to media coverage of the problem. ⁹ Claims that there are discrepancies between news and the actual gang problem are correct in the sense that the relationships are not perfect, but the LexisNexis measure used by Shelden is very strongly related to FBI youth gang homicide rates. Moreover, the trends in media coverage appear to follow trends in youth gang homicide rates. In contrast, the NYGS data do not follow as tight of a shared path as found for media coverage and FBI data on youth gang homicide rates. A stronger case can be made for real events and/or media attention preceding the NYGS data than for the NYGS data generating a media panic.

The common argument that police or law enforcement agencies are motivated to pursue resources by exaggerating crime problems, including the gang problem, can be countered by the argument that police attention tends to lag behind other measures of the problem and to sustain attention after real events begin to decline. Thus the widely stated emphasis on police initiated expansion can be countered with emphases on police inertia and momentum, generating lagged responses. There are discrepancies among temporal patterns that need to be addressed, but rather than point at discrepancies as evidence of “panics” that serve the interests of law enforcement, criminologists should be developing and testing hypotheses about the possible foundation of such patterns.

This analysis has implications for basic perspectives on social problems, including the gang problem. The current analysis suggested a perspective that, for want of a better term, will be called “neorealism.” The data are contrary to constructionist claims that the patterns involving measures of gang activities and gang membership observed are artificial. Constructionists look for the group interests behind media coverage and law enforcement promotion of statistics that do not conform to the objective harm of the problem (Goode & Ben-Yehuda, 2009, pp. 159-160). Constructionists often initiate their depiction of social problems with the assumption that they are largely artificial (e.g., surges in the gang problem are artificial panics). To some constructionists, “realism” (i.e., the notion that there are real problems that can be measured) is posited as serving the interests of various elites, including law enforcement and the media.

Both contextual constructionism and neorealism start with the assumption that there are real patterns but they must be discerned empirically and discrepancies must be addressed. When discrepant depictions of the reality of a problem are found, contextual constructionists exclusively focus on the possible interests involved in creating these discrepant depictions. However, there is an unwarranted assumption that discrepancies between patterns of reality and depictions of reality are caused by vested interests. The recognition that the concern over harm can be compared with the reality of harm must also concede that these measures of reality may be imperfect. When explaining discrepancies through underlying interests alone it is unclear whether different depictions of reality are due to different measurement processes of a similar underlying reality or whether depictions of reality are manufactured to serve group interests. The analysis in this article supports a compromise with considerable evidence that media gang news is tightly correlated or follows actual violence while also documenting variations and disparities in need of further examination.