Fear of Gang Crime: A Look at Three Theoretical Models

Abstract

Gang crime and resulting public fear became a major policy focus during the 1990s, yet few studies specifically focus on fear of gang crime. Guided by social disorganization theory, we test three theoretical models about the individual thought processes leading to fear of gang crime. Using structural equation models, we find that each of these three theories—diversity, disorder, and community concern—is an important predictor of gang-related fear. In addition, we find that the indirect relationships between demographic characteristics, theoretical variables, and fear depend upon which model is tested.

Fear of Gangs and Crime Policy

Crime has been a major focus of political campaigns for years (Johnson 1997; Warr 1995, 2000), but during the 1990s, policymakers routinely cited gang violence and the fear it invoked in the public as a primary justification for harsher laws and punishment policies (e.g., Clinton 1997; Senate 1994). Throughout the 1990s, both the U.S. House and the U.S. Senate held hearings about gang violence in an effort to “do something” about gangs, which, they believed, had the nation “caught in the grip of fear” (Senate 1994:2; see also House 1997a, 1997b). In his opening statement to the 1994 Senate hearing The Gang Problem in America, then-Senator Kohl summed up the Senate's concerns:

Too many of our young people are killing and being killed and breeding fear among all the honest people who try to walk our streets. Throughout the United States, gangs have much to do with all of this … today, we recognize that violent gang crime is a national problem.… (Senate 1994:1–2).

As a result of legislator concerns, the 1994 Federal Crime Bill made it a federal offense to be involved in gang-related crime and created minimum penalties for related offenses (Senate 1994:4, 19; Violent Crime Control and Law Enforcement Act of 1994, Public Law 103-322, §150001).

A few years later, President Clinton declared a war on gangs in his 1997 State of the Union address and announced that fighting gangs would be a top priority of his second administration (Clinton 1997; Peterson 1997). Later that year, the U.S. House Subcommittee on Crime held a hearing, Gang-Related Witness Intimidation and Retaliation, which reiterated the concern that gangs were still out of control. As a Los Angeles deputy district attorney told the subcommittee, “Fear and intimidation are the foundation of gang dominance in our communities…” (House 1997b:5). Clearly, during the 1990s, gangs rose to the center of the policy debate about crime, due, at least in part, to policymakers' perceptions that the public was terrified of gangs.

Few studies specifically examine fear of gang crime (see Lane 2002; Lane & Meeker 2000, 2003). Research regarding fear of nongang crimes indicates that the public's fear is complex and not likely to be lessened simply by passing more laws, such as Clinton's Anti-Gang and Youth Violence Initiative, and increasing punishments (see Peterson 1997; House 1997b). This lack of effect is due in part to the public's limited knowledge about the workings of the criminal justice system (Roberts & Stalans 1997). Findings also indicate that actual crime levels do not neatly translate into fear levels. Most fear of crime research indicates that perceptions of community factors such as diversity, disorder, decline, and crime are probably more important in predicting fear than is the objective “reality” of crime and victimization (e.g., Garofalo & Laub 1978; Lewis & Maxfield 1980; Taylor 2001; Warr 1994, 2000).

Fear of crime is functional if it helps people protect themselves from real threats (Warr 2000). But research has shown that people who are most at risk for victimization (e.g., young, minority males) are less fearful than those at less risk, such as women and the elderly (see Warr 1994 for a review). Social disorganization is a key theory that has been used to explain how real and perceived community characteristics can increase fear of crime, especially when victimization risk is low (Taylor & Covington 1993). Characteristics of social disorganization include poverty, residential mobility, racial heterogeneity, and the presence of gangs (see Bursik & Grasmick 1993; Sampson 1993).

Guided by the social disorganization framework, fear-of-crime theorists have developed a number of independent yet interrelated models using community factors other than crime to explain the public's fear. These models include subcultural diversity, disorder/incivilities, and community concern (Covington & Taylor 1991). Although all of these perspectives focus on contextual factors other than crime levels to explain individual fears of crime, each sees the key aspects of the individual thought process differently.

According to the subcultural diversity perspective, people are more likely to be afraid because they are worried about the behaviors of people who look or act different. The disorder model asserts that people see incivilities as an indicator of crime and therefore feel more vulnerable and afraid. The community concern (decline) model sees fear primarily as a result of residents' concern that the community is different, or less safe, than it was in the past (Covington & Taylor 1991; Merry 1981). Previous studies examining these three theoretical perspectives generally indicated that each is supported by empirical findings (e.g., Covington & Taylor 1991; McGarrell, Giacomazzi, & Thurman 1997; Taylor & Hale 1986). However, these studies generally have not measured fear of gang-related crime (Lane & Meeker 2000) and have not addressed fear of multiple crimes as we do here.

Understanding fear of gangs and their crimes specifically is important for reasons beyond those that are directly relevant to policy. First, few studies have examined the importance of perpetrator characteristics in predicting fear of crime, yet certain perpetrators (e.g., gangs or terrorists) may evoke more fear than others because of the images (e.g., hard, uncaring, ethnically/racially different) and crimes (e.g., random, violent) associated with them (Best 1999; Madriz 1997). Second, fear of gangs may be especially relevant to concerns about diversity, disorder, and decline. Theorists have considered gangs to be a key factor pointing to social disorganization because their presence points to the community's inability to watch and control their teenagers (e.g., Sampson 1993; Sampson & Groves 1989). Disorganized communities are also those that are most likely to experience more diversity, disorder, and decline, which are the key factors that research shows predict fear. For some residents, the presence of these other problems in the community may indicate that gangs are a serious threat to their safety, or that they will be soon (Lane 2002).

Using a survey conducted in September 1997 in Orange County, California, an area where local policymakers and the public struggled with gang crime during the 1990s, we examine the importance of these theoretical models in predicting individuals' fear of gang-related crime. Our study builds upon prior literature in three key ways. First, we study fear of gang crime specifically—a scarce topic in the literature, despite its popularity in policymaking circles. Second, we respond to other researchers' continuous calls for more research on fears of specific crimes (e.g., Ferraro 1995; Ferraro & LaGrange 1987, 1988; Warr 1994, 2000) as an improvement over general questions. Here we examine residents' fears of six gang-related crimes, ranging from graffiti to drive-bys. Third, we use structural equation models to examine the impact of each of these three models as intervening constructs between demographic characteristics and fear of gang crime.

Through such analyses, we can determine both direct and indirect relationships among the demographic variables, the theoretical variables, and fear of gangs. Our primary research questions are (1) do the relationships between our demographic and theoretical variables and fear of gang crime parallel the relationships found among these variables in other studies that examine nongang-related crimes? and (2) are there indirect relationships between the demographic characteristics and fear of gangs through the theoretical variables that might not be apparent in other types of analyses?

Social Disorganization Framework

The specific theories we are testing here—subcultural diversity, disorder/incivilities, and community concern (decline)—all have their origins in the social disorganization tradition. The tenets of social disorganization theory are well established in the literature (e.g., Bursik 1988; Bursik & Grasmick 1993; Sampson 1993; Shaw & McKay 1972 [1942]). Social disorganization has been defined as “the inability of local communities to realize the common values of their residents or solve commonly experienced problems,” such as disorder and crime (Bursik 1988:521). Shaw and McKay, early social disorganization theorists, noted that areas characterized by low socioeconomic status faced high residential mobility and racial heterogeneity, thereby decreasing their ability to keep new people out (e.g., immigrants), decreasing social control, and increasing crime (Bursik 1988; Shaw & McKay 1972 [1942]). Sampson has argued that a primary indicator of community social disorganization is how well communities “supervise and control teenage peer-groups—especially gangs” (1993:267; emphasis in original) and that this ability is related to crime levels (see also Sampson & Groves 1989). Taylor and Covington (1993) found that social disorganization and groups of teens were also an important predictor of fear of crime. In particular, they argued that incivilities—neighborhood characteristics that symbolize disorganization and disorder—invoke fear because people think they are linked to problems with local teens, including crime. Given the link between social disorganization, gangs, crime, and fear, we use this theory as the backdrop to understand the theoretical models about fear as they specifically apply to fear of gangs in Orange County.

Prior Research on Theoretical Factors and Fear

Three studies have compared the differential explanatory power of more than one fear of crime model, as we do here. Using path analyses, Taylor and Hale (1986) examined the effects of both disorder/incivilities and community concern and found that each of these models independently explained about 10% of the variance in fear of crime. McGarrell, Giacomazzi, and Thurman (1997) put these two perspectives with other variables into one statistical model and found that together their variables explained much more variance in fear (43%). Covington and Taylor (1991) independently tested fear of crime models, including subcultural diversity, disorder, and community concern, and found that each was predictive of fear. In a qualitative study of fear of gangs in Santa Ana, an Orange County city, we also have found support for all three theoretical models (Lane 2002). Our results indicated that long-term residents believe that Latino immigrants, especially undocumented ones, have brought disorder and decline, which in turn have brought gangs, thereby leading residents to be more afraid. Most studies have tested the components of one model at a time. The following sections briefly describe the essence of each “fear of crime” model and discuss related research findings.

Subcultural Diversity

A few studies have focused on racial and ethnic diversity as a key factor predicting fear of crime. The subcultural diversity model posits that in communities without strong friendship networks, people will be more afraid of strangers who are racially, ethnically, and/or culturally different than they are (Merry 1981; see also Bursik & Grasmick 1993; Covington & Taylor 1991; Skogan 1995). According to Merry (1981), this fear is prompted primarily by residents' inability to interpret the mannerisms and behaviors of people who look and act differently than they do. Residents often believe that the “others” do not share their own values and commitment to the community (Merry 1981; see also Lewis & Salem 1986). In her study of women, Madriz (1997) also found that stereotypes and perceptions of racial and cultural differences, including immigrant status, led her respondents to be more afraid. She argued that predominant ideologies in society lead citizens to see “[b]lack and Latino young males and females and members of ‘new’ immigrant groups [as] a dangerous class in the United States” (Madriz 1997:156). Her findings might be especially relevant to Southern California, where gangs have traditionally been associated with minorities—especially Latinos and immigrants (Jankowski 1991; Klein 1995; Moore 1978, 1991; Moore, Vigil, & Garcia 1983; Vigil 1988).

Some quantitative studies have supported the qualitative findings of Merry (1981) and Madriz (1997). One such study found that city racial composition affects fear of crime for both whites and minorities (Liska, Lawrence, & Sanchirico 1982). Covington and Taylor (1991) later found both that people who live in predominantly black neighborhoods are more afraid and that people who see themselves as racially different than the majority of their neighborhood are also more fearful. In comparing whites' and blacks' fear of crime, Chiricos, Hogan, & Gertz (1997) found that perceived neighborhood racial composition is a significant predictor for whites but not blacks. Taylor and Covington (1993) found that although disorder is a primary predictor of fear, residents' concerns about disorder are linked to racial composition changes in their neighborhoods (see also Taylor 2001). Another study indicated that whites are more afraid of black strangers than white ones, and this fear is intensified if the whites are prejudiced (St. John & Heald-Moore 1996).

We recently conducted a quantitative study that examines the relationship between diversity concerns and fear (Lane & Meeker 2000). We found that concerns about subcultural diversity—in particular, concern about foreign immigrants, racial and ethnic relations, and changing moral standards—together are important predictors of both general fear of crime and fear of gangs, but that diversity has a stronger impact on fear of gangs. For both types of fear, we found that older and less educated people are more concerned about diversity, leading them to be more afraid. Less-educated and minority respondents are more likely to live in the area with more social disorganization and gangs and therefore are more likely to be afraid of gangs, but not crime more generally. Our results showed that direct relationships between demographic factors and fear also vary across models. Older people are more afraid of crime generally, but respondents who are less educated, renters, or live in the high-crime area are more afraid of gangs, irrespective of other factors in the model. Interestingly, unlike most fear studies, we found that gender is not significant (Lane & Meeker 2000). Our earlier study illustrated that (1) concern about diversity is an important predictor of fear, especially fear of gangs; (2) predictors of different types of fear (e.g., gang v. nongang) can vary; and (3) the perpetrator of the crime is a relevant variable, even though fear of crime research generally has yet to pay much attention to this component of crime. This study improves on our previous Orange County research by using a better data set that was designed specifically to measure fear of gang crimes and by examining more of the other predictive factors that fear research has determined are important.

Disorder/Incivilities

Many recent fear of crime studies have examined the disorder/incivilities model—looking at the effects of problems such as graffiti, trash, abandoned buildings, homeless people, and youths hanging out (e.g., Skogan 1990, 1999; Taylor 1999, 2001; Wilson & Kelling 1982). The general idea behind this model is that people who see incivilities in their communities interpret them as symbols of “deeper, underlying problems” in the area (Taylor 2001:7; Skogan 1990). These symbols make them feel more vulnerable and therefore more afraid of crime (Lewis & Salem 1986; Skogan 1990). Perceived incivilities—or people's perceptions of community disorder problems—have consistently been significant predictors of their crime fears (e.g., Lewis & Maxfield 1980; LaGrange, Ferraro, & Supancic 1992; McGarrell, Giacomazzi, & Thurman 1997; Rountree & Land 1996; Skogan & Maxfield 1981; Taylor & Hale 1986; Taylor 2001). Some researchers, however, have studied the effects of objectively coded signs of disorder and compared fear of crime levels in different neighborhoods or on different blocks (Perkins et al. 1990; Perkins, Meeks, & Taylor 1992; Taylor, Shumaker, & Gottfredson 1985), but results have been inconsistent. Some studies have found that “actual” disorder problems predict fear (e.g., Perkins et al. 1990; Perkins, Meeks, & Taylor 1992; Perkins & Taylor 1996; Taylor 1996), but others have found that they generally do not once researchers control for other variables (e.g., Taylor, Shumaker, & Gottfredson 1985; Taylor 1999, 2001). Covington and Taylor (1991), for example, found that neighborhoods with more actual incivilities do have more fear of crime, but their research indicates that the relationship between perceived incivilities and individual fear has stronger statistical support. Interestingly, Taylor (1996) found that actual disorder is related to more fear but that it also is related to more investment and involvement in the community. He argued that disorder might draw some residents together while at the same time make them feel less safe.

Community Concern

The community concern (decline) model argues that fear of crime is primarily due to concern over community decay; that fear is heightened when local social ties are weak (Conklin 1975; Covington & Taylor 1991; Garofalo & Laub 1978). Researchers have generally seen community concern as another possible connection between disorder and fear. Some researchers have argued that when people see indications of disorder, it prompts them to believe the community is in decline—that it is not the way it “used to be”—which in turn triggers fear of crime (Garofalo & Laub 1978; Covington & Taylor 1991; Taylor & Hale 1986). As Taylor and Hale indicated:

The key causal sequence in the community concern model leads from objective characteristics (e.g., crime, physical conditions, socioeconomic status) to perceived problems which in turn lead to concern, which in turn leads to fear. (1986:164)

Even with these suggested ties between the theoretical constructs, the disorder and community concern models remain theoretically distinct because they are not inherently connected. We found no studies of the community concern model that focus exclusively on affluent neighborhoods or communities without much disorder, but other factors, such as diversity alone or increasing urbanization, might prompt residents to think the community is changing for the worse, thereby increasing fear of crime.

Not as many studies have focused exclusively on community concern (without disorder in the model), but those that do find it an important predictor. Hunter and Baumer (1982) found that people who are more integrated into their communities are less fearful (see also Lewis & Salem 1986). Covington and Taylor (1991) found that people who live in neighborhoods where they believe their neighbors would call the police if they saw graffiti painting in progress generally are less fearful. When they examined community concern as an intervening variable between disorder and fear, Taylor and Hale (1986) found that it was a significant and direct predictor of fear of crime.

Methods

Setting

The setting for this study was Orange County, California, an ideal location for research on fear of gangs. The area has experienced increasing racial and ethnic diversity, a long history of gang-related problems, and a decade-long, hard-hitting approach by local law enforcement to address increases in the number of gangs and gang members and to reduce the amount of gang crime (Baldassare 2000; Baldassare & Katz 1995a; Orange County Chiefs' and Sheriff's Association 1997; Rackauckas 1999).

Orange County is a primarily suburban area about 40 miles south of Los Angeles. In the mid-1990s, approximately 2.7 million people lived in 31 cities and unincorporated areas. The number of county residents increased 25% from 1980 to 1990 and another 13% from 1990 to 1998. Recent numbers from around the time of our survey (1998) indicated that the primary minority groups in the county were Latinos (29%) and Asians/Pacific Islanders (13%), primarily Vietnamese. African Americans constituted only about 2% of the population (Gaquin & DeBrandt 2000). In addition to a large number of documented Latino immigrants, there are many undocumented ones, whom residents associate with gangs (Lane 2002). Resident perceptions are correct in that regard, since most Orange County gangs are Latino, a characteristic that mirrors the remainder of Southern California (Capizzi 1998; Klein 1995; Moore 1978, Rackauckas 1999). In 1994, Californians, including those in Orange County, overwhelmingly expressed their displeasure with what they called “illegals”—and gave clear indications of their concerns about increasing diversity—when they voted to pass Proposition 187 (Ferrell & Lopez 1994). The purpose of this measure, which stalled in the courts and has never been implemented, was to restrict civil rights and social services for all undocumented immigrants (Nieves 1999; Purdum 1998). The county is mostly white and, in California, whites are most likely to consider immigrants, especially Latinos, a burden (Gaquin & DeBrandt 2000; Baldassare 2000).

Orange County also strengthened its efforts to eradicate gang activity as early as 1992, when the local Chiefs' and Sheriff's Association created a Gang Strategy Steering Committee (GSSC) to fight the “coming storm” of gang crime (Meeker & Vila 2002; Vila & Meeker 1997). The committee was, in part, a reaction to high-profile violent gang incidents (such as the gang-related murder of an innocent woman on her way home from the grocery store), to indications that gang activity was no longer simply an inner-city Latino problem, and to academic warnings about the gloomy future (Chow & Do 1991; Klein 1995; Lane 1998). As incidents continued and as the number of gang members more than doubled in about six years, the GSSC set out to warn the “naïve” public through press conferences and public awareness campaigns that they were now more at risk for being victimized by gang crime (Ellingwood 1995; Lane 1998). Media coverage of high-profile gang incidents and the efforts of the GSSC are some of the contextual elements that likely raised fear of gang crime among Orange County residents during this decade by making gangs more salient in the public's mind (Lane 1998; Lane & Meeker 2003). Although no trend studies indicate changes in fear of gangs specifically, one 1994 study indicated that 75% of respondents had heard about gangs or gang-related problems in their communities (Baldassare & Associates 1994). In 1996, California residents, including those in Orange County, expressed their concern about gangs by overwhelmingly voting in favor of propositions that created sentencing enhancements for the gang-related crimes of carjacking and drive-by shooting (Los Angeles Times 1996). In addition, annual polls indicated that crime was considered the most important problem facing the county during the 1990s. Interestingly, the year after the GSSC was implemented (1993), the percentage of residents who believed crime was the most important problem almost doubled from the prior year (Baldassare & Katz 1992–2000; Lane 1998). With this local backdrop, we study fear of gang crime among Orange County residents.

Sampling and Data Collection Procedures

From September 3 to 28, 1997, we conducted a random digit dial (RDD) survey of 1,000 Orange County residents. We ensured an equal split of men and women in the sample. The respondents had the option of taking the survey in English, Spanish, or Vietnamese. We translated the surveys into these languages and had bilingual interviewers. The survey was specifically designed to represent Orange County's population, but had a +/− 3% margin of error. Like the county, the sample was 50% female. The RDD sample was 63% white, 18% Latino, 6% Asian American, and about 11% others, which generally represented the racial/ethnic composition of the county.¹

For the current analyses, we used a subsample of the RDD group—specifically, we included only those respondents who had complete data on all variables of interest (n=694). This approach ensured that each of our path models represented the same subjects and were directly comparable. And t-tests indicated that the respondents in this subsample were not significantly different from people not included in this trimmed sample on any analyzed variables except age and education. Our trimmed sample was slightly younger (mean=4.04, mean for others=4.39) and slightly more educated (mean=5.23, mean for others=4.83) than other survey respondents not included in this subsample (see Table 1 for demographic characteristics of the sample, population comparisons, and variable codes).

Table 1. Variables and Sample Characteristics (n=694)

¹¹ Latino is not a racial category in the census. The census data indicate that 28.5% of the county is Latino and 71.5% is not.

¹² The census includes Vietnamese under Asian/Pacific Islander. We separated them from others because we were interested in studying fear of gangs among Vietnamese specifically. First, Orange County is home to many Vietnamese. Second, police are concerned about the rising number of Vietnamese gangs, especially in Westminster (including Little Saigon) and Garden Grove, and these gangs are known to primarily victimize their own community. Unfortunately, our random digit dial sample yielded only four people who called themselves Vietnamese (one was excluded here due to missing data). We do not know if our RDD sample just did not yield many Vietnamese or if they represented a disproportionate amount of refusals. According to local police, local Vietnamese are suspicious of justice system personnel and of people asking questions. This may have led to their being underrepresented in the sample.

Our trimmed sample also represented the county, except that Latinos and Vietnamese were underrepresented and that more of our respondents had college degrees.² Because we did not have any respondents under 18 years old, the percentages in our age categories were different from those of the county. We had a disproportionate number of respondents between ages 25 and 44, which may have affected our results (see Table 1 for population comparisons).

Measurement and Analysis Approach

The major focus of this study was to test the impact of perceived diversity, disorder, and community decline on fear of gang crime, which we measured with multiple indicators to improve the power of the analysis. However, we also wanted to control for key single-indicator variables (demographic characteristics) that have been shown to be important predictors of fear of crime. This created an analytical problem for latent construct structural equation models in that some of our measures had multiple indicators and some had single indicators. To manage this inconsistency, we used a two-step process to conduct our analyses. The first step was to construct the measurement models using confirmatory factor analysis, and the second step was to test the theoretical models with path analysis.

The Measurement Models

In the first step, we conducted latent construct confirmatory factor analysis (CFA) models to see if our indicators clearly represented our theoretical and fear constructs. We used the results and suggestions of prior studies to develop the survey questions included here. We constructed our dependent variable, fear of gang crime, from measures designed to respond to the concerns of some prominent fear researchers about the adequacy of the standard General Social Survey (GSS) and National Crime Victimization Survey (NCVS) questions in measuring fear (see Ferraro 1995; Warr 2000).³ These measures have been criticized as having limited validity and reliability because they are too general, do not mention the word crime, and do not gauge the complexities of people's fears about crime—for example, the likelihood that people fear different crimes differently (Ferraro 1995; Ferraro & LaGrange 1988; LaGrange & Ferraro 1987, 1989; Warr 1994, 2000). To prevent these measurement issues in our study, the survey asked respondents to indicate on a four-point scale (not afraid–very afraid) “how personally afraid” they were of being victimized by six specific gang crimes: graffiti, home-invasion robbery, drive-by or random gang-related shooting, physical assault by a gang member, harassment by gang members, and carjacking.⁴ We rotated the order of these crimes in the questionnaire to control for response bias. To ensure that we controlled for the likelihood that our respondents feared some crimes more than others, we used the factor weight results produced by our fear measurement model (Figure 1) in this first step for the measures used in the Step 2 analysis described later.

Figure 1. CFA for Gang Fear

We had three theoretical constructs: diversity, disorder, and community concern (decline). Our measures of diversity and disorder were created based upon responses to the following broad question:

We have a number of questions about your community as you define it. I will read you a list of some things that currently might be problems in your community. After I read each one, please tell me whether you think it is a big problem, somewhat of a problem, a small problem, or no problem in your community.

Our 12 subquestions were based on prior research, guided by social disorganization theory and designed specifically to measure concern about diversity and disorder (see Ferraro 1995; Skogan 1999; Skogan et al. 2000; Taylor 2001). The questionnaire rotated all 12 items. The scores ranged from 1, being “no problem,” to 4, being “a big problem.”

The items we used to measure diversity were people moving in and out without becoming attached to the community (“no attach”), language differences between residents (“lang diff”), cultural differences between residents (“cult diff”), and racial differences between residents (“race diff”).⁵ The items we designed to measure concern about disorder were poverty and economic hardship (“poverty”), people or landlords allowing their property to become run down (“rundown”), abandoned houses or other empty buildings (“abandon”), graffiti, too many people living in one residence (“mpeople”), gunfire, youths hanging out (“yhangout”), and gangs. All of these problems have been used as indicators of social disorganization or disorder in other studies except for “too many people living in one residence” (e.g., Ferraro 1995; Skogan 1999; Skogan & Hartnett 1997; Taylor 2001). We included this particular question in our study because of our knowledge of the Orange County context. Qualitative research in the county has shown that one problem concerning local residents is the presence of multiple families living in one household (see Lane 1998, 2002).

To measure community concern—or worry about community decline—we used responses to three questions that measured respondents' perception of negative community change during the past 2–3 years. The questions again were informed by previous studies and asked respondents whether their community had become a better place to live, had gotten worse, or stayed about the same; whether they felt more or less safe or about the same in the community; and whether they believed gang violence had increased, remained the same, or decreased (e.g., Baldassare & Associates 1994; Flanagan & Longmire 1996; Taylor & Hale 1986). These questions were coded so that a higher number indicated more negative change (e.g., crime had increased or the community had gotten worse or less safe).

Figures 1 through 4 present the CFA models for fear and the theoretical variables, including the unstandardized regression coefficients and goodness-of-fit measures. In constructing these models, we allowed the errors of the individual indicators for each theoretical construct to correlate if the model indicated that the correlation was significant. All of our CFA models fit well, producing nonsignificant chi-square values at or above the 0.05 significance level and resulting in strong values for other measures of fit (goodness-of-fit index [GFI], adjusted goodness-of-fit index [AGFI], Tucker-Lewis Index [TLI], and Incremental Fit Index [IFI]) (see Figures 1–4). In each of our models presented here (gang fear, diversity/gang fear, disorder/gang fear, community concern/gang fear), the paths between each indicator and the latent construct were significant, demonstrating that the indicators are indeed reflective of the single unmeasured construct. In addition, there were no direct paths between the indicators of the latent variable “gang fear” and any of the indicators of the other latent variables (diversity, disorder, or community concern). These results indicate that the relationships between the indicators of the different latent variables are only through their unmeasured constructs, permitting the treatment of these constructs as variables. We therefore felt justified in creating composite variables representing different latent constructs to be used in our next set of analyses.

Figure 2. CFA for Diversity and Gang Fear

Figure 3. CFA for Disorder and Gang Fear

Figure 4. CFA for Community Concern and Gang Fear

The Theoretical Models

The second step was to test the theoretical models in traditional-path analytical models using single indicators for all variables. This approach allowed us to test the direct and indirect effects of both the key demographic variables and the theoretical constructs on fear of gang crime. We allowed for the differential contribution of each particular community problem or offense to the broader construct by using factor score weights from Step 1 (the CFAs) to develop the theoretical and fear indexes for the path models.

The exogenous variables in the path analyses included the following demographic characteristics: sex (“female”), age (“older”), race/ethnicity (“white”), education (“education”), home ownership (“owner”), and district of residence (“central”) (see Table 1 for coding scheme). The central district of the county was the area with the most gang crime, and this variable served as our proxy for objective risk of victimization.⁶

The endogenous variables included the three theoretical constructs and fear of gang crime. We used the factor score weights produced in the CFA for gang fear (Figure 1) to construct our gang fear index, because we wanted to examine gang fear as one construct while also controlling for prior researchers' assertions that people fear different crimes differently (Ferraro 1995; Ferraro & LaGrange 1988; LaGrange & Ferraro 1987, 1989; Warr 1994, 2000). We created an index score by multiplying each individual indicator (offense) by its factor score weight, adding these values, and then dividing by the total number of items. The factor score weights for the offenses were as follows: graffiti=0.097, home invasion robbery=0.293, drive-by=0.119, gang assault=0.236, gang harassment=0.083, and carjacking=0.177.

We also created one variable to measure each theoretical construct (three variables in total) using the same process to produce indexes. The diversity index included the four community problems shown in Figure 2. The factor score weights produced in the CFA for these variables were as follows: language differences between residents=0.248, cultural differences between residents=0.267, racial differences between residents= 0.273, and people moving in and out without becoming attached to the community=0.089. The disorder index included the eight community problems shown in Figure 3. The factor weights for the individual indicators were as follows: poverty=0.132, rundown houses=0.079, abandoned houses=0.081, graffiti=0.213, too many people in one residence=0.075, gunfire=0.163, and youths hanging out=0.120. The community concern (decline) index variable used the following indicator factor weights from the CFA presented in Figure 4: community change=0.217, gang violence change=0.133, and safety change=0.340.

We ran four separate path models, one testing each theoretical model individually (total of three) and one testing the theoretically suggested connection between disorder and community concern in producing fear. Based on previous research findings, we expected women, minorities, people with less education, renters, and people living in the more crime-ridden area (central) to be more afraid of gang crime (see Warr 1994 for a review). In our sample, these groups were more likely to have lower household incomes and, we believed, were more likely to live in more socially disorganized areas (e.g., the central district).⁷ We also expected these groups to perceive more diversity and disorder problems in their communities and therefore to also perceive more community decline and consequently be more fearful (see Taylor & Hale 1986). Previous findings regarding the effects of age are inconsistent. Early studies found older people to be more afraid, but more recent studies, especially those measuring offense-specific fear, have found that younger people may be more fearful (e.g., Chiricos, Hogan, & Gertz 1997; Ferraro 1995; Ferraro & LaGrange 1987, 1988; LaGrange, Ferraro, & Supancic 1992; McCoy et al. 1996; Rountree 1998; Rountree & Land 1996; Warr 1994). We expected that younger people were more likely than older ones to come into contact with gang members during their everyday activities, based on younger people's greater likelihood of living in poorer areas (e.g., more likely to rent) and possibly based on lifestyles (e.g., seeing them in bars or at school). Consequently, we anticipated that age would be negatively related to fear of gangs as well as diversity, disorder, and community concern.

Findings

Diversity

In Figure 5, we present the path model testing the diversity perspective—or the idea that people who are more afraid are those who see cultural, racial, and language differences among local residents as a problem and who worry about people moving in and out without becoming attached to the community. To allow comparison across models, our figures report the standardized coefficients for the paths with significant t-values (see Table 2 for unstandardized coefficients and standard errors). Our figures report the following fit indexes: the chi-square goodness-of-fit statistic, the AGFI, the GFI, the TLI, and the IFI. The diversity model provides a good fit and explains 12.7% of the variance in gang fear. Being female, younger, and minority has a direct significant impact on gang fear, independent of concerns about diversity. Diversity also has a direct, significant effect on fear. Being a renter and living in the central district of the county have an indirect impact on gang fear through concerns about diversity. None of the other demographic variables have an indirect effect on fear through diversity. Education has no significant effects and was dropped from the model.

Figure 5. Path Analysis for Diversity and Gang Fear

Table 2. Coefficients for Path Models: Unstandardized Regression Estimates and Standard Errors

Disorder

We then tested the simple disorder model, or the idea that people who see incivilities—or evidence of disorder—in their community are more afraid of gangs. Our disorder model explains slightly more variance in fear (13.9%) than the diversity model and also provides a good fit (see Figure 6). As in the diversity model, this path model shows that being female, younger, and minority have a direct, significant, independent effect on fear, but being a renter and living in the central section of the county have only indirect effects, now through disorder. Disorder is positively and significantly related to fear, and its effect is stronger than that of diversity. This model is different from our diversity model in that here being a minority also has an indirect relationship through disorder—minorities who perceive more disorder in their communities are also more afraid of gangs. Another difference is that education has an indirect, negative effect on fear through perceptions of disorder.

Figure 6. Path Analysis for Disorder and Gang Fear

Community Concern

Our third model illustrates the results testing the simple community concern perspective, or the idea that people who are concerned about community decline are more afraid of gangs. This model explains even more variance in fear (16.7%) than do the previous two and again provides a good fit (see Figure 7). As in the previous two path models, females, younger people, and minorities are significantly more fearful, independent of their perceptions of community change. Community concern has a direct, significant effect on gang fear, and its effect is stronger than that of the other two theoretical variables. As in the disorder model, there is an indirect relationship between race/ethnicity and fear, but now whites are more likely to perceive community decline, and those that do are more afraid of gang crimes. This is also true for women. In addition, the indirect relationships between renters, those living in the central district, and theoretical variables do not hold in the community concern model. Both being a homeowner and living in the central district drop out of the model.

Figure 7. Path Analysis for Community Concern and Gang Fear

Disorder and Community Concern

To test the possibility that community concern is an intervening variable between perceptions of disorder and fear of gang crime (e.g., Garofalo & Laub 1978; Lewis & Salem 1986; Taylor & Hale 1986), we created another path model that included both theoretical variables (Figure 8). Perceptions of disorder do indeed significantly predict community concern and, therefore fear of gang crime. People who perceive more disorder problems in their communities also perceive more community decline, thereby causing them to be more afraid. This model explains 16.1% of the variance in gang fear and again provides a good fit.⁸

Figure 8. Path Analysis for Disorder, Community Concern, and Gang Fear

As in all three previous path models, females, younger people, and minorities are significantly more afraid of gangs independent of their perceptions of disorder or community decline. Both disorder and community concern still have direct, significant impacts on fear of gangs, but the community concern effect is stronger. Some paths in this model are similar to those in the simple disorder model. Specifically, minorities, less-educated people, renters, and those living in the central district perceive more disorder problems and, therefore, are more afraid. And in this model, they are also more concerned about community decline, which contributes to their fears. As in the simple community concern model, whites are significantly more concerned about community decline and hence are more afraid—regardless of their perceptions of disorder. Interestingly, in this model, the significant effect between sex and community concern drops out.

Conclusion and Discussion

Our Research Questions: Comparing Models

We began the study with two primary research questions. First, we asked whether the relationships between our demographic, theoretical variables and fear of gang crime parallel the relationships found among these variables in studies that examine nongang-related crimes. We found some similarities and some differences. Second, we wondered if there are indirect relationships among our variables that are not evident in other types of analyses (e.g., ordinary least-squares regression equations). We found that there are such relationships and that they are different depending upon the theoretical model tested (see Lane & Meeker 2000).

Our findings are similar to the findings of Taylor and Hale (1986) and Covington and Taylor (1991) in that our models indicated that each theoretical perspective helps explain fear of gang crime. In each path model, the theoretical variables had direct, positive, significant effects on fear of gangs. Demographic characteristics were also important. We found that our model R²s for the path analyses explain more variance in fear (ranging from 12.7 to 16.7%) than Taylor and Hale's (1986) models predicting general fear did (about 10%).⁹Covington and Taylor (1991) used different measures than we did here, but found that perceived incivilities (disorder) have the strongest impact on fear of crime. When we examined each theoretical construct independently, we found that community concern was our strongest theoretical predictor; yet the model with both disorder and community concern was a similarly strong model (Taylor & Hale 1986). Our findings for the combined disorder and community concern model are not as strong as those found by McGarrell, Giacomazzi, and Thurman (1997), possibly because their model included additional variables not examined in our models (e.g., neighborhood social control, police responsiveness, adult responsibility).

In our earlier fear of gangs study conducted in the same locale, we found that diversity was a stronger predictor of fear of gangs than fear of crime more generally (Lane & Meeker 2000). We again found here that diversity is a significant predictor of fear of gangs, but here its direct effect is smaller than the direct effects of disorder and community concern. Our findings are also similar to our earlier findings in that renters and those people living in the high-crime area are more concerned about diversity and, therefore, more afraid. In addition, both studies showed that minorities are more afraid. Our first study found that older people are more afraid, but sex is not a significant predictor (Lane & Meeker 2000). In contrast, this study found that females and younger people are both more afraid. We also found that education is not significant after controlling for the other factors. We expect that the differences between the two studies are related to differences in how they measured fear. Our former study asked about general “worry” about neighborhood gangs, but the current one asked about fear of specific gang-related offenses. More studies on the relationships between demographics, diversity, and fear will help further our understanding of the importance of these predictive factors.

Similar to most previous fear of crime studies, we found in all of our path models that women and minorities (primarily Latinos in this sample) are more afraid of gangs independent of other factors, but we also found evidence of interesting indirect relationships (e.g., Covington & Taylor 1991; McGarrell, Giacomazzi, & Thurman 1997; Taylor & Covington 1993; Taylor & Hale 1986; see Warr 1994 for review; cf. Lane & Meeker 2000). In the simple community concern model, women are also more afraid when they worry about negative community change. Interestingly, minorities perceive more disorder in their communities and are therefore more afraid of gangs, but whites are more worried about community decline, and those who perceive negative changes are more afraid. These findings may indicate an ethnic split in the relative importance of disorder versus community decline. It may be that minorities are more likely to have lived in neighborhoods with more disorder (and gangs) over the long term and may not experience as much “decline,” but whites may be more likely to see their neighborhoods changing for the worse—e.g., becoming more urban or less safe, whether or not disorder itself is believed to be a serious problem. It is also possible that many of our minority respondents were recent immigrants and were not in the area long enough to experience long-term negative change. Future studies might disaggregate samples by race and ethnicity and examine differences in length of stay in stable (bad or good) as opposed to negatively changing neighborhoods and how these experiences and perceptions affect fear of crime and gangs (see Lewis & Salem 1986; Taylor & Covington 1993; Skogan 1986; Taub, Taylor, & Dunham 1984).

We consistently found that younger people are more afraid of gangs independent of other factors. For decades, research found that older people were consistently more afraid, even though younger people faced more objective victimization risk (see Warr 1994 for a review; Covington & Taylor 1991). This was one of the findings that led researchers to focus on community factors as important predictors of fear in their attempt to explain this “paradox of fear” (Warr 1994:12). Recent studies that have measured fear of specific offenses (rather than using general fear questions) have found that younger people are more afraid or that the elderly are not as fearful as many assume (e.g., Ferraro 1995; Ferraro & LaGrange 1988; LaGrange & Ferraro 1987; McCoy et al. 1996; Rountree 1998; Rountree & Land 1996). In essence, the age–fear relationship is more complex than once thought, and we still have more to learn.¹⁰ We believe our finding that younger people are more afraid of gangs is logical, since we expect that younger people are more likely to come into contact with gang members in their everyday lives, because they are more likely to live in poorer areas (e.g., more likely to rent), and possibly because of their other activities (e.g., seeing gangs in bars or at school).

Another interesting finding is that being a renter and living in the central district of the county have no direct effects on fear of gang crime, but only indirect effects through concerns about diversity and disorder. Given the context, this is not surprising. The central district of the county has more renters than other areas and includes the cities experiencing the most ethnic diversity—especially Latino immigrants—social disorganization, crime, gangs, and gang crime (Capizzi 1998; Gaquin & DeBrandt 2000).

Theoretical Implications

This is the first study using the same sample to measure the impact of these theoretical perspectives on fear of gangs specifically. Our results confirm most of the results of previous studies. Using a more sensitive weighted fear index composed of specific crimes, we found that the key theoretical factors found in general fear of crime studies have value specifically for predicting fear of gangs (e.g., Covington & Taylor 1991; McGarrell, Giacomazzi, & Thurman 1997; Taylor & Hale 1986). Future studies that examine how these theoretical models apply to other specific crime-related fears, especially other seemingly random crimes that make the news (e.g., school shootings, abductions, other types of homicide), will give us important new information about how these theoretical perspectives apply to different people in different times, places, and social contexts.

We found that concern about diversity has a direct, significant effect on fear. Residents who are worried about cultural, language, and racial differences, as well as people moving in and out without becoming attached the community, are more afraid, which corroborates our earlier results (Lane & Meeker 2000). This finding supports Merry's (1981) argument that cultural distance is an important element in producing fear among strangers in urban environments (see also Covington & Taylor 1991). She argued that when residents do not understand others who are culturally and behaviorally different, they see these others as “dangerous,” especially in urban environments where strong friendship ties may be hard to form (Merry 1981:239). These findings confirm that her theory applies to fear of gangs, which in Orange County are typically both ethnically and culturally different from most residents.

Here, we find that the simple community concern (decline) model and combined disorder and community concern model are the most predictive of the models tested (Taylor & Hale 1986). Combining more theoretical perspectives into one statistical model seems to be an important next step (e.g., McGarrell, Giacomazzi, & Thurman 1997). Because we were interested in determining whether results from general fear of crime studies apply to fear of gangs, we created our theoretical models based upon prior literature and did not develop “new” models. However, our qualitative research in Orange County found that some residents believe that diversity specifically created by increasing undocumented Latino immigration to the area leads to disorder, which leads to community decline and, therefore, to their fear of gangs. These beliefs are fueled both by direct experiences with changing neighborhoods over time and through indirect victimization, primarily from local police officers and neighbors (Lane 1998, 2002). It may be that worry about diversity has indirect effects on fear through its effects on perceptions of disorder and decline (reflected in community concern), which may explain the weaker direct effect when diversity is in a model by itself. We plan to statistically test this possibility regarding fear of gangs but also wonder if combining models may yield important results regarding fear of other types of crime. It may also be that changes in diversity, rather than just diversity alone, is a key indicator causing people to believe their communities have more disorder and decline, which is another interesting avenue of study (Lane 2002; Taylor & Covington 1993).

Our indirect relationships, especially the finding that minorities are more concerned about disorder but that whites are more concerned about community decline, indicate that disaggregating samples by race/ethnicity and possibly other demographic characteristics (e.g., gender, income, neighborhood characteristics), might allow for different indirect relationships to emerge. As noted earlier, such analyses may allow us to refine our theories for different populations and different contexts.

Policy Implications

Like most prior fear of crime studies, our results showed that fear of gangs is not solely about crime and the objective risk of victimization. Consequently, focusing only on eradicating gangs and creating harsher suppression and punishment strategies likely will not eradicate fear of gang crime. Although being tough on crime can be a safe and successful political strategy, it is costly in both financial and human terms and has not substantially decreased the number of gangs or gang members in the United States (National Youth Gang Center 2000). Although we have no data on changes in fear of gangs over time and cannot know if specific suppression policies have made residents feel substantially safer, we expect that generally they have not.

Like other types of fear, fear of gangs is related to broader concerns about other problems in the community, such as diversity, disorder, and decline. Policymakers who want their constituents to be more satisfied with their lives, communities, and even their leaders may need to take a broader approach to addressing public concern and fears about gang crime. Rather than primarily focusing on separating the “bad” people from the “good” via suppression and incarceration, policymakers might be able to make citizens feel safer by helping to better the relationships among people who live and work together in communities. In other words, policies that focus on building and strengthening community ties may be as important as those focused on crime suppression (e.g., Merry 1981; Taylor 1996).

This study indicates three related community factors that community leaders can address toward this end—diversity, disorder, and community decline. Diversity issues such as language, cultural, racial differences, and residential mobility are one set of concerns leading to fear of gang crime in this setting, which may be tied to residents' ability to address their disorder and decline problems (see Merry 1981; Taylor 1996). In Orange County, most gang members are of color, primarily Latino, meaning that they are often culturally different and bilingual and sometimes monolingual Spanish-speakers (see Capizzi 1998). Residents express concerns about their ability to talk to their neighbors and relay expectations about behavior and lifestyles, which could help them predict what will happen in the future (Lane 1998; Lewis & Salem 1986).

Although encouraging participation may be a challenge, leaders can help by creating programs to increase cultural competency among residents. In addition, they might provide appropriate facilitators or mediators (e.g., bilingual and/or community residents themselves) to help residents work together to understand different priorities and solve community problems. Community policing and neighborhood watch programs could be expanded to include increasing understanding and cooperation among culturally different groups as a primary focus. Gang members themselves might be invited to join the efforts to decrease the social and cultural distance between them and other community members. These ethnic, cultural, interest, and law enforcement groups could work in collaboration, rather than against each other, to determine the causes and solutions of disorder, decline, and gang crime. Because these residents live in an urban environment where they may not have much social contact with their neighbors, they may need help getting to know them, especially those who seem scary and culturally different. As Taylor (1996) found, when residents know each other better, they are more likely to feel more invested and interested in the community, which makes them feel less vulnerable and better able to respond to disorder problems.

Our study and others indicate that concern about disorder and community decline can work together to produce fear, although not for everyone (e.g., Garofalo & Laub 1978; Skogan 1990; Skogan & Maxfield 1981; Taylor 1991; Taylor & Hale 1986). Previous research has shown that communities around the country have been able to reduce fear by addressing disorder and decline problems (Pate et al. 1987; Skogan 1990). Our research confirms that code enforcement and community improvement efforts are valid and important approaches to making people feel safer, because they may decrease perceptions of disorder and decline. For people who do not worry as much about disorder but more about community decline, it is important to know what other indicators tell them the community is changing. Helping people learn about each other and work together to address community problems may be a critical step in moving beyond the “us versus them” approach to improving community safety.