Urban Structures and Crime

.SIAK-Journal – Zeitschrift für Polizeiwissenschaft und polizeiliche Praxis

Hermann, Dieter/Laue, Christian (2010):

SIAK-Journal − Zeitschrift für Polizeiwissenschaft und polizeiliche Praxis (2), 99-107.

doi: 10.7396/2010_2_J

Um auf diesen Artikel als Quelle zu verweisen, verwenden Sie bitte folgende Angaben:

Hermann, Dieter/Laue, Christian (2010). Urban Structures and Crime, SIAK-Journal − Zeitschrift für Polizeiwissenschaft und polizeiliche Praxis (2), 99-107, Online: http://dx.doi.org/10.7396/2010_2_J.

Hinweis: Die gedruckte Ausgabe des Artikels ist in der Print-Version des SIAK-Journals im Verlag NWV (http://nwv.at) erschienen.

Online publiziert: 3/2013 2/2010 .SIAK-JOURNAL

Urban Structures and Crime

Understanding the broken windows approach as an ecological theory fo cussing on the interrelationships of urban structures and crime, it can be summarized by the following hypotheses: (a) Worsening urban structural conditions lead to higher crime rates, higher fear of crime and to worse per ception of quality of life. (b) An increase of delinquency rates, fear of crime and perceived negative quality of life in a district leads to migration and a change in the structural conditions. In this study cross-sectional data, repre DIETER HERMANN, sentative surveys, and also longitudinal data about structural changes in Professor at the Institute German cities were used. The results largely confirm the broken windows of Criminology, University approach; however, a modification should be taken into consideration. A of Heidelberg. supplementing of this approach can be reached by a combination with lifestyle approaches. Following this, crime rates in a district not only depend on structural variables, but also on the lifestyle of the inhabitants.

1. INTRODUCTION be interpreted in several ways, making it The broken windows approach can be in necessary to specify the hypotheses of the terpreted as an ecological theory that des approach more exactly prior to testing it cribes the interrelationship between urban empirically. Taking into consideration both structures and crime (Wilson/Kelling the Wilson and Kelling study (Wilson/ 1982; Bottoms/Wiles 1997; Laue 1999; Kelling 1982), the city-sociological stu Hermann/Laue 2001). The concept has at dies of the Chicago School (Shaw/ CHRISTIAN LAUE, tained a relatively high status due to seve Zorbaugh et al. 1929; Shaw/McKay 1931; Assistant Professor at the ral applications in crime-prevention pro Burgess/Bogue 1964; Shaw/McKay 1969) Institute of Criminology, University of Heidelberg. jects and to its attractive theoretical which, theoretically speaking, can be con plausibility. The approach underlies com sidered to be the basis for the broken pletely different prevention strategies, windows approach, and Stark’s theory such as the police tactics in New York that (Stark 1987), the following hypotheses caused a worldwide sensation with the slo seem to summarize the approach: gan “Zero Tolerance” (Kelling and Coles • Different structural conditions in a 1996; Green 1999), or “community poli neighbourhood influence the potential cing” projects (Pate 1986) while other for formal control in this neighbour communal crime prevention measures also hood. The more intense these structural refer to Wilson and Kelling’s broken wind conditions are, for example, the hetero ows approach (Wilson/Kelling 19821). The geneity and the degree of individualiza great significance of the approach for tion of the inhabitants, as well as popu crime policy justifies the attempt to test it lation density, the less intense is the empirically – and this is the central topic potential for control. As a result, the de of this essay. linquency rate and fear of crime increase The many different applications of the and quality of life are perceived to be broken windows approach show that it can worse by the inhabitants.

99 .SIAK-JOURNAL 2/2010

• The higher the crime rate and the fear We can conclude in simplified terms of crime in a neighbourhood and the that structural conditions in a neighbour worse the perceived quality of life, the hood are the reasons for crime rate, fear of higher the degree of migration, this crime and quality of life. These characteri leading to a reduced population density stics are the reason for a change in the po and, finally, to a change in the pulation structure, which then leads to a VOCABULARY neighbourhood’s structural conditions. change in structural conditions by a feed

Quelle: Hermann/Laue back process. This process is shown gra to perceive wahrnehmen phically by Figure 1. Anomic structural conditions such as anomic anomisch (Anomie = Zustand, high population density, and above average in dem die Stabilität der sozialen heterogeneity and individualization of the Beziehungen gestört ist) population make it more difficult to create to refute social networks and social control. widerlegen Neighbourhoods where social control doesn’t function show a higher crime rate and fear of crime, and the quality of life is considered worse by the inhabitants. Thus, they become less and less attractive as li ving areas, resulting in negative populati on development, which is the preliminary stage of increasing criminality. These rela tions describe an ecological theory of crime which can be considered as one of the Figure 1: Concept of an Ecological Theory of Crime elements of the broken windows approach. in the Tradition of the Broken Windows Approach

Quelle: Hermann/Laue 2. EMPIRICAL STUDIES ON THE Study Design Result BROKEN WINDOWS APPROACH Zimbardo Experiment: Confirmation of the theory So far, relatively few empirical studies (1969) An old car is left in the for the Bronx, however, Bronx (New York City) and refutation for Palo Alto have been made on the question conducted in Palo Alto (California) to to test the validity of the broken windows see how passers-by react approach. Table 1 shows examples of some Shaw, Criminal-ecological studies: In most parts confirmation of studies. The criminal-ecological studies of McKay Characterization of the theory Shaw/McKay (Shaw/McKay 1969) and (1931) neighbourhoods with high crime rate by dilapidation, Skogan (Skogan 1990) largely confirm the poverty, high percentage of approach; Zimbardo’s experiment (Zim foreigners, lacking social bardo 1969), however, which has been ta control and decreasing number of inhabitants ken as proof for the validity of the broken Skogan Secondary analysis of Confirmation of the approach: windows approach by Wilson and Kelling (1990) empirical criminal-ecological Anomic conditions in (Wilson/Kelling 1982), at least partially studies neighbourhoods (disorder) refutes it. As a whole, the approach needs decrease the ability of the to be completed or modified. community to react appropriately to criminality, and increase crime rate

Table 1: Empirical Studies on the Broken Windows Approach

100 2/2010 . SIAK-JOURNAL

Quelle: Hermann/Laue 3. DATA AND Study Number Year Location Sample OPERATIONALIZATIONS of Cases This study focusses on explaining the me Victim 2.930 1998 Heidelberg Random sample of chanisms of the ecological approach, rather Survey persons Freiburg inhabitants of Heidelberg than examining the complexities of the and Freiburg between 14 and 70 years entire broken windows theory. We have Structural 40 districts 1998 Heidelberg Total sample of all districts analysed whether the conditions put for Data Freiburg in Heidelberg and ward in the broken windows approach can Freiburg be refuted by an empirical study, and have Longitudinal 191 1994– Frankfurt Total sample of all concentrated purely on the existing hypo Data districts 1998 Heidelberg districts in the cities Köln theses of an environmental theory of crime. München The hypotheses were first examined Rostock using cross-sectional data, that is structural data of two university cities and represen Table 2: Description of data tative surveys of their inhabitants. We then used longitudinal data to look at structural 4. RESULTS changes in neighbourhoods of German ci ties. The cross-sectional survey includes a 4.1. CROSS-SECTIONAL DATA questioning of victims so that information The statistical analysis was made with the about the dark figure could be taken into help of data characterizing neighbour consideration. Table 2 describes the data hoods. For this reason, the individual data VOCABULARY which are the basis of the analysis. used had to be aggregated at this level. cross-sectional The structural data were taken from pu The crime rate was measured by determi Querschnitts blished statistics of the Offices for City ning the percentage of victims in the to aggregate Development and City Planning of the se neighbourhoods.2 Levels of fear of crime kumulieren lected cities (Amt für Stadtentwicklung and quality of life are given as the average 1999). The survey data were taken from a value from the survey of those interviewed project undertaken by Dieter Dölling and in the neighbourhood. Dieter Hermann at the Institute of Crimi Methodologically speaking, these mea nology of the University of Heidelberg surements at neighbourhood level provide (Dölling/Hermann 1998; Hermann 1999; total surveys for two cities. Generalizati Hermann/Laue 2001). ons of the survey results beyond these The cross-sectional data are supposed to communities are only possible on the determine the influence of structural cha plausibility level. However it is possible to racteristics on delinquency, fear of crime test the hypotheses with the regionally li and perceived quality of life. In addition, mited data, since the hypotheses of ecolo these data explain changes in the structure gical theories of crime claim to be valid of the population. As the interrelation bet for all communities. The universality of ween structural conditions in a neighbour the hypotheses could thus be refuted with hood and migration cannot be examined this survey. with these data, longitudinal data were In a statistical analysis of 40 cases, the used for this purpose. number of variables with multivariate pro cesses must be relatively low. For this rea son, a path model3 has been developed for testing the hypotheses which only contains variables with relevant effects. In the first

101 .SIAK-JOURNAL 2/2010

step, characteristics were selected from numbers assigned to the paths are stan structural data on population density4, le dardized partial regression coefficients. vel of rents5, ethnic, sex- and age-specific This diagram shows that structural con segregation (percentage of foreigners, per ditions of a neighbourhood influence its centage of men, percentage of 10–17 year crime rate. The denser the population and VOCABULARY old persons), degree of individualization higher the degree of individualization, the (percentage of 1-person-households), and higher the crime rate. Other structural con bivariate bivariat, zweidimensional migration (development of the population ditions, especially the ethnic heterogeneity by immigration and migration) which, in of the population6, do not have an influence to fall victim Opfer werden bivariate analyses, are relevant to the on this characteristic. In other words, the crime rate. The significance level of the probability of an inhabitant of the violent crime Gewaltverbrechen correlation coefficients was used as a rele neighbourhood falling victim to a crime vance criterion. In the next step, we used will not change with an increasing percen to hold true gelten the relevant structural characteristics; le tage of foreigners in the neighbourhood, vel of crime, fear of crime, perceived qua provided that population density and fami lity of life and structural population ly structure of the inhabitants remain un changes to define a path model similar to changed. Eisner (Eisner 1997, 121) came the conception of an environmental theory to the same result in an analysis on violent of crime shown previously. We defined crimes registered by the police in 24 can any insignificant paths as irrelevant and tons of Switzerland. excluded them from our model. The result The level of fear of crime in a neighbour of this analysis is shown in Figure 2. The hood is influenced by local structural con ditions as well as the actual crime rate. Quelle: Hermann/Laue The higher the heterogeneity and the crime rate, the higher the fear of crime. These results show the rational and irratio nal components of fear of crime. The rela tion between crime rate and fear of crime is rationally comprehensible, but the oppo site holds true for the relation between percentage of foreigners and fear of crime, since the percentage of foreigners has no influence on the crime rate. This presuma bly contradicts the everyday theories of many people. The aggregated data used here prove a relation between criminality and fear of crime. Questions such as whether beco ming a victim changes the level of fear of crime and whether victims differ from non-victims concerning their fear of crime were mostly examined on an individual le vel and were answered differently (e.g. Legend: The relations are standardized path coefficients. Boers 1991; Kury/Würger 1993). In this

Figure 2: Testing of the Hypotheses of an Ecological study, however, aggregate data have been Theory of Crime by a Path Analysis used and this explains, at least partially,

102 2/2010 .SIAK-JOURNAL

the discrepancies between the results of that the development of the population re studies. Supposing that becoming a victim ally has an influence on the structural con within the family and among neighbours ditions, which can, however, only be pro do affect fear of crime, this mechanism is ven by a longitudinal study. not taken into consideration in studies The longitudinal data can be used to ex using individual data. In aggregate data, amine the feedback effect of migration in however, this kind of experience partially terms of whether an increased or decrea do count in measuring crime rates. sed population affects the structural condi The level of the perceived quality of life tions in this neighbourhood. The structural in a neighbourhood depends on the hetero conditions are measured using density of geneity and the degree of individualizati population7, degree of individualization8, on in a neighbourhood. The higher the eth ethnic segregation9 and migration by per nic heterogeneity and the lower the degree centage changes of the population. As the of individualization, the worse the percep se factors have been recorded over five tion of the quality of life. This characteri years, it is possible to see the temporal re stic, as well as the crime rate, influences lationship between cause and effect in the the development of the population. The statistical analysis. Only previous measu VOCABULARY higher the crime rate and the worse the rements of the cause variables were used to afford quality of life in a neighbourhood, the to determine the influence of these factors. sich leisten higher the percentage of people leaving In order not to distort the result through decline the neighbourhood. territorial reforms and changes to Abnahme, Rückgang; As a whole, the results of the empirical neighbourhood boundaries, neighbour Verschlechterung analysis confirm to a great extent the hy hoods with changes in the development of to distort potheses of the ecological theory of crime population of 10 % and more were exclu verzerren presented here. ded from the study. If the basic area of a neighbourhood re 4.2. LONGITUDINAL DATA mains the same, immigration or migration Ecological theories of crime like the bro will, of course, change the population den ken windows approach describe a long sity. The empirically relevant question is, term, dynamic development process. The however, whether migration does not just different factors interplay with one another change the density, but also the compositi with different effects. Thus, structural on of the population. If, in movements of conditions in a neighbourhood influence the population, certain groups are syste the fear of crime and perceived quality of matically over- or under-represented, thus life and therefore the development of the leading to a change in the structure of the population. If it is assumed – as in the population in a neighbourhood, this can broken windows approach – that only those eventually have an effect on criminality, persons leave a neighbourhood who can fear of crime and quality of life. afford to do so, which means that migrati The questions were tested using autore on is selective, the development of the po gressive models. This statistical procedure pulation influences the structure of the po tests whether there are dependencies bet pulation and thus the structural conditions ween interval-scaled variables. The depen in a neighbourhood. In the worst case, this dent variable is not only explained by inde development is negative to such an extent pendent variables, but also by previous that it results in a decline of the neighbour measurings of the dependent variable. The hood. This theory rests on the assumption analysis assumes an autoregressive process

103 .SIAK-JOURNAL 2/2010

of first order.10 An autoregressive model of of population, ethnic segregation or de the first order assumes that the level of a gree of individualization. The 75 %-per variable at time t depends on the level of centiles have been used as limit values for this variable at time of the previous year this selection, this meaning that all t-1. neighbourhoods that display the following The degree of order of the autoregressive characteristics have been taken into consi model can be determined using theoretical deration: considerations and statistical procedures. 1. a higher density of population, For the present application of the statisti 2. a higher percentage of foreigners and cal procedure it can be assumed that the 3. a higher percentage of one-person-house degree of individualization and the ethnic holds than 75 % of all neighbourhoods. segregation only depend on the level of Table 3 shows the results of autoregres this variable of the previous year but not of sions with regard to the influence of mi a time prior to that. The autocorrelation gration on the degree of individualization functions of the variables taken into consi and ethnic segregation. deration show a pattern that speaks for au When we look at all neighbourhoods, toregressive models of the first order.11 the influence of migration and immigrati The influence of migration on the de on on ethnic segregation and individuali gree of individualization and ethnic segre zation is not significant. However, if we gation was determined both for all 191 take into consideration only those neighbourhoods and for a selection of neighbourhoods with anomic structural neighbourhoods. According to the concept characteristics, migration affects ethnic of the ecological theory of crime, anomic segregation significantly but has no effect structural conditions lead to a decrease in on individualization. Thus, changes in po population, in turn aggravating the anomy. pulation do not usually change the structu The interrelationship between structure ral conditions of neighbourhoods. In and migration is thus said to lead to an es neighbourhoods with an anomic situation, calation of anomic structural conditions. however, migration leads to an increase in We therefore analysed neighbourhoods ethnic segregation. showing a well-developed escalation pro The influence of migration on the densi cess. For this reason, neighbourhoods ty of population was not determined by were selected with a relatively high density autoregression, and the relationship bet-

Quelle: Hermann/Laue The effect of Autoregressive models Selection of districts migration (M) on ... Ethnic segregation (E) E(t) = 0,99(s)=~E(t-1) – 0,09(ns)=~M(t-1) All districts (n=191) Individualization (I) I(t) = 1,0(s)=~I(t-1) – 0,11(ns)=~M(t-1) Ethnic segregation (E) E(t) = 0.98(s)=~E(t-1) – 0,59(s)=~M(t-1) Selection: (s)~ (ns)~ Individualization (I) I(t) = 1,0 =I(t-1) – 0,24 =M(t-1) D1 E 71,8 or D2 E 58,2 or D1: Density of population (inhabitants per hectare developed area) D2: Density of population (inhabitants per hectare area) E E 23,6 or M: Migration (percental change of population) M E 52,0 E: Ethnic segregation (percentage of foreign inhabitants) I: Individualization (percentage of one-person households) (n=95) (s): Significant (p#0,05) (ns): Not significant (p>0,05)

Table 3: Testing of Hypotheses of an Ecological Theory of Crime by Autoregressive Models

104 2/2010 . SIAK-JOURNAL

Quelle: Hermann/Laue ween these characteristics can be derived logically. Each change in population has an effect on the number of inhabitants and thus on the population density, if the area itself remains unchanged. In 90 % of all neighbourhoods, the building covered area changed by less than 0.3 % a year, so we can conclude that there is a close relati onship between migration and population density.

4.3. SUMMARY OF THE CROSS SECTIONAL AND LONGITUDINAL DATA ANALYSES Figure 3 shows a summary of the results of the cross-sectional and the longitudinal studies: A high population density leads to a relatively high crime rate, and a high de gree of ethnic segregation leads to a nega tive assessment of the quality of life. A high crime rate and a low quality of life Figure 3: Tested Concept of an Ecological Theory of lead to migration away from the Crime neighbourhood, this leading on the one hand to a decrease in the population densi gation; this, however, does not have an in ty, on the other hand to an increase in eth fluence on the crime rate. The negative nic segregation. While a reduction in po evaluation of a negative population deve pulation density leads to a lower crime lopment can not therefore be proven. rate and thus to an increase in population, In addition, the hypothesis of a crime-es an increase in ethnic segregation leads to a calating momentum of anomic structural negative evaluation of the quality of life, situations cannot be confirmed. However, and thus to a decrease in population. There an interrelationship between the structural fore decreasing numbers of inhabitants of situation in a neighbourhood, criminality, neighbourhoods do have both an anomy fear of crime, evaluation of the quality of reducing as well as an anomy-increasing life and migration can be confirmed. How effect that is an escalating as well as a ever, the relations seem to be more com deescalating effect. plex than assumed in the broken windows The results of the empirical analyses approach. partially refute the broken windows ap Finally, we could not confirm that ano proach. A negative development of the po mic structural conditions increase the le pulation is not the first prerequisite for a vels of crime although a clear interrelati rising crime rate, as put forward by the onship between the structure of the broken windows approach. First, a reduc neighbourhood, fear of crime, evaluation tion in population leads to a reduction in of the quality of life and migration has the population density and thus to decrea been determined. The relationship is, how sing crime rates. Second, decreasing num ever, more complex than originally assu bers of inhabitants influence ethnic segre med in the broken windows approach.

105 .SIAK-JOURNAL 2/2010

1 For the original essay, see internet via 11 The autocorrelation function ACFi is the relati http://www.theatlantic.com/election/ on between the time-depending characteristic connection/crime/windows.htm. X(t) and the time-delaying values of this charac 2 The following delicts have been taken into con teristic: ACFi = Corr(Xt, Xt-i). If this correlation sideration in the questions on victimization: theft is getting ever smaller with increasing temporal of a car, a motorcycle or a bicycle, theft at/out of interval of the correlated variables, this speaks the car, damage of the car, residence burglary, for the use of an autoregressive model of first or attempted residence burglary, damage to proper der (SPSS Trends 10.0, Supplement). This is the ty, theft, robbery, physical injury, threatening and case with all variables taken into consideration sexual nuisance. The question was whether the here. event had taken place during the past twelve months. The question regarding affective fear of Sources of information crime was formulated as follows: “How often do Amt für Stadtentwicklung und Statistik Heidel you think that you yourself can become the victim berg (1999). Statistische Datenblätter. Statisti of a crime?” and “Are you afraid to become the sche Informationen zu den Stadtteilen 1997 und victim of a crime at night, alone in your 1998. http://www.heidelberg.de/index2.htm. neighbourhood?”. The measurement of the sub Boers, K. (1991). Kriminalitätsfurcht, Pfaffen jective quality of life was made by the question weiler. on the estimation of the quality of life for the Bottoms, A. E./Wiles, P. (1997). Environmental neighbourhood of the interviewed with the help criminology, in: Maguire, M./Morgan, R./Reiner, of the school mark scale. R. (eds.) The Oxford handbook of criminology, 3 The path analysis is a statistical procedure able Oxford, 305–359. to determine the level of influence relations in Burgess, E. W./Bogue, D. J. (1964). The delin complex relations of variables. The procedure is quency research of Clifford R. Shaw and Henry frequently used to develop causal models or to D. McKay and associates, in: Burgess, E. W./ test their validity (Hermann 1984; Opp/Schmidt Bogue, D. J. (eds.) Contributions to urban socio 1976). logy, Chicago, 591–615. 4 Inhabitants per hectare of built-up area. Dölling, D./Hermann, D. (1998). Der Einfluß von 5 Supplements and reductions according to Wertorientierungen und Lebensstilen auf Viktimi neighbourhood following the Heidelberg 1998 sierungen, Kriminalitätsfurcht und selbstberich average rent system. tete Delinquenz. Forschungsantrag, Heidelberg. 6 The heterogeneity of the population is measured Eisner, M. (1997). Das Ende der zivilisierten by the percentage of foreigners in a neighbour Stadt? Die Auswirkungen von Modernisierung hood. und urbaner Krise auf Gewaltdelinquenz, Frank 7 For some cities tested, data for “inhabitants per furt/New York. hectare of built-up area” were available; for Green, J. (1999). Zero tolerance: A case study of others, the density of population has been opera police policies and practices in New York City, tionalized as “inhabitants per hectare of total Crime & Delinquency (45), 171–187. area of a neighbourhood”. Hermann, D. (1984). Ausgewählte Probleme bei 8 Percentage of single-person households. der Anwendung der Pfadanalyse, Frankfurt/Main. 9 Percentage of foreigners. Hermann, D. (1999). Kriminalität und Lebens 10 A stochastic process Xt is called “autoregressive qualität in Heidelberg und Freiburg. Untersu process of first order i” if Xt = a1Xt-1 + a2Xt-2 chungen und Vorschläge zur Kommunalen Krimi + .... + aiXt-i + et with et being an accidental nalprävention in den Stadtteilen Boxberg, mistake, a “white-noise-process” with expected Emmertsgrund und Kirchheim. Abschlussbericht, value zero (Schlittgen/Streitberg 1999). Heidelberg.

106 2/2010 .SIAK-JOURNAL

Hermann, D./Laue, C. (2001). Ökologie und Le Shaw, C. R./McKay, H. D. (1931). Social factors bensstil. Empirische Analysen zum „broken in juvenile delinquency, National Commission on windows“-Paradigma, in: Jehle, J.-M. (Hg.) Law Observance and Enforcement, Publication Raum und Kriminalität, Mönchengladbach, No. 132, Volume II. 89–120. Shaw, C. R./McKay, H. D. (1969). Juvenile delin Kelling, G./Coles, C. (1996). Fixing Broken quency and urban areas, Chicago. Windows, New York. Skogan, W. J. (1990). Disorder and decline. Kury, H./Würger, M. (1993). Opfererfahrung und Crime and the spiral of decay in American Kriminalitätsfurcht. Ein Beitrag zur Viktimisie neighborhoods, New York. rungsperspektive, in Kaiser, G./Kury, H. (Hg.) Stark, R. (1987). Deviant places: A theory of the Kriminologische Forschung in den 90er Jahren. ecology of crime, Criminology (25), 893–909. Beiträge aus dem Max-Planck-Institut für aus Wilson, J. Q./Kelling, G. L. (1982). Broken wind ländisches und internationales Strafrecht, Frei ows. The police and neighborhood safety, The At burg i. Br., 411–462. lantic Monthly (March 1982), 29–39. Laue, C. (1999). Anmerkungen zu Broken Wind Zimbardo, P. G. (1969). The human choice: Indi ows, Monatsschrift für Kriminologie und Straf viduation, reason and order versus deindividuati rechtsreform (82), 277–291. on, impulse and chaos, in: Arnold, W. J./Levine, Opp, K.-D./Schmidt, P. (1976). Einführung in die D. (eds.) Nebraska symposium on motivation, Mehrvariablenanalyse. Grundlagen der Formu Lincoln, 237–305. lierung und Prüfung komplexer sozialwissen schaftlicher Aussagen, Reinbek. Further Literature and Links Pate, A. M. (1986). Experimenting with foot pa Hermann, D./Dölling, D. (2001). Kriminalprä trol: The Newark experiment, in: Rosenbaum, D. vention und Wertorientierungen in komplexen P. (ed.) Community crime prevention. Does it Gesellschaften. Analysen zum Einfluss von Wer work?, Beverly Hills, 137–156. ten und Lebensstilen auf Delinquenz, Viktimisie Schlittgen, R./Streitberg, B. H. J. (1999). Zeitrei rungen und Kriminalitätsfurcht, Mainzer Schrif henanalyse, München. ten, Mainz. Shaw, C. R./Zorbaugh, F. M. et al. (1929). Delin quency areas, Chicago.

107