Community Correlates of Rural Youth Violence - NCJRS · applied to communities at all points on the rural-urban continuum. To determine whether the theories are widely applica- ...

U.S. Department of Justice

Office of Justice Programs

Office of Juvenile Justice and Delinquency Prevention

May 2003

Access OJJDP publications online at ojjdp.ncjrs.org

J. Robert Flores, Administrator

ethnic diversity. These factors, it shouldbe noted, are statistical correlates and notcauses of such violence; nor are they theonly correlates.

Research limited to large urban areasleaves out as much of the U.S. populationas it captures. According to the 1990 cen-sus (U.S. Department of Commerce, 1992),only 49 percent of the U.S. population livesin urbanized areas of 500,000 or more, 25percent lives in fully rural settings (i.e.,places with populations of no more than2,500), and another 12 percent lives intowns or cities of fewer than 50,000 popu-lation. (The remaining 14 percent lives inmidsized urban areas with populationsbetween 50,000 and 500,000.) Althoughoverall crime rates are higher in urbanthan in rural areas (Maguire and Pastore,1995), this difference is not as large as iswidely assumed, and crime rates in smalltowns and rural areas vary considerably.

Several researchers on crime have calledfor more focus on rural settings, whichhave unique crime problems (e.g., the theftof agricultural equipment and commodi-ties) (Smith and Huff, 1982; Swanson, 1981;Weisheit, Wells, and Falcone, 1995). Equallyimportant are the striking similarities thatexist between urban and rural areas. Forinstance, there are comparable crimetrends over time, and the relationship of

Community Correlates ofRural Youth Violence

D. Wayne Osgood and Jeff M. Chambers

Rates of crime and delinquency vary wide-ly across communities, and research goingback many decades provides a good un-derstanding of the nature, correlates, andprobable causes of these community dif-ferences. Unfortunately, previous studieshave been limited in an important way.Virtually all studies of communities andcrime are based on large urban areas,almost totally excluding nonmetropolitanareas—that is, rural areas and smallercities and towns. The findings in this Bul-letin help to fill some gaps in the researchby examining variations in rates of juvenileviolence across nonmetropolitan commu-nities in Florida, Georgia, Nebraska, andSouth Carolina.

Social disorganization is the primary theoryby which criminologists account for ratesof crime in urban communities. If this theo-ry also applies to rural settings, then whatis known about crime in urban areas canprovide a basis for developing programsthat address the problem of delinquency insmaller communities. The research pre-sented in this Bulletin indicates that theprinciples of social disorganization theoryhold up quite well in rural settings. As inurban areas, rates of juvenile violence areconsiderably higher in rural communitiesthat have a large percentage of childrenliving in single-parent households, a highrate of population turnover, and significant

A Message From OJJDPAlthough decades of research haveshed considerable light on the natureof the causes and correlates of juve-nile delinquency, for the most partthese studies have focused on youthcrime in large metropolitan settingsand generally overlooked delinquencyin rural towns and smaller cities.

Since only half the U.S. population(49 percent) lives in urban areas of500,000 or more, while a quarter (25percent) lives in rural areas of 2,500or fewer and 1 in 10 (12 percent)lives in towns or cities of 2,500 to50,000, this leaves a considerablegap in research coverage, with virtu-ally as many excluded as included.

This Bulletin addresses the lack ofknowledge of rural youth violence byapplying social disorganization theoryto community correlates of youth vio-lence in nonmetropolitan communitiesin Florida, Georgia, Nebraska, andSouth Carolina.

According to this theory, rates of youthviolence are considerably higher incommunities that have large percent-ages of children living in single-parenthouseholds, a high rate of popula-tion turnover, and significant ethnicdiversity—whether in rural or urbansettings. Of course, these are but afew of the myraid of variables thatmay enter into the equation.

The findings on youth violence in non-metropolitan communities reported inthese pages will help guide programdevelopment and future researchto better serve rural youth and theirfamilies.

2

Chicago, IL; New York, NY; and San Diego,CA), but only one such study has beenconducted in a smaller city—Racine, WI.Nonmetropolitan areas have been includedin some studies of communities and crime,but that research is of limited value for thepurposes of this Bulletin. Some of thosestudies were based on national sampleswith both urban and rural respondents,but they did not separately examine pat-terns for nonmetropolitan communities(Sampson, 1985; Sampson and Groves,1989). Other studies analyzed crime ordelinquency in rural communities, butthey were either very limited in scope(e.g., Arthur, 1991, was limited to 13 coun-ties; Petee and Kowalski, 1993, is only 3pages long) or concerned with differentissues (Wilkinson, 1984).

Extending SocialDisorganization TheoryCurrent versions of social disorganizationtheory assume that strong networks ofsocial relationships prevent crime anddelinquency (Kornhauser, 1978; Bursikand Grasmick, 1993; Sampson and Groves,1989). When most community or neigh-borhood members are acquainted and ongood terms with one another, a substan-tial portion of the adult population hasthe potential to influence each child. Thelarger the network of acquaintances, thegreater the community’s capacity for in-formal surveillance (because residents areeasily distinguished from outsiders), forsupervision (because acquaintances arewilling to intervene when children andjuveniles behave unacceptably), and forshaping children’s values and interests.According to the current theory, commu-nity characteristics such as poverty andethnic diversity lead to higher delinquen-cy rates because they interfere with com-munity members’ abilities to worktogether (see citations above).

Just as in urban areas, systems of relation-ships are relevant to crime and delinquen-cy in small towns and rural communities.The only aspect of the theory specific tourban areas is the explanation of whysocial disorganization arises in somegeographic locations and not in others.

Rural sociologists concerned with the dis-ruptive effects of rapid population growthprovide some evidence that the processesof social disorganization apply in rural set-tings. Freudenberg (1986), for example,argued that the “boomtown” phenome-non brings high rates of crime and otherunacceptable behaviors but does not

crime to important factors such as age,sex, and race of the perpetrator and vic-tim is nearly identical (Bachman, 1992;Laub 1983a, 1983b).

Laub (1983b) concluded that mostindividual-level theories of crime and delin-quency, developed in reference to urbansettings, are likely to apply to rural set-tings. The question of whether the relation-ship between community characteristicsand rates of crime and delinquency is thesame in both urban and nonurban settingsrequires additional study. The rural-urbandimension (i.e., whether a community isrural, urban, or somewhere in between) isitself an essential aspect of communities,and current theories of communities andcrime would be far more useful if theyapplied to communities at all points onthe rural-urban continuum. To determinewhether the theories are widely applica-ble, the authors conducted a county-levelanalysis of youth violence to test whethersocial disorganization theory (Shaw andMcKay, 1942) applies to nonmetropolitancommunities.

Social Disorganizationand Rural CommunitiesSocial disorganization is defined as an in-ability of community members to achieveshared values or to solve jointly experi-enced problems (Bursik, 1988). In recentdecades, the themes of social disorgan-ization theory have been more clearlyarticulated and extended by Kornhauser(1978), Bursik and Grasmick (1993), andSampson and Groves (1989). Shaw andMcKay traced social disorganization toconditions endemic to the urban areasthat were the only places the newly arriv-ing poor could afford to live, in particular,a high rate of turnover in the population(residential instability) and mixes of peo-ple from different cultural backgrounds(ethnic diversity). Shaw and McKay’s analy-ses relating delinquency rates to thesestructural characteristics established keyfacts about the community correlates ofcrime and delinquency, and their work remains useful today as a guide for effortsto address crime and delinquency at thecommunity level.

Both theoretical development and empiri-cal research in the study of community in-fluences on crime and delinquency havefocused on urban settings. For instance,studies of neighborhood differences incrime rates have been conducted in manyof the largest cities in the United States(including Baltimore, MD; Boston, MA;

produce alienation or mental health difficul-ties. Furthermore, he explained thesenegative effects by the same logic associal disorganization theory: rapid growthgreatly diminishes the proportion of peo-ple who know one another, which in turninterferes with surveillance and socializa-tion of the young (Freudenberg, 1986).

Community Correlates ofYouth Violence Outsidethe CitySocial disorganization theory specifies thatseveral variables—residential instability,ethnic diversity, family disruption, eco-nomic status, population size or density,and proximity to urban areas—influencea community’s capacity to develop andmaintain strong systems of social relation-ships. To test the theory’s applicabilityto nonmetropolitan settings, this Bulletinexamines the relationships between thesecommunity variables and rates of offend-ing because the same relationships pro-vide the core empirical support for thetheory in urban settings. This section dis-cusses the relevance of each factor todelinquency rates in the social disorgani-zation framework.

Residential instability. Based on researchin urban settings, the authors expectedthat rates of juvenile violence in ruralcommunities would increase as rates ofresidential instability increased. Whenthe population of an area is constantlychanging, the residents have fewer op-portunities to develop strong, personalties to one another and to participate incommunity organizations (Bursik, 1988).This assumption has been central toresearch on social disorganization sinceits inception. Massive population changeis also the essential independent variableunderlying the boomtown research onrural settings (Freudenberg, 1986).

Ethnic diversity. According to social dis-organization theory, it could be expectedthat, as in urban areas, rates of juvenileviolence would be higher in rural communi-ties with greater ethnic diversity. Accord-ing to Shaw and McKay (1942), ethnic di-versity interferes with communicationamong adults. Effective communication isless likely in the face of ethnic diversitybecause differences in customs and a lackof shared experiences may breed fear andmistrust (Sampson and Groves, 1989). It isimportant to distinguish this theoreticallydriven hypothesis about heterogeneityfrom simple ethnic differences in offenserates. In other words, this hypothesis sees

3

accountability to neighbors. In the leastdense rural areas, it may be social isola-tion, instead, that limits social support tomonitor children and respond to problembehavior. On the other hand, Sampson(1983) suggested that density might bemore important in terms of opportunitiesfor offending than in terms of social disor-ganization. The relative isolation of livingin a sparsely populated area may reduceopportunities for offending because ofgreater distance from targets and from po-tential companions in crime (Cohen andFelson, 1979; Osgood et al., 1996). This pos-sibility is supported by Laub’s (1983b:189)finding that victimization rates are lowestin communities with the smallest popula-tions, but only for populations of 25,000 orless. In larger communities, the rates wereessentially unrelated to population size.

Proximity to urban areas. This final com-munity variable, which departs from thethemes of current social disorganizationtheory, considers an issue specific to ruralsettings and to the linkages among commu-nities. As Heitgerd and Bursik (1987) haveargued, it is important to look beyond theinternal dynamics of communities and con-sider ways in which rates of delinquencymight be influenced by relationships be-tween neighboring communities. Variousrural and suburban communities have verydifferent relationships with urban commu-nities, and this is an important theme ofresearch on rural settings. Heitgerd andBursik suggested that “less delinquentgroups of youths are being socialized intomore sophisticated types of criminal be-havior by youths in adjoining areas” (1987:785). Because average crime rates are high-er in communities with larger populations,this phenomenon would produce higherrates of delinquency in rural communitiesthat are adjacent to metropolitan areas.Previous research has not addressed thistopic, however, so it is not clear whethersuch diffusion actually occurs and, if itdoes, whether it is strong enough to pro-duce higher rates of juvenile violence incounties adjacent to urban areas.

Methods

SampleThe sample consists of the nonmetropoli-tan counties in Florida, Georgia, Nebraska,and South Carolina.1 The standard unit ofanalysis for research in the urban settinghas been neighborhoods no more thana few miles across. This conception ofcommunity does not generalize very well

for rural settings, where population densi-ty is much lower. The county is a con-venient unit of analysis for the study ofcommunity influences on rural crime ratesbecause both arrest data, taken from theFederal Bureau of Investigation’s UniformCrime Reports (UCR), and population char-acteristics, from U.S. Bureau of the Censuspopulation reports, are available at thecounty level. The county is also a com-mon unit of analysis in rural research ofall types because counties typically havestrong internal economic and governmen-tal structures. It should not be forgotten,however, that most counties include sev-eral distinct communities. The countylevel of analysis was necessitated by theavailability of data, but it is not ideal.

The analysis was limited to counties notincluded in metropolitan statistical areasby the U.S. Bureau of the Census. Thesecounties lack cities with populations of50,000 or more, and less than 50 percentof their population resides in metropoli-tan areas with a population of 100,000 ormore. Thus, residents of these countieslive in smaller cities, towns, and opencountry rather than in moderate to largecities and their suburbs.

The study sample included 264 countieswith populations ranging from 560 to98,000. Although these nonmetropolitancounties are much larger geographic unitsthan areas analyzed in community-levelresearch on crime in metropolitan settings,they are of equal or smaller size in termsof population. The average total populationof these nonmetropolitan counties wasroughly 10,000, which is comparable tothe smallest units used in research on ur-ban neighborhoods (Sampson, Rauden-bush, and Earls, 1997; Warner and Pierce,1993). This sample compares favorablywith those in studies of urban areas interms of the number of communities, thesize of the populations, and the variety ofcommunities included.

MeasuresDelinquency. UCR data (Federal Bureauof Investigation, 1998) were used to meas-ure each county’s delinquency rate. Thesedata are the logical starting point for analy-ses of crime and delinquency in rural areas,and previous community-level studies ofrural crime have relied on the same source.No measure of crime or delinquency is per-fect, and criminologists have long beenconcerned about potential biases in crimerates based on official records, especially

crime as arising from relations betweenethnic groups, not from some groups beingmore crime-prone than others.

Family disruption. Research in urbanareas has found that delinquency rates arehigher in communities with greater levelsof family disruption, and the authors ex-pected that this also would be true inrural areas. Sampson (1985; Sampson andGroves, 1989) argued that unshared par-enting strains parents’ resources of time,money, and energy, which interferes withtheir ability to supervise their childrenand communicate with other adults in theneighborhood. Furthermore, the smallerthe number of parents in a community rel-ative to the number of children, the morelimited the networks of adult supervisionwill be for all the children.

Economic status. Although rates of juve-nile violence are higher in urban areas withlower economic status, it was not clearthat this relationship should apply in ruralsettings. The role of economic status insocial disorganization theory is based onpatterns of growth in urban areas. In manymajor urban areas, growth leads to thephysical, economic, and social decline ofthe residential areas closest to the centralbusiness district. These areas then becomemost readily available to the poor and togroups who migrate to the area. As a re-sult, areas with the lowest average socio-economic status will also have the greatestresidential instability and ethnic diversity,which in turn will create social disorgani-zation (Bursik and Grasmick, 1993). Ac-cordingly, many studies have found thaturban neighborhoods with high rates ofpoverty also have greater rates of delin-quency (Warner and Pierce, 1993).

The processes that link poverty with pop-ulation turnover are specific to urban set-tings. In nonmetropolitan settings, poorpopulations may be stable and ethnicallyhomogeneous.

Population density. Population density israther different from the other communityfactors for two reasons. First, evidence ofa relationship between population densityand urban crime and delinquency is incon-sistent. Second, the meaning of densitybecomes quite different for nonurban com-munities, where, in the least dense areas,one must travel several miles to have sig-nificant contact with people outside ofone’s immediate family. The original rea-soning for the urban context was that highpopulation density creates problems byproducing anonymity that interferes with

4

that arrests might reflect the behavior oflaw enforcement officers more than thebehavior of offenders. Fortunately, find-ings relating social disorganization to ar-rests have been replicated by more recentstudies that measured offending throughcitizen calls for police assistance (Warnerand Pierce, 1993), self-reports of victims(Sampson, 1985; Sampson and Groves,1989), and self-reports of offenders (Elliottet al., 1996).

This study’s measure of delinquency wasthe per capita arrest rate of juveniles ages11–17 in each county, pooled over the 5-year period from 1989 through 1993. Theoutcome measures were as follows: ratesof arrest for homicide, forcible rape,aggravated assault, robbery, weaponsoffense, simple assault, and the UCR Vio-lent Crime Index, which comprises thefirst four offenses. The study consideredthe full spectrum of violent offenses (cap-turing a large range of offense serious-ness) for which recording is comparableacross the four states. This approach pro-vided a rich pool of information for estab-lishing the consistency of the findings.

Table 1 presents descriptive statistics forall of the measures, calculated separatelyfor each state. Rates of arrest for seriousviolent offenses in nonmetropolitan coun-ties in Florida and South Carolina areconsiderably higher than in Georgia andNebraska. Differences are less consistent

for simple assaults. Some of these incon-sistencies, such as the extremely low rateof simple assault, compared to the ViolentCrime Index, in Florida, suggest that policeand citizens give less attention to minoroffenses in areas with high rates of seri-ous offenses (as noted by Stark, 1987).

Explanatory variables. Data from the1990 census provide measures for mostof the explanatory variables (U.S. Depart-ment of Commerce, 1992). As is standardin research on communities and crime,the measure of residential instability wasthe proportion of households occupiedby persons who had moved from anotherdwelling in the previous 5 years (Samp-son, 1985; Warner and Pierce, 1993). Eth-nic diversity was measured in terms of theproportion of households occupied bywhite versus nonwhite persons. Ethnicdiversity was computed as the index ofdiversity, which reflects the probabilitythat two randomly drawn individualswould differ in ethnicity (Blau, 1977).Family disruption was indexed by female-headed households, expressed as a pro-portion of all households with children.Low economic status was defined as theproportion of persons living below thepoverty level. Proximity to urban areaswas coded “1” for counties adjacent to ametropolitan statistical area and “0” forcounties nonadjacent, based on census

classifications (U.S. Government Account-ing Office, 1989).

Also included in the analysis was the num-ber of youth ages 10–17, which is thepopulation at risk for juvenile arrests.Population size serves as a proxy measurefor population density because the twovariables are so strongly correlated thatthey are effectively indistinguishable. Be-cause states may differ in their statutesand in the organization, funding, and poli-cies of their justice systems, it was impor-tant to make sure that differences amongstates were not confused with the contri-butions of the explanatory variables.Therefore, the analysis controls for differ-ences among states in arrest rates foreach offense.

Data AnalysisThe outcome of interest in this study isthe arrest rate, defined as the number ofarrests in a county divided by the size ofthe juvenile population. Standard statisti-cal methods of analyzing crime rates areinappropriate for these data because thepopulation sizes are small relative to thearrest rates, so only very crude estimatesof arrest rates are available for the coun-ties with the smallest populations. Thisproblem is resolved with a specialized statistical technique (negative binomialregression) that takes into account the

Table 1: Descriptive Statistics for Nonmetropolitan Counties

Florida Georgia Nebraska South Carolina

Measure Mean SD* Mean SD* Mean SD* Mean SD*

Population at risk 2,941 2,074 2,287 1,940 1,091 1,152 4,926 2,621Number of counties 31 – 116 – 87 – 30 –

Explanatory variablesResidential instability 0.47 0.05 0.41 0.06 0.36 0.06 0.35 0.06Ethnic diversity 0.28 0.10 0.37 0.15 0.03 0.04 0.45 0.06Female-headed households 0.18 0.04 0.22 0.07 0.09 0.04 0.24 0.04Poverty rate 0.16 0.04 0.19 0.05 0.12 0.04 0.19 0.06Adjacent to urban area 0.74 0.44 0.53 0.50 0.14 0.35 0.80 0.41

Annual arrest rate per 100,000 populationViolent Crime Index 360.0 350.1 127.1 114.6 27.6 44.7 246.4 144.5

Homicide 12.2 16.8 4.8 9.9 1.0 4.1 10.7 12.2Forcible rape 19.5 24.7 8.2 12.3 2.8 8.3 25.7 20.0Robbery 78.5 99.6 23.4 36.0 2.9 9.0 42.3 31.6Aggravated assault 249.9 237.6 89.5 83.4 20.9 36.1 167.7 106.2

Weapons offense 45.2 52.6 36.9 49.6 22.9 46.5 88.8 47.9Simple assault 169.9 200.1 159.7 163.8 182.4 318.5 343.9 342.0

* Standard deviation.

5

contribution of population size to theaccuracy of arrest rates.2

Tables 2 and 3 present two versions of therelationships of the explanatory variablesto delinquency rates. Table 2 considerseach explanatory variable separately, con-trolling only for overall differences amongthe states. Table 3 presents the second es-timate of each relationship, which controlsfor all other explanatory variables. Thefirst estimate reflects the overall associa-tion of the variable with the rate of juve-nile violence (the bivariate relationship),and the second estimate reflects only theassociation that cannot be accounted forby the other variables (the multivariaterelationship). Comparing tables 2 and 3,one can see that the patterns of results areessentially the same, with the magnitudeof the relationships typically somewhathigher for the bivariate relationships, and

somewhat fewer of the multivariate re-lationships reaching statistical significance.

Tables 2 and 3 express the relationships interms of the proportional change in therate of arrests associated with an increasein the variable.3 Most of the explanatoryvariables reflect proportions of the popu-lation, such as the proportion living inpoverty. The tables indicate the changein arrest rate associated with a 10-percentincrease in each explanatory variable. Forinstance, the first entry in table 2 indi-cates that the arrest rate of juveniles forviolent offenses will average 45 percenthigher (e.g., 145 versus 100 per 100,000)for counties with 25-percent residentialinstability than for counties with 15-percent residential instability.

Overall, the analysis found that oneor more of the social disorganization

variables were significantly associatedwith arrest rates for all of the violentoffenses except homicide. Low numbersof homicides limited the researchers’ abili-ty to detect differences in the homiciderates; 69 percent of the counties in thesample recorded no homicides duringthe 5-year study period.

Results

Residential Instability,Ethnic Diversity, and Family DisruptionIn research on social disorganization inurban settings, the three variables moststrongly and consistently associated withrates of crime and delinquency are resi-dential instability, ethnic diversity, andfamily disruption (see pages 2 and 3). In

Table 2: Relationship of Explanatory Variables to Juvenile Arrest Rates, Controlling for Overall Differences Among States

Proportional Difference in the Arrest Rate AssociatedWith a 10-Percent Increase in the Variable

Violent Forcible Aggravated Weapons SimpleVariable Crime Index Homicide Rape Robbery Assault Offense Assault

Residential instability 45%* –9% 40% 29% 50%* 51%* 65%*Ethnic diversity 23* 27 27* 35* 20* 25* 20Female-headed households 82* 33 85* 100* 75* 75* 73*Poverty rate 3 49 2 19 –2 –8 –31*Counties adjacent to metropolitan

areas (versus counties nonadjacent) 2 45 –6 –21 9 –10 10

Note: The states explored are Florida, Georgia, Nebraska, and South Carolina. The relationships were estimated using negative binomial regression.

* p < .05

Table 3: Relationship of Explanatory Variables to Juvenile Arrest Rates, Controlling for All Other ExplanatoryVariables and Differences Among States

Proportional Difference in the Arrest Rate AssociatedWith a 10-Percent Increase in the Variable

Violent Forcible Aggravated Weapons SimpleVariable Crime Index Homicide Rape Robbery Assault Offense Assault

Residential instability 33%* 3% 45% 2% 44%* 20% 40%*Ethnic diversity 18* 27 12 33* 12 23* 21*Female-headed households 70* –29 167* 45 89* 72* 88*Poverty rate –18 84 –48* 0 –25 –32 –39*Counties adjacent to metropolitan

areas (versus counties nonadjacent) –13 45 –17 –37* –3 –27 –8

Note: The states explored are Florida, Georgia, Nebraska, and South Carolina. The relationships were estimated using negative binomial regression.

* p < .05

6

the four states in this study, a similar pat-tern was found in nonmetropolitan coun-ties (see tables 2 and 3).

Social disorganization theory holds thatwhen turnover in the membership of acommunity is high, social relationshipswill weaken and juvenile violence will in-crease. Consistent with this theory, thestudy data showed that residential insta-bility was significantly associated withhigher rates of aggravated assault, simpleassault, weapons violations (bivariaterelationship only), and the overall ViolentCrime Index. This relationship was margin-ally significant for rape (p < .10 for bothestimates). The connection between resi-dential instability and delinquency appearsto be quite strong. In the bivariate associ-ations of table 2, a 10-percent increase inresidential instability was associated with29- to 65-percent higher rates of arrest forthe various forms of juvenile violence,with the exception of homicide.

Ethnic diversity is also a key variable be-cause cultural differences tend to interferewith adults’ ability to work together in su-pervising and raising their children. Thecorrelation between ethnic diversity andviolent offenses was statistically signifi-cant in most instances. In the bivariaterelationships, a 10-percent increase inethnic diversity was associated with 20- to 35-percent higher rates of juvenileviolence.

The reader may wonder whether the re-sults for ethnic diversity truly reflect diver-sity or if that variable is merely a proxy forthe proportion of minority group membersin the population. These variables are toohighly correlated to address this questiondirectly by including both in the samemodel. To gain some perspective on theissue, the authors estimated models thatreplaced ethnic diversity with the propor-tion of the population that was nonwhite.The nonwhite percentage was less strong-ly related to arrest rates, suggesting thatdiversity is the more important variable.

Higher levels of family disruption, as in-dexed by the proportion of female-headedhouseholds, also were strongly and con-sistently associated with higher rates ofarrest for violent offenses other than homi-cide. According to social disorganizationtheory, this pattern arises from the burdenof single parenting, which interferes withparents’ abilities to work together andreduces the number of adults involved inthe joint supervision of children. Therelationship between family disruptionand juvenile arrest rates was the strongest

in the study’s results. In the bivariate rela-tionships, this relationship was significantfor all offenses except homicide, and inthe multivariate relationships, it was sig-nificant for all offenses except homicideand robbery. In the bivariate relation-ships, a 10-percent increase in female-headed households was associated with73- to 100-percent higher rates of arrestfor all offenses except homicide.

In combination, residential instability, eth-nic diversity, and family disruption stronglydifferentiated counties with high rates ofarrest from those with low rates. Com-pare, for example, a county with 35-percentresidential instability, ethnic diversity of0.23 (on a scale of 0 to 0.5), and 13-percentfemale-headed households, which wouldbe a moderately low level of social disor-ganization, with one that has 45-percentresidential instability, ethnic diversity of0.33, and 23-percent female-headed house-holds, which would be a moderately highlevel. The multivariate relationships shownin table 3 (which control for all other ex-planatory variables) indicate that the ar-rest rate for the Violent Crime Index in themore disorganized county would be 2 2/3

times as great as that of the less disorgan-ized county (217 per 100,000 versus 81 per100,000).

Economic StatusThe analysis did not find a meaningful re-lationship between rates of delinquencyand rates of poverty.4 Instead of showingpoverty to be associated with higher ratesof delinquency, the relationships wereeither very slight or indicated an associa-tion between poverty and lower delinquen-cy rates (significantly lower rates for sim-ple assault and rape).

To understand this finding, it is useful toexamine the association between povertyrates and the other community correlatesof juvenile violence. As research in urbanareas has typically found (Warner andPierce, 1993), poverty rates in the study’snonmetropolitan counties were positive-ly associated with both ethnic diversity(r = .48, controlling for state) and the rateof female-headed households (r = .55).In contrast to urban areas, however, thecorrelation between poverty and residen-tial instability in these nonmetropolitanareas was negative rather than positive(r = –.39). This finding contradicts theclassic pattern of relationships from Parkand Burgess’s (1924) theory of urbanecology, which was the basis for predict-ing that poverty would lead to social

disorganization. Also in contrast to find-ings in urban areas, poverty rates werehigher in nonmetropolitan counties withsmaller populations than in those withlarger populations (r = –.41). Poverty ratesincrease as ethnic diversity and the pro-portion of female-headed households in-crease, suggesting that delinquency rateswill increase along with poverty rates. How-ever, this source of positive correlationbetween the rates of poverty and delin-quency is canceled out in nonmetropoli-tan areas, where rates of poverty arelower in areas with high residential insta-bility and larger populations.

This pattern of relationships is consistentwith research conducted by Fitchen (1994),who found that poorer residents do notmake frequent moves in rural areas. Low-cost housing is often abundant, and resi-dents have a support network of familyand friends who can provide casual rentagreements and flexible payment schemes.It appears that—unlike in most urbanareas—poverty does not disrupt the socialfabric of small towns and rural communi-ties. The reasons that a high rate of ruralpoverty does not increase the delinquen-cy rate appear to be consistent with socialdisorganization theory.

Population Size and DensityArrest rates for juvenile violence varieddramatically with differences in the sizes(and densities) of juvenile populations. Thefigure illustrates these findings with graphsfor four of the studied offenses. For all vio-lent offenses except homicide, differencesin the size of county juvenile populationscorresponded to differences of at leastthreefold in juvenile arrest rates. The fig-ure shows that annual arrest rates for ju-venile violence were uniformly lower inthe rural counties with the smallest popu-lations. Per capita arrest rates rose withincreases in juvenile population, but onlyuntil the population size reached about4,000. Beyond this level, increasing popu-lation had little impact on arrest rates forviolent offenses other than robbery.5 Theseresults are comparable to Laub’s (1983b)finding that victimization rates increasedwith population size for total populations(rather than juvenile populations) up toabout 25,000, but did not increase furtherfor larger populations. Arrest rates for theViolent Crime Index, rape, and aggravatedassault appeared to decline somewhat inthe upper range of juvenile populationsizes, but it is unlikely that these decreas-es are statistically reliable because theyare small.

7

Proximity to MetropolitanAreasWhether a rural county was adjacent to ametropolitan area had little bearing on itsrate of juvenile arrests for violent offens-es. None of the relationships for this ex-planatory variable approached statisticalsignificance. If delinquency can spreadfrom one community to another, the rea-son is not simple enough to be explainedby the county’s proximity to a metropoli-tan area.

ConclusionThe principles of social disorganizationtheory, developed in studies of urbanneighborhoods, can be applied to ruralcommunities. In the nonmetropolitan coun-ties that made up the study sample, percapita rates of juvenile arrest for violent

offenses were significantly and consistent-ly associated with residential instability,ethnic diversity, and family disruption.Based on the strength and consistency ofthe findings, family disruption, in particu-lar, appears to be a critical element of so-cial disorganization in nonmetropolitancommunities.

The study results diverged from the stand-ard findings for urban areas in that theyindicated no association between povertyand delinquency. When the correlates ofpoverty for this sample of nonmetropolitancommunities are considered, however, thisfinding is consistent with the core logicof social disorganization theory. Shaw andMcKay (1942) concluded that the relation-ship of poverty to delinquency in urbanareas is produced by the connection ofpoverty with the combination of residen-tial instability and ethnic diversity. This

urban population dynamic does not existin small towns and rural areas; outside thecity, the populations of poorer communi-ties are more stable than average, notless. Thus, these findings support Shawand McKay’s contention that it is notpoverty per se but an association ofpoverty with other factors that weakenssystems of social relationships in acommunity, thereby producing socialdisorganization.

Population Size and DensityThe findings concerning the relationshipof juvenile violence to the size and densityof the juvenile population have interestingimplications. Based on social disorganiza-tion theory, the authors hypothesized thathigh population density would interferewith social organization by creating ano-nymity and by increasing the difficulty of

Rape Robbery

Aggravated Assault Simple Assault

2,000 4,000 6,000 8,000 10,000

Population Ages 10–17

00

4

8

12

16

Arr

ests

(p

er 1

00,0

00 p

opul

atio

n)

Arr

ests

(p

er 1

00,0

00 p

opul

atio

n)

0

50

100

150

200

2,000 4,000 6,000 8,000 10,000

Population Ages 10–17

0

2,000 4,000 6,000 8,000 10,000

Population Ages 10–17

0

Florida NebraskaGeorgia South Carolina

0

50

100

150

200

Arr

ests

(p

er 1

00,0

00 p

opul

atio

n)

2,000 4,000 6,000 8,000 10,000

Population Ages 10–17

00

200

400

600

800

Arr

ests

(p

er 1

00,0

00 p

opul

atio

n)

Relationship of Population Size to Arrest Rates for Four Violent Offenses, Controlling for OtherExplanatory Variables

8

supervising children and adolescents. Thisreasoning implies that problems would in-tensify in areas with especially high popu-lation densities. The findings show theopposite: after reaching the modest den-sity of about 4,000 juveniles in an entirecounty, population size makes little differ-ence in the rate of juvenile violence. Clear-ly, another dynamic must be at work.

The relationship between population sizeand juvenile violence is more likely due toincreased opportunities for offending inareas with larger populations (Sampson,1983). A small population reduces thechances that a potential robber will ran-domly encounter a likely victim or thattwo rivals will meet in an unguarded set-ting conducive to an assault (Cohen andFelson, 1979). Furthermore, the companyof peers provides support for engaging indelinquent behavior (Osgood et al., 1996),and a very low population density makesit more difficult for peers to get together.

Consistency Across ViolentOffensesThe findings are consistent across the setof violent offenses. Many researchers limittheir analyses to a few offenses that theypresume to be most reliably recorded,such as homicide and robbery. Indeed,there can be little doubt that law enforce-ment officers have less discretion aboutwhether to make arrests for these offensesor that victims and bystanders are morelikely to report them. Nevertheless, the re-lationships of community characteristicsto the rate of simple assaults are nearlyidentical to those for the other violentoffense categories such as rape and aggra-vated assault. Thus, instead of findinginconsistent results for less serious offens-es, the data provided additional confirma-tion for the overall pattern of results.

Directions for FutureResearchThis study of juvenile violence in nonmet-ropolitan communities has successfullyextended research on communities andcrime beyond urban centers to small cit-ies and rural communities. The themesfrom social disorganization theory havea broader application to communities ofall sizes. Data from nonmetropolitan com-munities can be especially useful for test-ing and expanding social disorganizationtheory because they present differentpatterns of community variables. Forinstance, the findings related to poverty

and crime suggest that nonmetropolitancommunities may provide the setting inwhich the direct impact of poverty oncommunity disorganization can be deter-mined. Thus, social disorganization andrelated theories are appropriate startingpoints for developing either theories ofcrime specific to rural settings or theoriesof communities and crime that are generalacross settings. Developing such theorieswill require a firm grounding in the mod-ern realities of settings ranging from smallcities to isolated farming communities tothe suburbs that surround urban cores.For too long, theories of communities andcrime have limited their attention to animage of small, dense urban neighbor-hoods that fully encompass the lives oftheir inhabitants, and that image is out ofsync with life in most communities in theUnited States today.

For Further InformationFor more information on youth violence inrural communities, contact:

D. Wayne Osgood, Ph.D.Crime, Law, and Justice ProgramDepartment of SociologyPennsylvania State University1001 Oswald TowerUniversity Park, PA 16802–6207814–865–1304814–863–7216 (fax)[email protected] (e-mail)

Endnotes1. Many other states would be appropri-ate for this purpose. Florida, Georgia, Ne-braska, and South Carolina were chosenbecause the larger project through whichthis research was funded focused on thesoutheastern United States. Because ofregional variations in both crime and thestructural correlates specified in socialdisorganization theory, this study includesa midwestern plains state to assess, for asecond region, the generalizability of thefindings.

2. For a detailed discussion of these statis-tical problems and their resolution, seeOsgood, 2000.

3. This is a simple means of conveying theinformation in the regression coefficientsof the negative binomial analysis. Thatstatistical model assumes a logarithmicrelationship between the explanatoryvariable and the outcome, which impliesthat unit differences on the social disor-ganization variables are associated with

proportional differences on delinquencyrates.

4. Although the analysis included the un-employment rate as a second index of eco-nomic status, the results for this variablewere not very informative because therates varied so little within each state thatthe estimates were too imprecise to bemeaningful. Although unemployment wasassociated with higher rates of most ofthe offenses examined, none of those rela-tionships approached statistical signifi-cance (p > .35 in all cases). For a completepresentation of these analyses, see Os-good and Chambers, 2000.

5. This implies that the relationship wascurvilinear. In technical terms, the analy-ses allowed for curvilinearity by addingthe square and cube of population size asadditional terms in the regression model.There was significant evidence of a curvi-linear relationship for the Violent CrimeIndex, aggravated assault, and simple as-sault. For rape, the deviation from lineari-ty was of borderline significance, as wasthe overall relationship of population sizeto offending (p < .10).

ReferencesArthur, J.A. 1991. Socioeconomic predic-tors of crime in rural Georgia. Criminal Justice Review 16(1):29–41.

Bachman, R. 1992. Crime in nonmetro-politan America: A national accounting of trends, incidence rates, and idiosyn-cratic vulnerabilities. Rural Sociology57(4):546–560.

Blau, P.M. 1977. Inequality and Heterogene-ity: A Primitive Theory of Social Structure.New York, NY: Free Press.

Bursik, R.J., Jr. 1988. Social disorganizationand theory of crime and delinquency:Problems and prospects. Criminology26(4):519–551.

Bursik, R.J., Jr., and Grasmick, H.G. 1993.Neighborhoods and Crime: The Dimensionsof Effective Community Control. New York,NY: Lexington Books.

Cohen, L.E., and Felson, M. 1979. Socialchange and crime rate trends: A routineactivity approach. American SociologicalReview 44(4):588–608.

Elliott, D.S., Wilson, W.J., Huizinga, D.,Sampson, R.J., Elliott, A., and Rankin, B.1996. The effects of neighborhood dis-advantage on adolescent development.Journal of Research in Crime and Delin-quency 33(4):389–426.

9

Federal Bureau of Investigation. 1998.Crime in the United States, 1997: UniformCrime Reports. Washington, DC: U.S. Gov-ernment Printing Office.

Fitchen, J.M. 1994. Residential mobilityamong the rural poor. Rural Sociology59(3):417–436.

Freudenberg, W.R. 1986. The density ofacquaintanceship: An overlooked variablein community research. American Journalof Sociology 92(1):27–63.

Heitgerd, J.L., and Bursik, R.J., Jr. 1987.Extracommunity dynamics and the ecolo-gy of delinquency. American Journal ofSociology 92(4):775–787.

Kornhauser, R.R. 1978. Social Sources ofDelinquency. Chicago, IL: University ofChicago Press.

Laub, J.H. 1983a. Patterns of offending inurban and rural areas. Journal of CriminalJustice 11(2):129–142.

Laub, J.H. 1983b. Urbanism, race, andcrime. Journal of Research in Crime andDelinquency 20(2):183–198.

Maguire, K., and Pastore, A.L. 1995. Source-book of Criminal Justice Statistics 1994.Washington, DC: U.S. Department of Jus-tice, Office of Justice Programs, Bureau ofJustice Statistics.

Osgood, D.W. 2000. Poisson-based regres-sion analysis of aggregate crime rates.Journal of Quantitative Criminology16(1):21–43.

Osgood, D.W., and Chambers, J.M. 2000.Social disorganization outside the metrop-olis: An analysis of rural youth violence.Criminology 38(1):81–115.

Osgood, D.W., Wilson, J.K., Bachman, J.G.,O’Malley, P.M., and Johnston, L.D. 1996.Routine activities and individual deviant

behavior. American Sociological Review61(4):635–655.

Park, R.E., and Burgess, E.W. 1924. Intro-duction to the Science of Sociology. 2d ed.Chicago, IL: University of Chicago Press.

Petee, T.A., and Kowalski, G.S. 1993. Mod-eling rural violent crime rates: A test ofsocial disorganization theory. SociologicalFocus 26(1):87–89.

Sampson, R.J. 1983. Structural density and criminal victimization. Criminology21(2):276–293.

Sampson, R.J. 1985. Neighborhood andcrime: The structural determinants of per-sonal victimization. Journal of Research inCrime and Delinquency 22(1):7–40.

Sampson, R.J., and Groves, W.B. 1989.Community structure and crime: Testingsocial-disorganization theory. AmericanJournal of Sociology 94(4):774–802.

Sampson, R.J., Raudenbush, S.W., andEarls, F. 1997. Neighborhoods and violentcrime: A multilevel study of collectiveefficacy. Science 277(5328):918–924.

Shaw, C.R., and McKay, H.D. 1942. JuvenileDelinquency and Urban Areas. Chicago, IL:University of Chicago Press.

Smith, B.L., and Huff, C.R. 1982. Crime inthe country: The vulnerability and victim-ization of rural citizens. Journal of Crimi-nal Justice 10(4):271–282.

Stark, R. 1987. Deviant places: A theory ofthe ecology of crime. Criminology 25(4):893–909.

Swanson, C.R. 1981. Rural and agriculturecrime. Journal of Criminal Justice 9(1):19–27.

U.S. Department of Commerce. 1992. Sum-mary Tape Files 1 and 3, 1990 Census.Washington, DC: U.S. Government Print-ing Office.

U.S. Government Accounting Office. 1989.Rural Development: Federal Programs ThatFocus on Rural America and Its EconomicDevelopment. Washington, DC: U.S. Gov-ernment Accounting Office.

Warner, B.D., and Pierce, G.L. 1993. Re-examining social disorganization theoryusing calls to the police as a measure ofcrime. Criminology 31(4):493–517.

Weisheit, R.A., Wells, L.E., and Falcone,D.N. 1995. Crime and Policing in Rural andSmall-Town America: An Overview of theIssues. Washington, DC: U.S. Department ofJustice, Office of Justice Programs, Nation-al Institute of Justice.

Wilkinson, K.P. 1984. Rurality and patternsof social disruption. Rural Sociology 49(1):25–36.

The Office of Juvenile Justice and DelinquencyPrevention is a component of the Office ofJustice Programs, which also includes theBureau of Justice Assistance, the Bureau ofJustice Statistics, the National Institute ofJustice, and the Office for Victims of Crime.

This Bulletin was prepared under grant number94–JN–CX–0005 from the Office of JuvenileJustice and Delinquency Prevention, U.S. De-partment of Justice.

Points of view or opinions expressed in thisdocument are those of the authors and do notnecessarily represent the official position orpolicies of OJJDP or the U.S. Department ofJustice.

10

Related OJJDP Publications

If you are interested in more infor-mation about juvenile violence,please consult the recent OJJDPpublications listed below. Thesepublications are available in AdobeAcrobat and HTML online formatsand can be downloaded fromOJJDP’s Web site (ojjdp.ncjrs.org).Click on “Publications” to accessan alphabetical list of titles or tobrowse documents by topic area.You may also order copies of theseand other publications by callingthe Juvenile Justice Clearinghouseat 800–638–8736.

Animal Abuse and Youth Vio-lence (NCJ 188677). 2001. ThisOJJDP Bulletin describes psychi-atric, psychological, and criminalresearch linking animal abuse tointerpersonal violence perpetratedby juveniles and adults. Particularattention is focused on the preva-lence of cruelty to animals by chil-dren and adolescents and on therole of animal abuse as a possiblesymptom of conduct disorder. Inaddition, the motivations and etiolo-gy underlying the maltreatment ofanimals are thoroughly reviewed.The Bulletin includes recommen-dations to curb such cruelty andprovides contact information foradditional resources concernedwith violence perpetrated againstanimals and people. Available atojjdp.ncjrs.org/pubs/violvict.html#188677.

Gun Use by Male Juveniles:Research and Prevention (NCJ188992). 2001. This OJJDP Bulletinexamines patterns of gun owner-ship and gun carrying among ado-lescents, drawing on data fromOJJDP’s Rochester Youth Develop-ment Study. The Bulletin, whichis part of the Youth DevelopmentSeries, also addresses the inter-relationship between gangs andguns and describes preventionprograms, including the BostonGun Initiative, the Office of Com-munity Oriented Policing Services’Youth Firearms Violence Initiative,

and OJJDP’s Partnerships To ReduceJuvenile Gun Violence Program. Effec-tive efforts to reduce illegal gun carry-ing and gun violence among youthrequire the support and participationof multiple community agencies. Theinformation presented in this Bulletinis intended to enhance those efforts.Available at ojjdp.ncjrs.org/pubs/violvict.html#188992.

Juvenile Delinquency and SeriousInjury Victimization (NCJ 188676).2001. This OJJDP Bulletin draws ondata from two OJJDP longitudinalstudies on the causes and correlatesof juvenile delinquency—the DenverYouth Survey and the PittsburghYouth Study—to explore the interrela-tionship between delinquency andvictimization. The Bulletin, part of theYouth Development Series, focuseson victims of violence who sustainedserious injuries as a result of the vic-timization. Being victimized may leadto victimizing others. The studiesfound that many victims were proneto engage in illegal activities, associ-ate with delinquent peers, victimizeother delinquents, and avoid legalrecourse in resolving conflicts. Aclearer understanding of the patternsand predictors of victimization offersthe potential for increased effective-ness in designing and implementingstrategies to reduce both victimizationand offending. Available at ojjdp.ncjrs.org/pubs/violvict.html#188676.

Short- and Long-Term Conse-quences of Adolescent Victimi-zation (NCJ 191210). 2002. ThisBulletin is part of the Youth ViolenceResearch Bulletin Series, which isproduced jointly by OJJDP and theCenters for Disease Control and Pre-vention. It analyzes National YouthSurvey (NYS) data to explore howbeing a victim of crime during adoles-cence affects the likelihood of certainnegative outcomes in adulthood, in-cluding violent and property offendingand victimization, domestic violenceperpetration and victimization, druguse, and mental health problems. TheNYS data reflect the experiences of

approximately 1,700 respondents atages 11–17 and 21–29. Availableat www.ncjrs.org/pdffiles1/ojjdp/191210.pdf.

Violent Victimization as a RiskFactor for Violent OffendingAmong Juveniles (NCJ 195737).2002. This Bulletin analyzes therelationships between violent vic-timization and violent offendingamong juveniles across a 2-yearperiod, using data for 5,003 youthwho participated in the NationalLongitudinal Study of AdolescentHealth. It looks at victimization andoffending experiences in subgroupsof juveniles classified by age, gen-der, race, and level of physical de-velopment and also identifies riskand protective factors. The analysisdemonstrates that violent victimiza-tion is a warning signal for futureviolent offending and that victimiza-tion and offending share many ofthe same risk factors. The authorsdiscuss policy implications of theirfindings and suggest directions forfuture research. Available at www.ncjrs.org/html/ojjdp/jjbul2002_12_1/contents.html

Trends in Juvenile Violent Offend-ing: An Analysis of Victim SurveyData (NCJ 191052). 2002. ThisOJJDP Bulletin presents informa-tion on trends in juvenile violentoffending over the past two decades,based on data collected from thevictims of those offenses by theNational Crime Victimization Survey(NCVS). Unlike the data derivedfrom the FBI’s Uniform Crime Re-ports, which drive traditional assess-ments, the information provided byNCVS is not limited to cases thatcome to the attention of local lawenforcement officials. This Bulletinsuggests that examining informa-tion from a variety of sources abouta range of activities related to ju-venile offending will assist effortsto prevent and intervene in suchdelinquency. Available at www.ncjrs.org/html/ojjdp/jjbul2002_10_1/contents.html.

11

AcknowledgmentsD. Wayne Osgood, Ph.D., is Profes-sor in the Crime, Law, and JusticeProgram of Pennsylvania State Uni-versity’s Department of Sociology.Jeff M. Chambers, M.A., is ResearchManager in the Center on Children,Families, and the Law at the Univer-sity of Nebraska-Lincoln.

This Bulletin is an abbreviated ver-sion of an article that appeared inthe February 2000 issue of the jour-nal Criminology (Osgood and Cham-bers, 2000). The authors thank GaryMelton, Director of the Institute forFamilies and Neighborhood Life atClemson University, and Susan Lim-ber, Director of the Center for YouthParticipation and Human Rights atClemson University, for their supportof this project as Principal Investi-gator and Project Director, respec-tively, of the larger grant of which thisstudy was a part. Mike Overton ofthe Nebraska Commission on LawEnforcement and Criminal Justice,Dave Pfiefer of the Center for PublicAffairs Research at the University ofNebraska-Omaha, Mary Sik of theGeorgia Division of Demographic andStatistical Services, Mike Macfarlaneof the South Carolina Division of Re-search and Statistical Services, andSteven Kimble of the Florida StateData Center assisted the authors inobtaining arrest data.

It’s FastWant to know more about the issuesin this Bulletin or related information?Log on to ojjdp.ncjrs.org:

➤ Browse titles alphabetically orby topic.

➤ Discover the latest OJJDP releases.

➤ Subscribe to OJJDP’s listserv JUVJUST and the electronic newsletter JUSTINFO.

➤ Link to the NCJRS Abstracts Data-base to search for publicationsof interest.

It’s Easy

It’s Free

Share With Your ColleaguesUnless otherwise noted, OJJDP publications are not copyright protected. Weencourage you to reproduce this document, share it with your colleagues, andreprint it in your newsletter or journal. However, if you reprint, please cite OJJDPand the authors of this Bulletin. We are also interested in your feedback, such ashow you received a copy, how you intend to use the information, and how OJJDPmaterials meet your individual or agency needs. Please direct your comments andquestions to:

Juvenile Justice ClearinghousePublication Reprint/FeedbackP.O. Box 6000Rockville, MD 20849–6000800–638–8736301–519–5600 (fax)E-mail: [email protected]

NCJ 193591Bulletin

Electronic subscriptions are now available for OJJDP News @ aGlance, a bimonthly newsletter that presents up-to-date notices ofand quick access to agency activities, recent publications, fundingopportunities, and upcoming events.

To subscribe,

Go to OJJDP’s Web site (ojjdp.ncjrs.org).

Click on “Subscribe to OJJDP News @ a Glance.”

Fill in your name and e-mail address.

OrDownload individual issues fromojjdp.ncjrs.org/about/newsletter.html.

If you would like a copy mailed to you, contact the Juvenile Justice Clearinghouse by phone at 800–638–8736, by e-mail at [email protected],

or by mail at P.O. Box 6000, Rockville, MD 20849–6000.

*NCJ~193591*