fear of crime among korean americans in ... - Wiley Online Library

4 downloads 0 Views 2MB Size Report
Furthermore, in Chicago, blacks picketed Korean stores in the Roseland area in 1991 as a ...... Bryk, Anthony S. and Stephen W. Raudenbush. 1980. 1992 .... Thompson, Carol Y., William B. Bankston, and Roberta L. St. Pierre. 1992. Walker ...
FEAR OF CRIME AMONG KOREAN AMERICANS IN CHICAGO COMMUNITIES MIN SIK LEE Korean Institute of Criminology

JEFFERY T. ULMER Penn State University

This study examines Korean Americans’ perceived incivilities, perceived crime risk, and fear of crime using an explanatory model combining group threat theory of racial hostility and risk interpretation theory of fear of crime. In particular, our hierarchical linear models show strong effects on fear of crime for English proficiency, length of U.S. residence, preference for ethnic Korean media, perceived risk of future black rioting, and anti-black prejudice. We discuss the importance of cultural factors and the dynamics of race and ethnic conflicts in explaining fear of crime, and suggest directions for future research on race relations, perceived victimization risks, and fear of crime. Fear of crime is conceptually and empirically distinct from actual rates and severity of crime (Lindquist and Duke, 1982; Smith, 1987). Fear of crime is widespread in the United States, and it has detrimental consequences for both individuals and communities (Hale, 1996). Scholars have noted a link between race and ethnic group relations and fear of crime, but prior research on racial and ethnic differences in fear of crime has mostly compared whites and blacks (Braungart et al., 1980; Clemente and Kleiman, 1977; Ferraro, 1995; Parker, 1988; Parker and Ray, 1990; Rohe and Burby, 1988). Very little is known, however, about fear of crime among other minority groups, such as Asian Americans, and how such fear might be related to race relations (but see Walker, 1994). This constitutes a gap in the literature on fear of crime in particular and race and ethnic relations in general, because Asian Americans constitute a diverse and rapidly growing minority in the United States. This study draws on perceived group threat theories of racial hostility in combination with Ferraro’s (1995) risk interpretation theory to explain fear of crime among KoreanAmericans, using survey data from a sample drawn from Korean-American ethnic churches1 1. We use the term Korean Americans to indicate Koreans who reside in the United States (i.e., Chicago area) at the time of the survey. They may or may not be U.S. citizens.

CRIMINOLOGY VOLUME38 NUMBER4 2000

1173

1174

LEE AND ULMER

GROUP THREAT THEORIES OF RACIAL HOSTILITY Herbert Blumer’s (1958) theory of racial conflict and prejudice emphasizes a given group’s sense of superior group position, and a perception of threat to that group’s interest or position from another group that is seen as inferior. Blumer’s theory has recently received strong empirical support from Quillian (1995) and Bob0 and Hutchings (1996). Bob0 and Hutchings (1996:968) state: Blumer’s group-position model provides the most comprehensive theoretical leverage and goes further to emphasize that identity, stereotypes, values, and assessments of interests are shaped historically and involve a collective and relational dimension between groups that powerfully engages emergent normative ideas about appropriate group statuses and entitlements. Marylee Taylor’s (1998) study of whites’ anti-black prejudice examines the role of perceived threat in such prejudice. Although Taylor (1998) fails to find a significant role of economic or political threat in whites’ antiblack prejudice, she calls for further research assessing the role of perceived “status, physical, or cultural threats” in racial hostility. One kind of perceived threat Taylor (1998) singles out in particular is the personal threat embodied in fear of crime. We thus answer Taylor’s (1998) call by examining race and ethnic group position, relations, and perceived threat in the form of fear of crime. THE KOREAN COMMUNITY IN CHICAGO Contemporary Korean communities in major US.cities, including Chicago, are largely the byproduct of the Immigration Act of 1965 (Min, 1996).2 Language barriers, lack of cultural knowledge about American society, lack of economic resources, and other labor market disadvantages pushed many Korean Americans into an economic niche of small retail business (Min, 1984). Survey data show that about 30% of working Koreans in the Chicagoland area were self-employed in small retail or service businesses (Kim and Hurh, 1993). Most of these businesses are small and 2. A unique group of Koreans, coal miners and nurses, came to Chicago from West Germany beginning in 1968. These Koreans not only remigrated from Germany to the United States but intended to settle permanently (in contrast to earlier waves of Korean immigrants, which primarily consisted of students who did not intend to stay). In this sense, Korean miners and nurses from West Germany have played an important role in establishing a permanent Korean community in Chicago, and to some extent in other United States cities (Ahne, 1995). It is interesting to note that these Korean miners were not miners by training or by experience in their homeland. The majority of them were highly educated-in college or at least in high school-but often unemployed young adults in their twenties. They were nicknamed Haksa Kwangbu, meaning “miners with college degrees.”

FEAR OF CRIME AMONG KOREAN AMERICANS 1175 family-owned, and they often serve low-income black communities. They located there mainly because of the opportunities available after the exodus of the white, mostly Jewish, merchants in the 1970s (Kim and Hurh, 1993). The 1997 survey by the Korea Central Daily of Chicago (a KoreanAmerican newspaper) identified about 500 of these types of Koreanowned businesses in 17 major Chicagoland areas, primarily in the predominantly black South and West sides of Chicago City. The president of the Federation of the Southside Korean Merchants estimates the number to be even higher-between 800 and 900-including those unidentified stores in scattered areas. In white neighborhoods, Korean-owned small stores have been unable to compete with large chain supermarkets. On the other hand, the low purchasing power of residents and high crime rates, coupled with negative racial stereotyping and stigmatization of black inner-city neighborhoods, have resulted in a dearth of investment in black inner-city neighborhoods by large chains and independent white business owners (Wilson, 1996). Moreover, capital requirements to operate a store, including rents, are significantly lower in low-income minority areas. Thus, comparatively disadvantaged Korean immigrants can open small stores in these areas without much competition (Light and Bonacich, 1988; Min, 1996; Min and Kolodny, 1994). In addition, according to a 1998 survey by the Korea Central Daily of Chicago, the Chicago Korean community has the largest number of ethnic media among other Asian ethnic groups in the Chicago area. There are three daily Korean-language newspapers, two weekly newspapers, three Korean television programs, and two Korean-language radio shows. The Korean-American media are probably the most developed ethnic media in the Chicago area, and they have the highest home subscription rates (approximately 10,000 daily subscribers) (Ahne, 1995). These media go beyond mere news services; they are a very influential mechanism in Korean immigrant life. They provide publicity about Korean community issues, and they air opinions, provide educational information about American life, and news from South Korea. Videotape rentals of Korean drama and movies are also highly popular. Korean immigrants in the United States have confronted comparatively little overt hostility from whites and Hispanics (Cheng and Espiritu, 1989; Min, 1996). However, as symbolized by the highly publicized events surrounding the 1992 L.A. riots, relations between Korean Americans and African Americans have been strained in many cities for at least two decades (Min and Kolodny, 1994). In the 1 9 8 0 ~€or ~ example, New York City Korean communities experienced four long-term boycotts of Korean stores by African-American groups (Min and Kolodny, 1994). Events

1176

LEE AND ULMER

associated with the 1992 L.A. riots provide an especially prominent example. During the riots, approximately 2,300 Korean businesses were looted or burned, one Korean was killed, and 46 Koreans were injured (Korea Times Los Angeles, May 23, 1992). The psychological impacts of the L.A. riots ran deep and were not confined to the Korean-American victims who experienced personal financial loss or physical injury. A situation of vicarious victimization was evident among the general Korean population who lived in other large American cities, including Chicago, because many of them engaged in small businesses in black communities, read Korean media accounts of the riots, and perhaps communicated with friends or relatives in the L.A. area (see KimGoh et al., 1995). Furthermore, in Chicago, blacks picketed Korean stores in the Roseland area in 1991 as a reaction to a highly publicized conflict between Korean greengrocers and a group of blacks in New York City. More importantly, Korean businesses in Chicago were also major targets of looting and rioting in black neighborhoods after the Chicago Bulls won the 1992 NBA Championship. In fact, the Chicago Bulls riots of 1992 and similar but smaller incidents are said to have seriously weakened the Korean business community, especially in Southside Chicago (The Korean Central Daily, Chicago, November 19, 1997).

FEAR OF CRIME: A MODIFIED RISK INTERPRETATION MODEL Although a great deal of research exists on fear of crime, perhaps the most complete general theory of fear of crime is Ferraro’s (1995) risk interpretation model. An individual’s fear of crime is one alternative emotional response to hidher definition of situations and social objects as presenting risks of criminal victimization. According to Blumer (19693, “the actor selects, checks, suspends, regroups, and transforms the meanings [of things] in the light of the situation in which he is placed and then directs his action.” Individuals thus form definitions of situations from their preexisting experiences, knowledge, and attitudes, and interpretations of the information presented by social environments and social relations. According to Ferraro (1993, an individual’s assessment of risk involves interpretation of one’s exposure to the chance of injury or loss. Estimation of risk thus entails defining a potentially problematic situation. One can never be certain of the risks of victimization. One can only gather the information related to the risk and make a judgment about it, and fear is only one of several reactions to judgments of high risk. The advantage of Ferraro’s model is that it is more comprehensive than-and in fact encompasses-personal vulnerability and victimization

FEAR OF CRIME AMONG KOREAN AMERICANS 1177 models of fear of crime, and it is more parsimonious than other general models (for example, see Garofalo, 1981; Gordon and Riger, 1989; Skogan and Maxfield, 1981). Another advantage is that the risk interpretation model specifies and disentangles distinct interpretive processes: judgment of risk and emotions of fear. Furthermore, the risk interpretation model is an individual- and community-level theory. To leave out either individual experiences and perceptions or community-level characteristics is to have an impoverished and incomplete explanation of fear of crime. Investigations of crime and fear of crime that focus on community environmental factors typically emphasize the breakdown of social cohesion and formal and informal social control as consequences of social disorganization. Although the risk interpretation model clearly recognizes a variety of potential community environmental influences, Ferraro’s (1995) research only examined two community-level factors-crime rates and metropolitanhonmetropolitan status. This study thus extends Ferraro’s risk interpretation approach by incorporating and testing a wider variety of ecological factors. High crime rates, racial and ethnic heterogeneity, poverty, single-headed households, and a high concentration of young people in the “crime prone” age group (15-24 years) are said to be indicators of social disorganization (Bursik and Grasmick, 1993; Sampson and Groves, 1989; Skogan and Maxfield, 1981). These factors may also lead to increased perceived incivilities, perceived risk, and fear, and we therefore include them in our measures and models. We adapt Ferraro’s model to include factors particularly relevant to the Korean-American situation. The historical, economic, and cultural characteristics of Korean immigrants in the United States are unique among Asian Americans. First, Koreans are relatively more recent immigrants to the United States compared with most other Asian ethnic groups (Min, 1996), and they have been relatively slower to assimilate into American society culturally and economically. They also display a very high degree of ethnic cohesion (Kim and Hurh, 1993; Mangiafico, 1988). Among the several ethnic institutions in Korean-American communities, two play crucial roles in Korean-American ethnic cohesion. First, Korean ethnic churches serve four major social purposes beyond just religious activities (Ahne, 1995; Min, 1992): (1) providing fellowship for Korean immigrants, (2) maintaining Korean cultural traditions, (3) providing social services for church members and the Korean community as a whole, and (4) providing social status and positions for Korean immigrants.3 These features of ~~~~~~~

~

~

3. Various types of Korean-American ethnic associations, such as “Dong Chang Hoe” (graduates’ associations), “Hyang Woo Hoe” (association based on regionalism), and “Jong Chin Hoe” (association based on kinship) play important roles in producing

1178

LEE AND ULMER

Korean ethnic churches have key implications for this study’s sampling and data collection. Second, Korean-American newspapers serve as a crucial source of information about the Korean-American community as well as the mainstream society, and they are much more widely read among Korean Americans than are mainstream English newspapers (Kim and Hurh, 1993). Although acculturation and ethnic attachment-the two processes of immigrants’ adaptation to the host society-are not mutually exclusive, they may be contradictory to each other in their influences on fear of crime among immigrants. For example, a high degree of assimilation toward the host society means greater social ties with the host, which are, in turn, associated with less mutual prejudice and hostility. Therefore, a higher level of acculturation is likely to be linked to a lower level of perceived risk and fear. In contrast, a strong ethnic attachment may be associated with in-group insularity and suspicion of outsiders, and it may thus foster perceived risk and fear. We use the following four factors as indicators of acculturation: (1) English language proficiency, (2) friendship with non-Korean Americans, (3) nativityhmmigration status, and (4)length of residence in the United States. We use consumption of Korean ethnic media as an indicator for degree of ethnic attachment. In explaining fear of crime among Korean Americans, the exposure to ethnic media measures the degree of a respondent’s ethnic attachment, whereas the exposure to the American media measures the degree of a respondent’s exposure to violence in the media as a potential factor of heightened fear. Generally, Korean ethnic media (especially TV programs or movies) tend to be less violent than are American media. Korean Americans may prefer Korean ethnic media because of greater comfort with the language, repulsion at widespread violence seen in ethnic cohesion. However, the Korean-American church is crucial because it serves as a “community center,” often encompassing these other types of associations. The role of Korean Christian churches in the Chicago area provides an important rational ground for our church-based sampling strategy. According to Ahne (1995),the Korean church has been a center of Korean culture and social support in America. During the heavy immigration influx of the 1970s and early 1980s, Korean pastors provided personal and social services for members and nonmembers alike, including driving lessons, job hunting help, translation, help in dealing with utility companies and government offices, and business consultations. Many contemporary Korean churches, in addition to the typical church functions, offer classes in Korean language, history, music, and customs for the American-born generation in an effort to retain their heritage. Thus, a part of the remarkable Christian church attendance among Korean Americans may be explained by the apportunities for preserving and participating in Korean culture that the churches provided. Overall, the author concludes that in the United States, “the history of the Korean community is the history of the Korean church” (Ahne, 1995468).

FEAR OF CRIME AMONG KOREAN AMERICANS 1179 American media, and desire for news, people, amusements, and culture from their mother country. This preference for ethnic media further hampers their acculturation toward the host society. In addition, Korean media also carry news of organized ethnic associations among Korean Americans, and they provide them with common topics of conversation. Thus, ethnic media strengthen ethnic cohesion among Korean Americans, and therefore, they may foster perceptions of high victimization risk and fearful images about U.S. crime, particularly black crime. Earlier, we described the often strained relations between Korean Americans and African Americans in several urban areas, including the Chicagoland area. Jo (1992) argues that Korean-black conflicts come from a variety of sources, such as the lack of understanding of different cultures, communication problems, economic competition, anti-black prejudice on the part of Koreans, Koreans’ status anxiety arising from incongruence between their previous status4 and their current one, blacks’ experience with nonblack shop owners who have exploited black communities in the past, and blacks’ feeling of powerlessness in stopping the influx of both whites and Asian groups into their community. In conclusion, Jo expressed the tension between Korean and African Americans as “prejudice between the victims of prejudice.” Stereotypes and hostilities toward African Americans might be one important factor in predicting fear of crime among Korean Americans. Therefore, we expect that Korean Americans’ perceptions of the severity of racial conflicts, as well as their negative prejudices toward African Americans, will be positively associated with their perceived risk and fear of victimization.

DATA AND METHODS The research sites of this study are Chicago City and three suburban counties-Cook, DuPage, and Lake counties-in the Chicago metropolitan area. According to the 1990 Census of Population and Housing, as of 1989. the total population of Korean Americans dwelling in the three counties, including Chicago City, is 35,197.5 It is very difficult to design and practice a strict probability sampling in a study of recently immigrated, small-minority groups, such as Korean Americans, because recent 4. As a result of the post-1965 Immigration Act, Korean immigrants (particular those who came in the 1970s) were highly educated and generally held white-collar and prafessianal occupations in Korea. 5. About 84% (29,687 persons) of this population reside in Cook County, 10.2% (3,587 persons) in BuPage County, and 5.5% (1,923 persons) in Lake County. Also, about 47% (13,863 persons) of Korean Americans in Cook County live in Chicago City.

1180

LEE AND ULMER

and accurate population data for them are not typically available and the members of such groups tend to migrate frequently (see Ackah, 1992). The present study therefore uses a purposive sampling, based on a key Korean-American ethnic association, Korean ethnic Christian churches. As noted earlier, Korean Americans are well known for their active participation in ethnic Christian churches. Approximately 70% of Korean immigrants regularly attend ethnic Christian churches (Hurh and Kim, 1984,1988; Kim and Hurh, 1993; Min, 1988,1989,1992). In Korea, Christians constitute only about 21% of the population (Korean National Bureau of Statistics, 1987), but the majority of Korean immigrants in the United States have had a Christian background in their home country (Park et al., 1989, cited in Min, 1992:1376). A 1986 survey in Chicago (Hurh and Kim, 1990) also indicates that about 53% of Korean immigrants were Christians in Korea. By contrast, Buddhists constitute a very small fraction of Korean immigrants. The 1986 Chicago survey shows that only 4.2% of the subjects were affiliated with Buddhist temples. The growth in the proportion of Christians among Korean immigrants from approximately 50% in Korea to 70% in the United States indicates that roughly 40% of non-Christian Korean immigrants have become Christians since their immigration (Min, 1992). According to the 1997 directory of Korean churches in the Chicago area (from the Korean Churches Federation of Greater Chicago), there are 187 Korean churches in the three counties. From these, we selected 33 relatively large churches (18%). The numbers of churches selected by area roughly follow the proportions of Korean-American population by area.6 However, relatively more churches were selected for Chicago City, because Chicago City contains a large number of Korean churches with smaller memberships than do those in suburban areas. The data were collected through a questionnaire survey, and respondents could select either an English or Korean version of the questionnaire. The first author administered the survey from August through October 1997. He visited each sampled church and distributed questionnaires to adults (18 years old or over) attending church services. They were returned by postage-paid mail. A total of 3,195 questionnaires were distributed, and 780 responses were collected. Although this response rate of roughly 25% is low, the actual response rate may have been somewhat higher to some unknown degree. A total of 3,195 surveys were distributed at the churches, but this does not mean that there were actually 3,195 potential respondents. In any case, this response rate (along with the fact 6. Sixteen churches were selected for Chicago City, 15 for suburban Cook County, and 1 for each of DuPage and Lake Counties.

FEAR OF CRIME AMONG KOREAN AMERICANS 1181 that this is a nonprobability sample) should be a caution against generalization to the general Korean-American population. Fifty-nine cases were excluded from the analyses because of missing data, or because respondents resided outside of research sites. This left 721 cases for analysis, and those subjects were nested within 63 out of 199 official community areas (as recognized by city and county government maps and political districts) in Chicago City and Cook, DuPage, and Lake counties.’ The distribution of the sample roughly follows the proportions of the Korean-American population by area, but suburban areas were somewhat overrepresented.8 MEASURING FEAR OF CRIME Many recent studies critique “global measures” of fear of crime (such as measures used in the National Crime Survey or the General Social Survey) and suggest using more specific measures (see Ferraro and LaGrange, 1987; Kahle, 1994; Thompson et al., 1992; Warr and Stafford, 1983). Ferraro and LaGrange (1987) argue that such global measures generally fail to distinguish perceived risk from fear, and they tend to be conceptually nebulous. More appropriate measures, developed by Warr and Stafford (1983) and Warr (1984, 1985, 1987) and used by Ferraro (1995), ask respondents to describe how afraid they were of becoming a victim of each of 16 (or 15) different criminal offenses. Also, by asking the levels of fear by type of crime, such questions allow researchers to make empirical observations concerning the relative intensity of fear caused by different kinds of potential victimization. Although Warr and Stafford (1983) used 16 items, the present study uses 11 offense items, including six items for the Uniform Crime Reports index crimes (except larceny-theft and arson). Those items were as follows: (1) being assaulted by someone with a weapon, (2) being cheated or conned out of your money, (3) being murdered, (4) having your car stolen, (5) being robbed or mugged on the street, (6) having someone break into your home while you are away, (7) having someone break into your home while you are there, (8) having your property damaged by vandals, (9) being sexually assaulted, (10) being approached on the street by a beggar, and (11) receiving a strange or obscene call. For each question item, respondents were asked to rate their fear on a scale of 1 to 10, where 1 7. We recognize that neighborhoods or census tracts, as smaller units, are more proximal contexts for shaping fear of crime. In this study, however, community areas are preferable as the level-2 units because of the small number of Korean Americans within each census tract (many census tracts in the research sites have no Korean Americans). 8. Among the 721 respondents, 35% (254 persons) live in 23 communities of Chicago City and 65% (467 persons) reside in 40 communities of the three suburban counties.

1182

LEE AND ULMER

means “not afraid at all” and 10 means “very afraid.” All 11 items were summed for the fear of crime measure (Cronbach’s alpha in the present study is .94), and the overall fear scale varies from 11 to 110. In addition, we construct similar scales for fear of personal and property crime. The personal fear scale consists of items 1 , 3 , 5 , 7, and 9, and it varies from 5 to 50 (alpha = .93). The property fear scale contains items 2 , 4 , 6 , and 8, and it varies from 4 to 40 (alpha = .90). The fear measures are all standardized (i.e., mean = 0, standard deviation = 1). INDEPENDENT VARIABLES We investigate the influences of personal- and community-level factors on fear of crime among our sample of Chicago-area Korean Americans, as suggested by our modified risk interpretation model. First, at the community level, we control for the effects of crime rates, population density, youth concentration, housing conditions, percent Black, percent Hispanic, percent Korean, poverty, single-headed households, and whether communities are suburban or Chicago metropolitan. Our later analysis tables present the results for only the community-level factors that exhibited statistically significant relationships with perceived incivilities, perceived risk, and fear. Such community characteristics are among the indicators of social disorganization in a local community (see Bursik and Grasmick, 1993). These data are from the 1990 Census of Population and Housing (issued by Bureau of the Census and City of Chicago9) and IJniform Crime Reports (1990-1996). At the individual level, we examine the influences of vulnerability or status characteristics (such as age, gender, health, education, and income), victimization (direct and indirect), self-employment versus employment by others (a proxy for small business ownership), media effects (such as exposure to media violence and general perception of crime severity in the United Stateslo), and perceptions of incivility and risk (which we standardize when we treat them as the outcome variables). Appendix 1 describes the measurements of the community and individuallevel variables. DATA ANALYSIS The data are analyzed by Hierarchical Linear Modeling (HLM). HLM 9. We use census data from Chicago City for statistics of the 77 official community areas in Chicago City. 10. We also include the subjects’ general perceptions of severity of crime in American society as a measure of indirect media effects because such perceptions can be partially formed by the amount and nature of depictions of violence in news and entertainment media (see Altheide and Michalowski, 1999).

FEAR OF CRIME AMONG KOREAN AMERICANS 1183 has proved to be a very effective analytic technique for data having multilevel structures. HLM allows researchers to overcome a number of conceptual and technical difficulties (e.g., aggregation bias, misestimated standard errors, and heterogeneity of regression) that have plagued past analyses of multilevel data (for details, see Bryk and Raudenbush, 1992). Our data exhibit a two-level hierarchical structure. People are nested within a community whose characteristics might influence their perceptions and fear. Therefore, both persons (level-1) and communities (level2) are our units of analysis. In HLM, each level is formally represented by its own submodels (Bryk and Raudenbush, 1992). These submodels express the relationships among variables and residual variability within a given level, and they specify how predictors at one level influence relationships at another. Thus, we can examine relationships at each level and assess the amount of variation explained at each level.11 HLM encompasses a variety of submodels that enable multilevel analyses of nested data (Bryk and Raudenbush, 1992). In this study, we use three kinds of HLM submodels: one-way analysis of variance (ANOVA) with random effects, analysis of covariance (ANCOVA) with random effects, and a random-intercept model with level-1 covariates. First, the ANOVA model with random effects (i.e., fully unconditional model) is the simplest HLM, and it shows what proportion of total variation in fear of crime exists between and within communities. The estimates of variance components at each level are used as criteria in evaluating the results of subsequent models. Second, the ANCOVA model with random effects shows the effects of person-level predictors on the dependent variables. We evaluate the effects by computing the proportion of reduction in variance of the outcome variable from the previous ANOVA model. Third, and most importantly, the random-intercept model with level-1 covariates represents our full model, and it shows the influences of both communitylevel and person-level predictors. The data in this study have a small average number of level-1 units per level-2 unit and, thus, an unbalanced structure. Because of this data limitation, we focus on the common effects that community characteristics (such as crime rate) exert on each person’s fear within the community, rather than on any differentiating effects (i.e., we cannot examine crosslevel interaction effects). Such community effects modify only the mean level (i.e., the level-1 intercept, Po, of fear for each community. That is, we fix the slopes of level-1 predictors and do not allow them to vary across 11. These analyses use full-information maximum likelihood methods with a combination of Empirical Bayes (Dempster et al., 1977) and Fisher scoring algorithms (Longford, 1987) for fast and stable convergence to the solution optimum.

1184

LEE AND ULMER

level-2 units. We do this because of the small average number of level-1 units (i.e., about 11 persons) per level-2 unit, and because we have no theoretical rationale for allowing specific slopes to vary.

FINDINGS Table 1 shows descriptive statistics on both individual and community characteristics. The age of respondents is distributed between 18 and 82, and mean age is 43. The gender distribution is 50.5% men and 49.5% women. About 26% of respondents reported being victimized during the past three years in the United States, whereas 60% of respondents reported victimization of their family members (or relatives) or other Korean acquaintances. Among the respondents who are engaged in economic activities, 36.2% (73.2% of the total sample) are self-employers or their unpaid family members. About 55% of respondents are US. citizens, whereas 30% of the people have permanent residentship. Among the U.S. citizens, only 8% are native-born Americans. About 65% of respondents have been in America over 10 years, but only 24% have resided at present residence over 7 years. About 42% of respondents report that they read Korean newspaper(s) regularly, and 30% watch Korean TV programs regularly. A total of 70% of respondents are concerned about the potentiality of violent racial conflicts in the Chicago area, and 75% of respondents consider their black neighbors a potential threat. Finally, the average level of fear among respondents was 7.4 points per item. This fear level, along with the levels of property and especially personal crime fear, are much higher than are those reported in previous research on the general American population with similar measures (Ferraro, 1995; Warr, 1984, 1985,1987; Warr and Stafford, 1983). This suggests that fear of each type of crime is prevalent among these Korean Americans. Because the present study adopted a purposive sampling based on Korean ethnic churches, we first examined the effect of religiosity on the outcome variables. The results of these analyses, however, indicate that religiosity is not a significant predictor, and so we do not control for religiosity in the final analyses (the sample is fairly homogeneous in terms of religiosity-about 93% of the respondents reported that they attend church at least once a week). Appendix 2 reports bivariate correlations between fear of crime, perceptions, and person-level predictors. UNCONDITIONAL MODEL The one-way ANOVA with random effects (i.e., unconditional model) provides useful preliminary information about how much variation in the outcome variables (i.e., perceived incivilities, perceived risk, and fear of

FEAR OF CRIME AMONG KOREAN AMERICANS 1185

Table 1. Descriptive Statistics of Variables Variables Community-Level Variables ( N = 63) Crime Rate Density

Youth Concentration House Value Person-Level Variables ( N = 721) Fear of Crime Fear of Personal Crime Fear of Property Crime Perceived Risk Perceived Incivility Age

Gender (Female) Health (Defensibility) Education Income Victimization Indirect Victimization Crime on TV Crime Severity Self-Employed English Proficiency American Friend Natural Citizenship Length of Residence Ethnic Media Future Riot Fear of Blacks

Mean

S.D.

Minimum

Maximum

7249.00 3522.00 13.41 162714.00

11191.00 3158.00 2.75 71681.00

1789.00 650.00 8.75 57000.00

65834.00 13796.00 21.14 500000.00

81.05 41.87 29.55 8.02 12.98 42.54 0.50 2.28 3.63 8.28 0.27 0.81 2.56 3.40 0.26 3.20 0.99 0.05 14.11 4.42 2.77 0.75

23.41 11.65 9.45 4.26 4.88 14.71 0.50 0.76 1.09 4.20 0.44 0.73 0.92 0.62 0.44 1.04 1.13 0.21 7.93 1.09 0.68 0.44

11.00 5.00 4.00 2.00 9.00 18.00 0.00 1.OO 1.oo 1.oo 0.00 0.00 1.00 2.00 0.00 1.oo 0.00 0.00 0.00 2.00 1.OO 0.00

110.00 50.00 40.00 20.00 27.00 82.00 1.00 4.00 5.00 15.00 1.oo 2.00 4.00 4.00 1.oo 5.00 3.00 1.OO 45.00 6.00 4.00 1.oo

crime) lies within and between communities. The estimates of variance components in each level work as criteria in evaluating the results of subsequent models. These models have no predictors at either level. That is, the level-1 or person-level model characterizes the outcome (Y,, of individual i in community j with just an intercept (Po,, which is the community mean of the outcome, and the random effect (r,,. A t the level-2 or community-level, each community’s mean outcome (Po,) is represented as a function of the grand mean (yoo plus a random error (uo, associated with community j . Table 2 reports the results from the ANOVA models for perceived incivilities, perceived risk, overall fear of crime, and fear of personal and property crime. The estimates for the within-community or level-1 variance [i.e., Var (r,, = .”I in the outcomes are .884, .903, .973, .978, and .970,

1186

LEE AND ULMER

respectively. The overall variability among the community means [i.e., ] .116, .129, .043, .040, and .041, respectively. These values, Var (poi = T ~ ) is along with the intraclass correlation (IC r = T,, / T~ + a’), indicate that about 12% of the variation in perceived incivilities and 13% of the variation in perceived risk exists between communities. However, only about 4% of the variation in overall fear, personal fear, and property fear exists between communities. In other words, most of the variance in perceived incivilities, risk, and each fear outcome is at the individual, not the community, level. Next, we test whether the estimated values of between-community variance, T ~ are , significantly greater than zero (H”: T,,,, = 0), using a large sample chi-square distribution with J - 1 degrees of freedom. In Table 2, the test statistic for perceived incivilities is 162, with 62 degrees of freedom (J = 63 communities). We thus reject the null hypothesis (p < .001); significant variation to be explained exists between community means of perceived incivilities. The null hypotheses for perceived risk, overall fear, and fear of personal and property crime are also rejected, but the betweencommunity variances are very small for the fear variables. The overall reliabilities of the sample means (Po, of the outcomes in any community j are .474, .490, .268, .260, and .261, respectively. The overall measure of reliability is the average of the community reliabilities (the formula of the reliability of the sample mean in any community j is A, = reliability (Y.,)= TOO/ [ T W + (a’ln,)]). T H E INDIVIDUAL-LEVEL MODEL The random-effects-ANCOVA models in Table 3 show the effects of individual-level predictors on the outcome variables. In the model for perceived incivilities (the first column in the table), the effects of income, indirect victimization, and general perception of the severity of crime in the United States are significant, and the directions of associations are the same as those in previous studies for the general U.S. population. For example, Korean Americans who have a high level of household income are less likely to perceive incivilities in their communities because they tend to reside in more affluent neighborhoods. Knowledge of others’ victimization appears to increase the perception of incivilities. Korean Americans who have family members or Korean friends who have been victimized may interpret the same environmental clues more threatening than others do. Perceptions of the general severity of U.S. crime are also positively associated with perceived incivilities. Such an effect may be especially amplified when subjects compare the severity of crime in America with the relatively less severe crime of Korea. In the model for perceived risk (the second column), the influences of gender (female), health (defensibility), victimization experience, indirect

FEAR OF CRIME AMONG KOREAN AMERICANS 1187

Table 2. Results from Unconditional Models (One-way ANOVAa)b Fixed Effect' Between Communities Grand Mean, yoo Perceived incivility Perceived risk Fear of crime Personal crime fear Property crime fear Random Effect

Coefficient

S.E.

-.004

,062 .065 .050 ,050 .050

-.018 -.035 -.031 -.036 Perceived Perceived Incivility Risk

Fear of Crime

Personal Crime

Property Crime

Level 2: Between Communities, %

dflX2/P Level 1: Within Communities,

,116 ,129 ,043 .041 .041 6211621.OOO 62/149/.000 621951.005 621941.005 621931.007

rll ,884 Intraclass Correlation ,116 Reliability, Po, .474 'ANOVA = Analysis of variance. bAll outcome variables are standardized. 'All fixed effects were tested by t-tests.

,903 ,125 ,490

,973 ,042 ,268

.978 .040

,260

.970 ,041 .261

victimization, perceived crime severity, evaluation of the likelihood of racial riots, and perceived incivility are all significant. That females perceive less risk is somewhat unexpected. However, evidence exists to demonstrate that women are more likely to restrict their activities to decrease the degree of exposure to potential crime risk. For instance, in his survey of Seattle residents, Warr (1985) found that 40% of women avoided going out at night, whereas 9% of men did so (see also Balkin, 1979; Garofalo, 1981; Gordon and Riger, 1989; Stafford and Galle, 1984). Further, a high level of self-evaluated defensibility against an attacker is associated with decreased perceived risk, and this is congruent with our expectations. The positive influences of victimization experience, vicarious victimization, and perceived crime severity on perceived risk can be understood in the same context as their relationships to perceived incivilities. As we hypothesized, Korean Americans' evaluation of the potentiality of racial riots in the Chicago area, as a measure of racial conflicts, greatly increases perceptions of risk. In the ANCOVA model for perceived risk, the variance-explained statistic is .185 for within communities and .403 for between communities, and these are equivalent to a 21% reduction from the total residual variation in perceived risk.

1188

LEE AND ULMER

Table 3. Results from the Person-Level Models (ANCOVA" with Random Effects)b Fixed Effect Between Communities Grand mean, yw Within Communities Age, y10 Gender (Female), yz0 Health (Defensibility), y30 Education, ya Income. ys0 Victimization, ym Indirect Victimization, y7" Crime on TV,yRm Crime Severity, yw Self-Employed, ylm English Proficiency, yllo American Friend, yI2,, Natural Citizenship, ylia Length of Residence, yla Ethnic Media, y15" Future Riot, y,% Fear of Black, y,," Perceived Incivility. yI8" Perceived Risk, ylal Random Effects Level 2: Between Communities,

Perceived Incivility ,007

Perceived

Risk ,001

,003 -.062 ,074 ,005

-003 -.203** -.155** ,028

-.025*

.003

,141 .119* -.016

.218*** -.006 -.042 ,029 -.189 -.006 -.040 ,036 -.120

Perceived Incivilitv

Personal Fear of Crime Crime Fear

.391*** .156** .042 .199*** -042 -.033 ,049 ,254 -.011 ,054 .339*** -.057 .022**

Perceived Risk

-.010 .014*** .384*** -.245*** ,057 .011 ,001 ,007 -.029

.164*** ,002 -.142*** -.030 -277 -.021*** .132*** .134** .264*** ,010 .029***

-.007 .013***

.358*** -.247** *

.063* .016* ,017 ,003 -.052

.zoo*** ,016 -.146*** -.035 -.642*** -.020*** .117*** .137** .219*** ,003 .020**

Personal Fear of Crime Crime Fear

Property Crime Fear -.013 .013*** ,283"; -.197*** .045

,005 ,052 ,025 -.013 .139** ,004 -.171*** -.028 -.080 -.015** .112*** .131** .259*** .014* .024**

Property Crime Fear

,065

,077

.002

.003

,002

dflx'lp 62/121/.000 Level 1: Within Communities, ,868

62/125/.000 .736

62/68/.282 ,590

621731.167 .579

62/65/.360 ,677

,403

,953 ,394 ,417

.929 ,408 ,429

.950 ,302 ,328

&I

1'1

Proportion of Variance Explained: Level 2 ( R Z 2 ) Level 1 (R,') Total ( R f \

,440 .018 ,067

,185

,212

'ANCOVA = Analysis of covariance. bAll outcome variables are standardized. All level-1 predictors are centered around their grand means. * p 5 .05. **p 5 .Ol. ***p 5 ,001.

The third column of Table 3 presents the individual-level model of overall fear of crime. In this model, the variance-explained statistics are .394 within communities and .953 between communities. This means that the predictors explain about 42% of the total variation in fear of crime. The findings indicate the importance of perceived incivility and perceived risk-the two key intermediate processes in the risk interpretation

FEAR OF CRIME AMONG KOREAN AMERICANS 1189 model-in predicting fear of crime among these Korean Americans. Perceived incivility has significant direct and positive relationships with perceived risk and fear of property crime. Perceived risk, in turn, significantly effects all three types of crime fear. A pair of findings that has been consistently supported throughout the fear literature-that females and the elderly are more afraid of crime and that health (defensibility) dampens fear of crime-turned out to be applicable to these Korean Americans as well. Korean Americans who perceive greater crime severity in American society tend to be more afraid of crime. Further, perceived risk, as a reaction distinct from fear, appeared to be a very important predictor of each type of crime fear, just as the risk interpretation model predicts. Acculturation measures (English proficiency and length of residence in the United States) are strongly negatively associated with fear of crime. For example, Korean Americans who speak less English are more likely to be afraid of crime victimization than are those who are more fluent. As Poole and Pogrebin (1990) point out, language barriers can frustrate Korean Americans when they encounter crime victimization. Frustrated by language barriers, they often give up on calling police and often believe they will not be understood and will not receive an appropriate response or assistance. These kinds of language-related situations may amplify fear of crime among Korean Americans who speak little English. Also, Korean Americans with shorter lengths of residence in the United States tend to be more afraid of crime. Just-immigrated Korean Americans may compare the severity of crime in American society with that in Korea. Because they have not yet become immune to unfamiliar and perhaps uncomfortable social environments they encounter in everyday life in the United States, they may be more likely to interpret such environmental clues as risky. The effects of friendship with non-Korean Americans and natural citizenship, the other indicators of acculturation, are not significant. Second, Korean Americans who prefer Korean ethnic media (our indicator of ethnic attachment) are more likely to be afraid of crime victimization. Finally, the two variables measuring interracial conflicts and prejudices-evaluation of the potentiality of racial riots and prejudice against blacks-positively influence fear of crime. Fear of crime victimization is much higher among Korean Americans who assess the likelihood of violent conflicts among racial groups highly and who perceive blacks as a potential threat in terms of crime. In terms of the associations between racial conflicts and fear of crime, an especially interesting group may be Korean Americans who manage small businesses in minority neighborhoods. There were no meaningful differences in fear between self-employed persons and other subjects

1190

LEE AND ULMER

(including those who are not employed outside of the home). However, when we compared the fear levels of self-employed respondents with those who are employed by others (omitting those not employed outside of the home), a modest but significant difference emerged (p < .05, analysis available on request). Thus, self-employed Korean Americans (the majority of whom own small businesses) are more afraid of crime victimization than are those who are employed by others. To the extent that those self-employed here are small business owners, this suggests that the Korean-American small business owners in this sample are relatively more afraid of crime. The results for personal and property crime fear differ from those for overall fear only modestly. Of the three fear models, the personal fear model has the best explained variance at the individual level (.41) and the property fear model has the least (.30). The only notable difference between the overall fear model and the property fear model is that perceived incivility shows a small but significant positive effect on fear of property crime. The results for fear of personal crime are also very similar to those for overall fear. Income and education show modest but significant positive influences on personal crime fear. More notably, U.S. nativity exhibits a strong negative influence: Korean Americans born in the United States in this sample tend to be considerably less afraid of personal crime than are those born abroad. This finding is congruent with our expectations regarding the potentially different effects of acculturation versus ethnic attachment on fear of crime. THE FULL MODEL We now examine the common contextual effects of community environment on the outcomes with the random-intercept models with level-1 covariates. In Table 4,the individual-level models are the same as those in Table 3, but the level-2 models for community means (i.e., poi of the outcomes include a set of community characteristics as predictors. In Table 4, each model includes four ecological factors, selected from our preliminary analyses. However, the full models for the fear of crime variables are not superior to the person-level models in terms of explained-variance statistics and model parsimony. None of the level-2 predictors of fear of crime was significant because there was no significant between-community variance to explain after controlling for the effects of person-level predictors (p > .05 in Table 3). In other words, individual-level factors almost entirely accounted for the level-2 variation in overall fear of crime, as well as fear of personal and property crime. The contextual effects on fear of crime may be indirect through intermediate processes (i.e., perceived risk, perceived incivilities), as the risk interpretation approach suggests (LaGrange et al., 1992). Although we cannot

FEAR OF CRIME AMONG KOREAN AMERICANS 1191

Table 4. Results from the Full Models (Random-Intercept Model)" Fixed Effect Between Communities Grand Mean, ya, Crime Rate Density Youth Concentration House Value Within Communities Age. Y I O Gender (Female), yz0 Health (Defensibility), y w Education, y4 Income, ym Victimization, ym Indirect Victimization, yT0 Crime on TV, yso Crime Severity, yw Self-Employed, ylw English Proficiency, yl10 American Friend, ylz0 Natural Citizenship, yIM Length of Residence, y l a Ethnic Media, ylJO Future Riot, ylm Fear of Black, ylT0 Perceived Incivility, . ylso . Perceived Risk, ylw Random Effects Between Communities, UO, dflX*/P Within Communities, ry Proportion of Variance Explained: Level 2 (R?) Level 1 (R,') Total (R:)

Perceived Incivility

,053 .226** .143** ,046 -.062 ,002 -.085

.074 ,012 -.023* ,111 .112* -.014 .236*** ,017 -.053 ,024 -.147 -.003 -.018 ,048 -.119

Perceived Incivility

Perceived Risk

Personal Fear of Crime Crime Fear

,024 ,052

.114* .139** -.045 -.004 -.203** -.150** ,027 ,007 .376*** .155** ,043 .213*** -.018 -.041 ,045 ,269 -.008

,063 .360*** -.056 .018* Perceived Risk

,003 ,018 -.047 .021

,007

-.OM

,010 -.050 ,053 -.040

.015*** .391*** -.244*** ,054 ,011 ,009 ,007 -.030 ,160" -.001 -.137** -.031 -.272 -.022* * * .129*** .133** .263*** ,011 .029***

.014*** .366*** -.246*** ,059 .016* ,025 ,003 -.053 .198*** ,017 -.142*** -.035 -.639*** -.020*** .116*** .141** .217** ,002 ,019'

Personal Fear of Crime Crime Fear

Property Crime Fear

,011 ,057 -.054 ,014 -.067 .013*** .292*** -. 195* * * .045

.005 ,062 ,026 -.014 .138* ,001 -.165*** -.030 -.069 -.015* .109** .130* .260*** .013 .024** Property Crime Fear

.006 58/71/.121 ,867

,022 58/81/.025 .739

58/65/.236 ,589

,007 58/70/.138 .575

,003 58/60/.393 .676

,948 .019 ,127

A29 ,182 ,263

.907 395 .416

,829 ,412 ,429

,927 303 ,329

,004

'All of the outcome variables and level-2 predictors were standardized. All of the level-1 predictors were centered around their grand means. * p 5 .05. **p 5 -01.***p 5 .001.

fully test the indirect effects of the community characteristics,lz results are certainly congruent with this notion in that the community characteristics

12. HLM cannot assess indirect effects between perceived incivilities, perceived risk, and other variables. HLM cannot test a multistep causal model having several outcome variables simultaneously, nor does it provide overall goodness-of-fit statistics for the whole model.

1192

LEE AND ULMER

significantly affect perceived incivilities and perceived risk, which in turn are strongly related to fear of crime. In the full models, controlling for community characteristics did not produce remarkable changes in the effects of the individual-level predictors. This is because, in the random-intercept models in Table 4, the level-2 predictors explain only the level-2 variance, and almost none of the individual-level variance in incivilities and risk. Table 4 shows significant positive effects of crime rate and population density on community mean perceived incivility. Controlling for crime rates and population density, the residual between-community variance becomes almost zero (.006). Population density and youth concentration also show significant positive effects on perceived risk in the full model. These ecological factors explain about 43% of between-community variance in perceived risk, and they add about 5% to the total explained variance of perceived risk, compared with the model in Table 3. The chi-square test of significance of residual between-community variance indicates that there is still significant residual variance between communities in perceived risk (p < .05), but its actual amount is very small (.022). Housing values, Yo Black, Yo Hispanic, % Korean, poverty, single-headed households, and suburban neighborhood status did not significantly predict perceived risk, perceived incivilities, or fear of crime (results available on request). In the foregoing analyses, we classified the person-level factors into several subsets that measured different aspects of a risk interpretation model adapted for Korean Americans. However, the goodness-of-fit statistics (i.e., explained-variance) in Tables 2-4 do not allow us to determine whether the joint effect of a subset of predictors is statistically significant. Table 5 therefore reports multiparameter hypothesis tests for the fixed effects in the full model for overall fear of crime. We omit personal and property fear because the joint subset effects were substantially the same as for overall fear. All of the results, except for two victimization variables and self-employment, are significant. The influences of the acculturation versus ethnic cohesion and racial conflict and prejudice variables were especially remarkable. The significant effects of perceived risk and perceived incivilities (“perceptions” in Table 5 ) , as intermediate processes between the other level-1 predictors and fear of crime, support the risk interpretation model. Finally, acculturation, ethnic cohesion, racial conflict, and anti-black prejudice were important factors in explaining Korean Americans’ fear of crime, even though their influences on perceived incivilities and perceived risk were minimal.

FEAR OF CRIME AMONG KOREAN AMERICANS 1193

Table 5. Results of Multiparameter Hypothesis Tests for Fixed Effects in the Model (ANCOVA) for Fear of Crime Null Hypothesis (H,) Status Characteristics:

Chi-square

df

119.74 118.33 6.14

5 3

.ooo .ooo

2

,045

0.03

2

>.500

11.64

2

.003

19.47

2

.OW

0.01

1

>.500

112.05

5

.ooo

23.80

2

.ooo

= y20 = YU, = y40 = y50 = 0 Physical: ylo= y2,, = y30 = 0 Social: yd0 = yso = 0

yl0

p-Value

Victimization: y60

=

y70

=

0

Media Effects: yso

=

yw = 0

Perceptions: y1so = y1w = 0

Ethnicity and Racial Conflict: Self-Employment: yloo= 0 Acculturation vs. Ethnic Cohesion: y110 = y120 = y130 = y140 = y1so = 0 Racial Conjlict and Prejudice: y1M) = y170 = 0

CONCLUSION Drawing from Ferraro’s (1995) risk interpretation theory, we investigated influences on fear of crime among Korean Americans at both individual and community levels. We summarize our key findings as follows: (1) English proficiency was a strong predictor of fear of crimeKorean Americans who speak English well were remarkably less afraid of crime than were their less fluent counterparts. ( 2 ) The length of residence in the United States exhibited a negative relationship to fear of crime among these Korean Americans. In addition, native-born U.S. citizens were markedly less fearful of personal crime. ( 3 ) Preference for Korean ethnic media was a substantial factor in explaining fear of crime. Korean Americans who watchedhead ethnic TV programs or newspapers more frequently were potentially more attached to their ethnic people and culture and, in turn, more afraid of crime. (4) Racial conflict and anti-black prejudice significantly influenced these Korean Americans’ fear of crime. Korean Americans who were concerned about the potentiality of violent racial conflicts appeared to perceive more victimization risk and tended to be

1194

LEE AND ULMER

more afraid of crime. Additionally, Korean Americans who harbored prejudices against African Americans as a potential threat were strikingly more afraid of victimization. ( 5 ) Self-employment and the number of close American friends were not significant predictors in explaining fear among these Korean Americans, nor were community-level variables, such as percent black, percent Hispanic, poverty, or single-headed households. Although the effects of victimization factors on perceptions of incivilities and risk are moderately supported by our data, a strong direct influence of victimization (personal or vicarious) on fear of crime, which has been widely recognized through the fear literature, was not found with this Korean sample. Korean Americans who have been victimized or know family members or other Korean acquaintances who have been victimized tend to perceive more incivilities and risk in their environments (and this may indirectly affect fear), but neither of these factors is significant in explaining fear of crime per se. A possible explanation may lie in Korean Americans’ ethnic characteristics and unique history of victimization. As discussed above, among Korean Americans, fear of crime may be part of a prevalent collective discourse.13 This suggests that in explaining Korean Americans’ fear, some group-level dynamics, such as interracial conflicts and collective perceptions, may be more important than is individual experience or knowledge. For instance, although Korean Americans’ perceptions of the probability of violent racial conflicts like the 1992 L.A. riots were used as an indicator of racial conflicts, it could also represent group-based vicarious victimization. The present study has some limitations. Most obviously, the data represent a nonprobability sample, and this limits the statistical generalizability of these findings to a general Korean-American population. Second, small numbers of individuals per community prevented us from making optimal use of the merits of HLM, such as testing for cross-level interaction effects. For example, we could not permit the relationships (i.e,, level-1 slopes) between individual characteristics and the outcomes (e.g., fear of crime) to vary across communities, and we could not examine whether community characteristics affect the relationships between individual characteristics and the outcomes, either amplifying or attenuating

13. This assumption may be supported by comparing their average fear levels in this study to those in other studies (Ferraro, 1995; Warr, 1984, 1985, 1987; Warr and Stafford, 1983) that surveyed the general American population, using similar fear measures.

FEAR OF CRIME AMONG KOREAN AMERICANS 1195 them. Future studies that do so will obviously make significant contributions to our understanding of interrelationships between the social organization of communities and individual-level factors as they influence perceived risk, perceived incivilities, and fear of crime. Future research that addresses perceptions of crime among immigrant minority groups should focus on historical, economic, and cultural background characteristics of those groups (see also Ackah, 1992). Our study illustrates the importance of acculturation, ethnic cohesion, and dynamics of interracial conflicts in predicting fear of crime among ethnic communities in similar structural, cultural, and demographic circumstances to Korean Americans (i.e., relatively small, relatively recently immigrated, from home cultures that are more racially homogeneous). In particular, it would be interesting to replicate (and improve on) our study among such ethnic communities as Vietnamese Americans, Cambodian Americans, or Thai Americans. These communities may have similarities to Korean Americans in that they, too, are relatively small, are even more recent arrivals than Korean Americans, and come from relatively racially homogeneous home cultures. Such similarities might lead us to expect similarities in the effects of acculturation, ethnic cohesion, or racial prejudice on fear of crime. On the other hand, these groups also differ in important ways (historically, culturally, socioeconomically) from Korean Americans, and these differences would suggest differences in the dynamics of fear of crime. For example, to our knowledge, Vietnamese Americans, Cambodian Americans, Thai Americans, or others have not experienced similar tensions and strained relations with the African-American community as Korean Americans have. Thus, anti-black prejudice and hostility might not play the same role in fear of crime among these groups as it does among the Korean Americans we have studied. These are interesting and important empirical questions for future research. We further suggest that fear of crime should be incorporated as a key emphasis in group threat theories of racial prejudice and hostility (e.g., Blumer, 1958). That is, fear of crime at the hands of another racial group seems to us to constitute a very emotional, personalized, and visceral type of perceived group threat. This kind of threat is, in turn, potentially important in explaining racial prejudice more broadly. As Taylor’s (1998) study of anti-black prejudice suggests, fear of crime as a perceived group threat has been largely unexplored in the literature on racial prejudice and hostility. This literature has focused mostly on the role of perceived economic or political group threats (e.g., the role of fear of blacks taking jobs from whites as a factor in white anti-black prejudice, or fear of Korean small businesses in black neighborhoods displacing black-owned businesses as a factor in black anti-Korean prejudice; see Jo, 1992). However, this same literature has largely ignored fear of crime (see Taylor, 1998).

1196

LEE AND ULMER

We have contributed to group threat theory by extending its application beyond the context of black-white relations, just as Blumer (1958) intended. We also suggest that future research empirically examine the role of fear of crime in the relationship between perceived group threat and racial prejudice involving any group (including whites, blacks, and Hispanics). Although our study focuses on Korean Americans, we speculate that fear of crime is likely to be tightly intertwined with racial and ethnic stereotypes, attributions, and tensions much more broadly (see also Altheide and Michalowski, 1999). It is worth recalling the Thomas and Thomas (1928) theorem: What people define as real becomes real in its consequences. Fear of crime and perceived group threats are individual and collective definitions of situations. To the extent that such fears, senses oi threat, and prejudices are widespread, they poison interracial relations between all kinds of racial communities (Korean Americans and blacks, blacks and whites, whites and Hispanics, etc.), and they potentially undermine efforts toward a more harmonious multicultural society. Ic may be that debunking unsupported race-specific fears of crime could be an important antidote to racial prejudice. Likewise, debunking racial prejudice could be an important antidote for overwrought fears of interracial crime.

REFERENCES Ackah, Samuel An empirical investigation of the correlates of fear of crime among immi1992 grants in the Washington D.C. metropolitan area: A case study of Ghanaians. Ph.D. dissertation, George Washington University. Ahne, Joseph Koreans of Chicago: The New Entrepreneurial Immigrants. In Melvin G. 1995 Holli and Peter d’A. Jones (eds.), Ethnic Chicago: A Multicultural Portrait. Grand Rapids, Mich.: William B. Eerdmans. Altheide, David and Raymond S . Michalowski 1999 Fear in the news: A discourse of control. The Sociological Quarterly 40:475-503. Balkin, Steven Victimization rates, safety and fear of crime. Social Problems 26:343-358. 1979 Blumer, Herbert Race prejudice as a sense of group position. Pacific Sociological Review 1958 1:3-7. Symbolic Interactionism: Perspective and Method. Englewood Cliffs, N.J.: 1969 Prentice-Hall. Bobo, Lawrence and Vincent Hutchings 1996 Perceptions of racial group competition: Extending Blumer’s theory of group position to a multiracial social context. American Sociological Review 61:951-972.

FEAR OF CRIME AMONG KOREAN AMERICANS 1197 Braungart, Margaret M., Richard G. Braungart, and William J. Hoyer Age, sex, and social factors in fear of crime. Sociological Focus 1355-66. 1980 Bryk, Anthony S. and Stephen W. Raudenbush Hierarchical Linear Models: Applications and Data Analysis Methods. 1992 Newbury Park, Calif.: Sage. Bursik, Robert J., Jr. and Harold G. Grasmick Neighborhoods and Crime. New York: Lexington Books. 1993 Cheng, Lucie and Yen Espiritu Korean businesses in Black and Hispanic neighborhoods: A study of inter1989 group relations. Sociological Perspectives 32521434. Clemente, Frank and Michael B. Kleiman Fear of crime in the United States: A multivariate analysis. Social Forces 1977 56:519-531. Dempster, A. P., N. M. Laird, and D. B. Rubin 1977 Maximum likelihood from incomplete data via the EM algorithm. Journal of the Royal Statistical Theory, Series B 39:l-8. Ferraro, Kenneth F. Fear of Crime: Interpreting Victimization Risk. New York: State Univer1995 sity of New York Press. Ferraro, Kenneth F. and Randy L. LaGrange The measurement of fear of crime. Sociological Inquiry 57:70-101. 1987 Garofalo, James The fear of crime: Causes and consequences. Journal of Criminal Law and 1981 Criminology 72:839-857. Gordon, Margaret T. and Stephanie Riger The Female Fear. New York: The Free Press. 1989 Hale, Chris 1996 Fear of crime: A review of the literature. International Review of Victimology 4:79-150. Hurh, Won Moo and Kwang Chung Kim Korean Immigrants in America: A Structural Analysis of Ethnic Confine1984 ment and Adhesive Adaptation. London: Associated University Presses. Uprooting and Adjustment: A Sociological Study of Korean Immigrants’ 1988 Mental Health. Final Report Submitted to National Institute of Mental Health. Religious participation of Korean immigrants in the United States. Journal 1990 of the Scientific Study of Religion 1919-34. Jo, Moon H. 1992 Korean merchants in the Black community: Prejudice among the victims of prejudice. Ethnic and Racial Studies E395-411. Kahle, Robert Wayne 1994 The multiple dimensions of fear of crime, neighborhood cohesion and protective weapon ownership. Ph.D. dissertation, University of Michigan. Kim, Kwang Chung and Won Moo Hurh 1993 Beyond assimilation and pluralism: Syncretic sociocultural adaptation of Korean Immigrants in the US. Ethnic and Racial Studies 16:696-713.

1198

LEE AND ULMER

Kim-Goh, Mikyong, Chong Suh, Dudley David Blake, and Bruce Hiley-Young 1995 Psychological impact of the Los Angeles riots on Korean-American victims: Implications for treatment. American Journal of Orthopsychiatry 65: 138-146. Korean National Bureau of Statistics 1987 Korean General Population Statistics. Seoul, Republic of Korea (South Korea): Korean National Bureau of Statistics. LaGrange, Randy L., Kenneth F. Ferraro, and Michael Supancic Perceived risk and fear of crime: Role of social and physical incivilities. 1992 Journal of Research in Crime and Delinquency 29:311-334. Light, Ivan and Edna Bonacich 1988 Immigrant Entrepreneurs: Koreans in Los Angeles, 1965-1982. Berkeley, Calif.: University of California Press. Lindquist, John H. and Janice M. Duke The elderly victim at risk: Explaining the fear-victimization paradox. Crim1982 inology 20:115-126. Longford, Nicholas T. A fast scoring algorithm for maximum likelihood estimation in unbalanced 1987 mixed models with nested random effects. Biometrika 74:817-827. Mangiafico, Luciano 1988 Contemporary Asian Immigrants: Patterns of Filipino, Korean, and Chinese Settlement in the United States. New York: Praeger. Min. Pyong Gap 1984 From white-collar occupations to small business: Korean immigrants’ occupational adjustment. Sociological Quarterly 25:333-352. 1988 Ethnic Business Enterprise: Korean Small Business in Atlanta. Staten Island, N.Y.: Center for Migration Studies. 1989 Some Positive Effects of Ethnic Business for an Immigrant Community: Koreans in Los Angeles. Final Report Submitted to the National Science Foundation. The structure and social functions of Korean immigrant churches in the 1992 United States. International Migration Review 26:1370-1394. 1996 Caught in the Middle: Korean Merchants in America’s Multiethnic Cities. Berkeley, Calif.: University of California Press. Min, Pyong Gap and Andrew Kolodny 1994 The middleman minority characteristics of Korean immigrants in the United States. Korea Journal of Population and Development 23:179-202. Park, I. S.. J. Fawcett, F. Arnold, and R. Gardner 1989 Koreans Immigrating to the United States: A Pre-Departure Analysis. Paper No. 114. Hawaii: Population Institute, East-West Center. Parker, Keith D. 1988 Black-white differences in perceptions of fear of crime. Journal of Social Psychology 128:487-494. Parker, Keith D. and Melvin C. Ray 1990 Fear of crime: An assessment of related factors. Sociological Spectrum 1029-40.

FEAR OF CRIME AMONG KOREAN AMERICANS 1199 Poole, Eric D. and Mark R. Pogrebin Crime and law enforcement policy in the Korean American community. 1990 Police Studies 1357-66. Quillian, Lincoln Prejudice as a response to a perceived group threat: Population composi1995 tion and anti-immigrant and racial prejudice in Europe. American Sociological Review 60586-611. Rohe, William M. and Raymond J. Burby 1988 Fear of crime in public housing. Environment and Behavior 20:700-720. Sampson, Robert and W. Byron Groves 1989 Community structure and crime: Testing social disorganization theory. American Journal of Sociology 94:774-802. Skogan, Wesley G. and Michael G. Maxfield 1981 Coping with Crime: Individual and Neighborhood Reactions. Beverly Hills, Calif.: Sage Publications. Smith, Susan J. Fear of crime: Beyond a geography of deviance. Progress in Human Geog1987 raphy 11:l-23. Stafford, Mark C. and Omer R. Galle Victimization rates, exposure to risk, and fear of crime. Criminology 1984 22: 173-185. Taylor, Marylee 1998 How white attitudes vary with the racial composition of local populations: Numbers count. American Sociological Review 63512-536. Thomas, William Isaac and Dorothy Swain Thomas 1928 The Child in America. New York: Knopf. Thompson, Carol Y., William B. Bankston, and Roberta L. St. Pierre Parity and disparity among three measures of fear of crime: A research 1992 note. Deviant Behavior 13373-389. Walker, Monica A. 1994 Measuring concern about crime: Some inter-racial comparisons. British Journal of Criminology 34:366-378. Warr, Mark Fear of victimization: Why are women and the elderly more afraid? Social 1984 Science Quarterly 65:681-702. Fear of rape among urban women. Social Problems 32:238-250. 1985 1987 Fear of victimization and sensitivity to risk. Journal of Quantitative Criminology 3:29-46. Warr, Mark and Mark Stafford 1983 Fear of victimization: A look at the proximate causes. Social Forces 61:1033-1043. Wilson, William J. 1996 When Work Disappears. Cambridge, Mass.: Harvard University Press.

1200

LEE AND ULMER

Min-Sik Lee received his Ph.D. in Sociology at Purdue University in 1998. He is a Senior Research Scientist in the Division of International Criminal Justice at the Korean Institute of Criminology in Seoul, South Korea. His research interests include the ecology of crime, crime victimization, criminal careers, sentencing, and hierarchical linear modeling methods. His other publications on victimization and fear of crime appear in Sociology of Crime, Law and Deviance and Korean Journal of Victimology. His e-mail address is [email protected]. Jeffery T. Ulmer is Associate Professor of Sociology and Crime, Law, and Justice at Penn State University. His interests include courts and sentencing, the relationship between sanctions and future deviance, theories of crime and deviance, and symbolic interactionism’s general contributions to sociological theory and methodology. His recent research projects include studies of community corrections and recidivism, fear of crime, sentencing practices under sentencing guidelines, and an exploratory study of crime and sentencing in southern Brazil. His recent articles have appeared in Social Problems, Criminology, The Sociological Quarterly, Theoretical Criminology, Justice Quarterly, Sociological Inquiry, Journal of Contemporary Ethnography, and Applied Behavioral Science Review. His book, Social Worlds of Sentencing, was published in 1997 by SUNY Press. He is also editor of Sociology of Crime, Law, and Deviance, a biannual series published by Elsevier Sciences. Direct correspondence to Ulmer at Penn State.

FEAR OF CRIME AMONG KOREAN AMERICANS 1201

Appendix 1. Description of Variables Variable

Description

Community-Level

Crime ratea

[7 index crimes (murder, robbery, aggravated assault, burglary, larceny-theft, motor vehicle theft, arson) / population] * 100,000 Population / km2 Density Youth concentration [youth (15 to 24 years old) population / total population] * 100 Mean value of specified owner-occupied housing House value units YOBlack [black population/total population] * 100 YOHispanic [Hispanic population/total population] * 100 YO Korean [Korean population/total population] * 100 Poverty YOof population living at or below poverty line Single-headed YOof families headed by a single parent households Suburban 1 if suburban community, 0 if Chicago metro community Person-Level

A. Status Characteristics: Age Gender (Female) Health (Defensibility)

Education

Income

Age in years Dummy variable (1 = female, 0 = male) Self-evaluated ability to defend oneself effectively against an attacker: 4 = definitely yes 3 = probably yes 2 = probably no 1 = definitely no Completed schooling: 5 = graduate school 4 = college 3 = some college 2 = high school 1 = under high school Total household income for the previous year: 1 = less than $10,000, 2 = $10,000-$14,999, 3 = $15,OO0-$19,999, . . ., 14 = $70,000-$74,999, 15 = $75,000 and over

1202

LEE AND ULMER

B. Victimization:

Victimization Indirect victimization

Experience of victimization during the past 3 years; dummy variable (1 = yes, 0 = no) This variable is composed of two question items: one for victimization of family members or relatives, the other for victimization of Korean acquaintances. Each question item has two response categories (1 = yes, 0 = no). The range of the variable is between 0 and 2.

C. Exposure to the Media Violence and General Perception of the Severity of Crime:

Crime on TV

Crime severity

Frequency of watching reports of crime or other programs related to crime on TV: 4 = very frequently, 3 = frequently, 2 = sometimes, 1= rarely or never General perception (as a media effect) of the severity of crime in the U.S.: 4 = very serious problem, 3 = serious problem, 2 = small problem, 1 = not a problem at all

D. Small Businesses: Self-employed

Dummy variable (1 = self-employed, 0 = all others)

E. Acculturation: English proficiency American friend Natural citizenship

Length of residence

Self-evaluated English-speaking ability: 5 = fluent, 4 = good, 3 = fair, 2 = poor, 1 = not at all Number of close non-Korean-American friends. Nativityhmmigration status: This variable has been measured by the question: Were you born in the U.S.? Dummy variable (1 = yes, 0 = no) Length (years) of residence in the US.: Less than 6 months was counted as 0.

F. Ethnic Cohesion: Ethnic media

Frequency of watchinglreading media (TV program and newspaper) in Korean language: For each kind of media: 3 = regularly, 2 = occasionally, 1 = not at all. The range of the variable is between 2 and 6.

FEAR OF CRIME AMONG KOREAN AMERICANS 1203 G. Racial Conflict and Prejudice:

Future riot

Fear of Blacks

Opinion on the potentiality of violent racial conflicts in Chicago area: 4 = very likely, 3 = somewhat likely, 2 = somewhat unlikely, 1 = very unlikely. Based on the question, “HOW likely do you think that a violent racial conflict like the 1992 L.A. riots will happen in the Chicago area in the near future?” Prejudice against blacks as a potential threat: Based on the responses to the question, “What ethnic group are you most afraid of in terms of crime? dummy variable (1 = blacks, 0 = others)

H. Perceptions: A composite variable consisted of the following nine items that represent social and physical incivilities in a neighborhood: (1) Trash and litter lying around your neighborhood (2) Neighborhood dogs running loose (3) Inconsiderate or disruptive neighbors (4) Graffiti on sidewalks and walls ( 5 ) Vacant houses and unkempt lots (6) Unsupervised youths hanging out on the street (7) Too much noise (8) People drunk or high on drugs in public (9) Abandoned cars or car parts lying around Each item has three response categories: 1 = not a problem, 2 = somewhat of a problem, 3 = very serious problem. The range of the variable is between 9 and 27. Alpha = .94. This variable is standardized when it is used as the outcome variable in a model.

Perceived incivility

Perceived risk

This variable is measured by the following two question items asking perceived risk of violent crimes and property crimes, respectively: (1) How would you rate the chance that any violent crime such as murder, rape, and assault will happen to you during the coming year? (2) How would you rate the chance that any property crime such as burglary, theft, and vandalism will happen to you during the coming year?

1204

LEE AND ULMER

Respondents are required to indicate their feelings on a scale of 1 to 10, where 1 means ‘it’s not likely at all’ and 10 means ‘it’s very likely.’ The range of the variable is between 2 and 20. This variable is standardized when it is used as the outcome variable in a model. aForcible rape statistics furnished by the state-level Uniform Crime Reporting Program administered by the Illinois Department of State Police were not in accordance with national UCR guidelines. Therefore, the figures are excluded from the total index crime rates. The data for suburban communities are collected from 1990 through 1996 Uniform Crime Reports, on the basis of latest availability. The data for 77 community areas in Chicago City are acquired from 1996 index crime data by 25 police districts, from the Chicago Police Department. Although the borders of those districts do not coincide with the community areas. they can be matched with minimal discrepancies, by breaking down community areas to census tracts. All of the population data used in the computation of crime rates come from the 1990 Census.

FEAR OF CRIME AMONG KOREAN AMERICANS 1205

Appendix 2. A General Two-Level Hierarchical Linear Model Person-Level Model: We denote the outcome (e.g., fear of crime) for individual i in community j as Yll. This outcome is represented as a function of individual characteristics, XQe,and a model error rll:

where we assume that the level-1 error, r,, is normally distributed with a for i = 1, ..., n, mean of zero and a constant variance u2 [i.e., rll - N(0, d)], individuals in community j . We refer to u2 as the person-level variance. The regression coefficient, Pql, q = 0, ..., Q, indicate how the outcome is distributed in community j as a function of the measured individual characteristics. The intercept, pol, is the mean outcome for community j . Community-Level Model: The effects for each community, captured in the set of Pql’s in (l), are presumed to vary across communities. This variation is in turn modeled in a set of Q + 1 level-2 equations (i.e., one equation for each of the regression coefficients in the level-l model). Each pql is conceived as an outcome variable that depends on a set of communitylevel variables. W,, and a unique community effect, uql:

P,

=

yqo

+

yq1

WI] +

yq2W21

+ ... + yqsqws, + uq],

(2)

where a unique set of W, (s = 1, ..., s,) may be specified for each p,. The yqs coefficients capture the influence of community-level variables, W,, on the within-community relationships represented by p,. We assume that the set of Q + 1 level-2 random effects is multivariate normally distributed. , Each uql has a mean of zero, some variance T~~ [i.e., uql - N ( 0 , T ~ ~ ) ]and , , any two random effects q and q’. We refer to T ~ , covariance T ~ ~between as the community-level variance. The present study uses the random-intercept models with level- 1 covariates as the full model. Such models fix the slopes of level-1 predictors, not allowing them to vary across communities. Also, we center the level-1 predictors around their respective grand mean, XQ.. (see Bryk and Raudenbush, 1992:18-20). The shape of such models is basically the same as that of the ANCOVA model, except that the community mean (Pol of the outcome is now predicted by community characteristics. Therefore, the model becomes at the person-level:

Y, = Poj + P I j XQ..)+ r,.

(Xljj

- XI..) + pZj(X2;j - Xz..) + ... + pQj(XQll(3)

1206

LEE AND ULMER

The community-level model is now modified like this:

Because each level-1 predictor is centered around its grand mean, Poi is the mean outcome for community j, adjusted for differences among communities in XQipThe intercept, yOO,is the grand mean of outcome, whereas yoo is the pooled within-community regression coefficient for level-1 predictors.