12
Using regression analysis of community areas of Chicago, patterns of Abstract homicide, robbery, and aggravated assault are analyzed for the mid-1970s. It is found that neighborhoods in which very poor and middle-class people live in close proximity are those in which rates of all three types of criminal violence are highest. Of all demographic and crime variables analyzed, proximity was by far the most strongly related to crime rates. This relation- ship was taken to be another indicator of the extreme burden placed on blacks in heavily segregated cities of the urban North. 0 COMMUNITY, ENVIRONMENT, AND VIOLENT CRIME RICHARD BLOCK Loyola University rban dwellers have known for many years, at least since the U Renaissance, that some neighborhoods are more dangerous than others. The newcomer to a city is often told which neigh- borhoods are safe, which are dangerous at night, and which should be avoided altogether. To many suburbanites and nonurban dwellers, the whole city is defined as dangerous. Although communities believed to be dangerous may not be overrun with crime, and although suburban communities may not always be crime-free, folk wisdom often does reflect real crime counts. Thus, it is not surprising that included among AUTHORS NOTE The research reported in the paper was completed at the Center for Studies in Criminal Justice of the University of Chicago Low School. This study was supported by PHS Research Grant No. IROIMH27575, NIMH (Center for Studies of Crime and Delinquency) and by grants from the Ford Foundation. It could not have been completed without the help and cooperation of the Chicago Police DepartmentS Superintendent James Rochford. Deputy Superintendent Michael Spiotto. ond Homicide Division Commander Joseph Di Leonardi. I would like ro thank Franklin Zimring for his constant help. patience. and advice. Thanks go to all my assistants on this project. especially Ruth (OBrien) Perrin ohd Nancy Hoverfield, and to Ronald Gilbert for his computer assistance. CRIMINOLOGY, Vol. 17 No. 1. May 1979 46-57 0 1979 American Society of Criminology 46

COMMUNITY, ENVIRONMENT, AND VIOLENT CRIME

Embed Size (px)

Citation preview

Using regression analysis of community areas of Chicago, patterns of Abstract homicide, robbery, and aggravated assault are analyzed for the mid-1970s.

It is found that neighborhoods in which very poor and middle-class people live in close proximity are those in which rates of all three types of criminal violence are highest. Of all demographic and crime variables analyzed, proximity was by far the most strongly related to crime rates. This relation- ship was taken to be another indicator of the extreme burden placed on blacks in heavily segregated cities of the urban North.

0 COMMUNITY, ENVIRONMENT, AND VIOLENT CRIME

RICHARD BLOCK Loyola University

rban dwellers have known for many years, a t least since the U Renaissance, that some neighborhoods are more dangerous than others. The newcomer to a city is often told which neigh- borhoods are safe, which are dangerous at night, and which should be avoided altogether. To many suburbanites and nonurban dwellers, the whole city is defined as dangerous. Although communities believed to be dangerous may not be overrun with crime, and although suburban communities may not always be crime-free, folk wisdom often does reflect real crime counts. Thus, it is not surprising that included among

A U T H O R S NOTE The research reported in the paper was completed at the Center for Studies in Criminal Justice of the University of Chicago Low School. This study was supported by PHS Research Grant No. IROIMH27575, NIMH (Center for Studies of Crime and Delinquency) and by grants from the Ford Foundation. It could not have been completed without the help and cooperation of the Chicago Police DepartmentS Superintendent James Rochford. Deputy Superintendent Michael Spiotto. ond Homicide Division Commander Joseph Di Leonardi. I would like ro thank Franklin Zimring for his constant help. patience. and advice. Thanks go to all my assistants on this project. especially Ruth (OBrien) Perrin ohd Nancy Hoverfield, and to Ronald Gilbert for his computer assistance.

CRIMINOLOGY, Vol. 17 No. 1. May 1979 46-57 0 1979 American Society of Criminology

46

Block / COMMUNITY, ENVIRONMENT, AND CRIME 47

the earliest research on criminal behavior were studies of the geographic distribution of urban crime (Levin and Lindesmith, 1937). This research predated even that of physical anthro- pologists, such as Lombroso. From this early start research has proceeded fitfully. In the United States the work of the Chicago School in the 1920s and 1930s was a good beginning and in many ways served as a basis for current research in the geogra- phic distribution of urban crime (Shaw and McKay, 1972). Between World War I1 and the beginning of large-scale fund- ing of criminological research in the mid-l960s, geographic research concentrated on census tract variation in crime and used factor analysis or multiple regression (Shevky and Bcll, 1955; Boggs, 1965; Ebert and Schwirian, 1968; Chilton, 1964).

From the mid-1960s to the present, both geographers and criminologists have developed increasingly sophisticated research methods for analyzing spatial distributions of crime. Among these are isoline analysis, models of spatial distance of victim and offender and crime site, and increasingly precise methods for differentiating and factoring areas of high and low crime (Curtis, 1975; Harries, 1974; Pyle, 1974; Capone, 1975; Reppetto, 1974).

The research presented here is a reflection of all these traditions of criminological analysis. Using data based upon records of the Chicago Police Department, community area variation in the distribution of violent crime and correlates of that variation are analyzed.

With the cooperation of the Chicago Police Department, the Center for Studies in Criminal Justice was able to obtain detailed and complete codings of all homicides in 1974 and large samples of robbery and aggravated assault in 1975. Coded information included a description of the crime, demographic information for victim and offender, and the crime’s location (Block, 1977). This location was later recoded into census tracts and aggregated into 76 community areas as uniformly defined by the early Chicago School and the 1970 census (Hunter, 1976). After coding was completed, informa-

48 CRIMINOLOGY / MAY 1979

tion was grouped by census tract and community area. Thus, for each area census data, crime rate, and crime characteristic data (e.g., percentage of robberies with guns) were available. The aggregated community area file forms the base of the analysis presented here. Community areas were chosen as units of analysis because of the inclusion of data on crime charac- teristics. Even with large samples, most census tracts had no- or only a few-violent crimes. These low crime tracts distorted analysis of the correlation of crime characteristics and crimes. Larger aggregation units reduced this problem.

The definition of most variables used in this analysis is straightforward. A few need explanation. All rates of crime are logged because of their nonlinear nature. Robberies and aggravated assault rates are projected for an entire year, and community areas in which no cases of a violent crime occurred were set to a value of 0. Although technically incorrect, this decision had little statistical effect. The variable “proximity” consists of the ratio of families earning more than three times poverty level to those earning 75% or less. This type of measure is usually taken as an indicator of income disparity. In terms of crime analysis, however, it can be more properly thought of as an indicator of spatial proximity of middle-class and poor.

The evidence and findings of this paper should be viewed with great caution. Inherent in the data and methodology are the following problems:

0 Some findings may be unique to Chicago or to Chicago in the mid-1970s.

0 Census data were collected in 1970, crime data in 1974 and 1975. The character of a few community areas may have changed.

0 The crime data are based upon police records. Crimes not reported to the police and events which were downgraded or not accepted as crimes by the police are excluded. This bias seems to be related to the seriousness of a crime. Police and victim survey data tend to converge for more serious crimes (Block, 1979).

Block / COMMUNITY, ENVIRONMENT, AND CRIME 49

0 This analysis is subject to both ecologically false interpretation and to overinterpretation due to the tendency for ecological correlations to be higher than individual correlations.

0 The interpretation of all findings are post hoc and should therefore be retested.

ANALYSIS OF DATA

Explanations of area variation in rates of violent crime are usually cultural, subcultural, or economic (Miller, 1958; Wolfgang and Ferracuti, 1967). Thus, high rates of violence in Houston are said to reflect a frontier tradition and a heritage of quick and easy justice (Lundsgarde, 1977). High rates of violence in the South are said to represent either a tradition of violence (Gastil, 1971) or the presence of vast rural poverty among both whites and blacks (Loftin and Hill, 1974).

In any study of violence in the United States race is always an important variable. Rates of both victimization and offense are consistently higher for blacks than for whites (Wolfgang, 1958; Brearley, 1932). These differences are explained by economics, by frustration and aggression resulting from the prejudiced and discriminatory nature of American society, or by southernness and the relative recency of black emigration to urban areas.

In Chicago, community areas which are predominantly black have higher rates of violence than those which are predominantly white. The 16 community areas with a 75% or larger black population have a logged homicide rate of 3.92 (5 1 per lO0,OOO) while the 47 community areas with fewer than 25% black population have a logged homicide rate of 1.77 (6 per 100,000). It is not surprising that virtually all victims in black communities are black and those in white communities are white. Chicago is an extremely segregated city.

Still, as seen in Table I, for Chicago’s community areas, racial composition is not the strongest zero-order correlate of crime

50 CRIMINOLOGY / MAY 1979

TABLE 1 Pearson Correlation Violent Crime Rates With

Demographic and Crime Characteristics of Community Areas

P r o ~ i m i t y Perc. Black Perc. South Perc. Old Perc. H.S. Educ Perc. 75% PO" Perc. 300% Pov Room Density Perc. 16-21 Unemp Perc. Fem. Head Perc. No. Fam Perc. S t a b l e Perc. Rob. Gun Perc. Rob. Out Perc. Rob. Mul. O f f . Perc. Rob. Resis Perc. Rob. I n j Perc. Rob. Teen O f f . Perc. H o m i . Rob Perc. H o m i . Unk Perc. H o m i . Gun Perc. H o m i . Out Perc. H o m i . Teen Perc. A s s . Unpro Perc. A s s . Uk Perc. Ass. Gun Perc. Ass. Dead Perc. Ass. Teen Perc. Ass. Out Lg. Nat. Homi Lg. Nat. Rob Lg. Nat. Ass

N a t u r a l Log Rate Per 100.000 Pop

Homi c i d r Robbery Aggravated A s s a u l t

~.

.7s** .61** .62**

.69** . s2** .47**

.14** . S 7 * ' . S8** - .Is** - . I2 -.16 -.6S** -. s4** - .60**

.64** .06** .48** -.67** - .48** - . 5s* * - .02 -. 24 -.13

.72** . 53* * . 5 2 * *

.63** .48** .49**

.29* .39** . 3 3 * -.47** -.so** -.40**

.03 - .06 -.22

.09 .18 .21 -.os -.04 . 0s - . 01 .02 .20 - .04 .os .26 - . 01

.22

.11

.36* -.01 - .06

. 2 s -.07

.27'

.01 - .28* - .01 1.00

.70**

.64**

-.20 . 2 1 . 2 s .36*

- . 2 1 - . I S

. 2 3 - .04

. 2 2

.06 -.16 -.04

.70 1.00

. 5 4 * *

-.11 . 2 2

- . 14 .25

-.12 - . 3 1 * -. 37** - .21

.45** - .02 - .39**

.16

.64**

.54"* 1.00

'p < .01 "p < ,001

rates. I Two variables are consistently more strongly related to violent crime rates-percentage of residents from the South and residential proximity of poor and middle-class families. Percent- age of residents who are black and percentage who are southern immigrants are very strongly correlated. Initially only one of these variables was included in the analysis. But the meaning of the two variabies is different in predominantly white and predominantly black neighborhoods. In predominantly black

Block / COMMUNITY, ENVIRONMENT, A N D CRIME 51

neighborhoods percentage southern is negatively correlated with poverty (-0.43), while in predominantly white neighbor- hoods the correlation is positive (poverty 0.76). In black neighborhoods southernness is related to stable residential patterns, while in white neighborhoods it is related to unstable patterns. The emigration of southern whites to Chicago is a more recent phenomenon than that of southern blacks. Thus, the differential meaning of southernness in white and black communities in Chicago probably accounts for the slightly stronger correlation of percentage southern born and crime rates than percentage black.

Residential proximity of poor and middle-class families is also strongly correlated with racial composition (0.62) and percentage southern born (0.67). Yet neither correlation is so strong as to entirely explain variation in residential proximity. And residential proximity is more strongly correlated to crime rates than either of the other two variables.

What is the meaning of close residential proximity of poor and middle-class families? Ulf Hannerz and others have noted that a particular problem for middle-class blacks is their close residential proximity to poorer community members (Hannerz, 1969). This close residential pattern mostly stems from discriminatory housing patterns and results in clashing and conflictual life styles. This economically based proximity conflict might result in higher crime rates in any community regardless of whether or not the neighborhood were pre- dominantly white or black. Crime rates are high and fear is great in communities in which the middle class has not created a buffer to keep out the poor (Silver, 1967).

A second explanation of the correlation of crime rates and residential proximity is that proximity may merely reflect the combined effects of neighborhood racial and income composi- tion. It is known that poor neighborhoods and communities which are predominantly black have higher crime rates than those which are predominantly white. A variable which combines both correlates of crime may be more strongly

52 CRIMINOLOGY / MAY 1979

related to crime rates than either variable alone. Both explana- tions of the strong correlation of residential proximity and violent crime rates are probably correct.

Given the high correlation of many of the independent variables under analysis, bivariate analysis is insufficient for understanding the relationship between community area and crime rate variables. In Table 2 a multiple regression analysis is presented for three logged crime rates. In this table additional variance explained by the inclusion of each variable and its standardized Beta weight is included for each independent variable. Each regression equation is limited to five steps.

It is clear that neighborhoods in which poor and middle- class families live in close proximity are likely to have higher crime rates than other neighborhoods. More than half the neighborhood variance in homicide rates and close to 40% of the variance in robbery and aggravated assault rates are explained by variation in residential proximity. The mean homicide rate for community areas above the median in proximity was 3.48; for those below the median it was 1.58.

Only percentage of residents southern born also enters into all three equations, and it explains relatively little of the variance in either robbery or aggravated assault rates. About 10% of the variance in robbery rates is accounted for by the percentage of homicides with unknown assailants. Individual level analysis indicates that most homicides with unknown assailants probably result from robberies. Thus, a high per- centage of homicides with unknown assailants is probably a result of a high robbery rate with death as a potential outcome of any act of violent crime. No other variable entered these equations either strongly or consistently.

Residential proximity of the poor and middle class so dominate the initial regression analysis that it was decided to replicate the multiple regression for community areas above and below the median for class proximity. Given the high percentage of variance already explained by proximity, it is not surprising that there is little variance left to explain. This is

Block / COMMUNITY, ENVIRONMENT, AND CRIME 53

TABLE 2 Multiple Regression Violent Crime with Community

Area Characteristics

Natural Log Homicide Rate Per 100.000 Pop- Variable R2 Chance Beta

Proximity Perc. South Perc. Stable Perc. H o m i . Uk. O f f . Perc. Rob Teen O f f .

Total R2

. 5 6

.10

.05

.02

.01

. 7 3

-42 - 4 2 .. 18 .18 .09

Natural Log Robbery Rate Per 100.000 Pop.

Proximity Perc. Homi. Uk. O f f . Perc. Homi. Teen O f f . Perc. South Perc. Homi. Outside

.37

.10

.ll

.04

.04

.41 - 5 6

-. 34 . 3 2 .22

Total R2 . 6 6

Natural Log Aggravated Assault Rate Per 100,000 Pop.

Proximity Perc. Rob. Inj . Perc. Ass. Unprov. Perc. South Perc. No. Family

. 39

.08

.06

.06 -04

Total R' - 6 3

30 32 28 34 21

especially true in high proximity neighborhoods. In these neighborhoods the mean logged homicide rate was 3.48 (32.5), and the standard deviation was 0.73. The mean logged homi- cide rate for low proximity neighborhoods was 1.58 (4.9) more than two standard deviations below that for high proximity neighborhoods.

Given the large difference in crime rates, it is not surprising that crime rate regression equations for high proximity neigh- borhoods are almost completely different from those for low proximity neighborhoods. For high proximity neighborhoods, percentage of residents southern born is strongly related to rates of robbery and homicides. This variation may reflect

54 CRIMINOLOGY / MAY 1979

both southernness and racial segregation in these neighbor- hoods.

Gun use during robberies is negatively related to both homicide and robbery rates. As the percentage of robberies using guns increases, the rate of violence declines. These results seem to contradict common sense. At an individual level both police data and victim survey data indicate a positive relationship between gun use and the probability of a robber’s success. Additionally, gun use in robbery is positively corre- lated with death while negatively correlated with injury.

The negative relationship between gun use and rates of robbery and homicide at a community level, however, may be explained by some of the correlates of gun use in robbery. Both at an individual and community level, gun use is posi- tively related to commercial robberies. Street crimes are less likely to involve gun use than commercial crimes. If neighborhoods with many street crimes are different from those with many commercial crimes, then the negative rela- tionship of gun use in robbery and rates of robbery may be explained. On the other hand, the correlation could be inter- preted to depict a dominance of fear in high-crime commu- nities. If there are fewer robberies in communities where the percentage of gun robberies is high, it may mean that fear of gun use has resulted in a reduction of crime or that robbery is so frequent that only armed robberies are reported to the police.

The range of variation of crime rates in low proximity neighborhoods is much larger than in high proximity commu- nities. Demographic variables related to violent crime rates are similar to those found in earlier research. And rates of violent crime seem to be dominated by robbery and pseudo- robbery indicators. Thus, rates of both robbery and homicide are positively correlated with youth unemployment, while robbery and aggravated assault rates are negatively correlated with family stability.

Block / COMMUNITY, ENVIRONMENT, AND CRIME 55

As previously noted, the percentage of homicides with unknown offenders is strongly related to the percentage of homicides which results from robberies. Thus, in low prox- imity neighborhoods, this variable is positively correlated with rates of robbery and homicide. However, percentage of homicide offenders unknown is negatively correlated with rates of aggravated assault. probably this reflects a high rate of robbery relative to assaults in some low proximity commu- nities.

Percentage of assaults resulting from unprovoked attacks enter the aggravated assault equation first. Communities with a high probability of random violence have higher overall rates of aggravated assault. Probably this reflects a high rate of community stability and assault is negative, it is positive in the regression equation. The data available for analysis could provide no explanations for variation in random vio- lence prevalence or the positive Beta of stability and aggra- vated assault.

While the matrix of variables entering regression equations to explain variation in crime rates in neighborhoods with high and low proximity of poor and middle-class residents is clearly different, and sometimes confusing, it should be remembered that proximity explains more of the variation in crime rates than any other variable.

Does the strength of the relationship between proximity of poor and middle-class residents and rates of violent crime invalidate the known relationships between race, southern migrant status, or economic conditions and criminal violence?

I think not. The power of proximity of residence to explain crime rates probably comes from its relationship to both the unequal distribution of income in some neighborhoods and the racial composition of communities. Proximity represents two aspects of racial discrimination-the extreme racial segrega- tion of housing units in Chicago and the lack of housing choices for blacks. Most violent crime in Chicago, including robbery, is between members of the same race. Rates of vio-

56 CRIMINOLOGY / MAY 1979

lence against blacks are far higher than against whites. Much of this difference probably reflects the burden of discrimination of black people in a northern urban community.

NOTE

I . A complete copy of this matrix can be obtained from the author.

REFERENCES

BLOCK, R. (1977) Violent Crime: Environment, Interaction and Death. Lexington, MA: Lexington Books.

_-- and C. BLOCK (1979) A Look in the Black Box: The Transformation of Robbery Incidents into Official Robbery Statistics. Paper delivered at the con- vention of the American Sociological Association, Boston, August 1979.

BOOGS, S. (1965) "Urban crime patterns." Amer. SOC. Rev. 30 (December). BREARLEY, H. C. (1932) Homicide in the United States. Chapel Hill: University

of North Carolina Press. CAPONE, D. and W. NICHOLS (1975) "Urban structure and criminal mobility."

Paper presented at the 27th annual meeting of the American Society of Crimi- nology, Toronto, Canada.

CHILTON. R. J. (1964) "Continuity in delinquency area research: a comparison of studies for Baltimore, Detroit, and Indianapolis." Amer. Soc. Rev. 29 (Feb- ruary): 71-83.

CURTIS, L. (1975) Criminal Violence: National Patterns and Behavior. Lexington, MA: Lexington Books.

EBERT, P. and K. SCHWlRlAN (1968) "Metropolitan crime rates and relative deprivation." Criminologia 5 (February): 43-52.

FLEISHER, B. (1956) The Economics of Delinquency. Chicago: Quadrangle. GASTIL, R. D. (197 1)"Homicide and a regional culture of violence." Amer. Soc. Rev.

HANNERZ, U. (1969) Soulside. New York: Columbia University Press. HARRIES, K. (1974) The Geography of Crime and Justice. New York: McGraw-Hill. HUNTER, A. (1976) Symbolic Communities. Chicago: University of Chicago Press. LEVIN, Y. and A. LINDESMITH (1937) "English ecology and criminology of the

past century." J. of Criminal Law and Criminology 27 (March): 801-816. LOFTIN, C. and R. H. HILL( 1974)"Regional subculture and homicide: anexamina-

tion of the Gastil-Hackney thesis." Amer. SOC. Rev. 39 (October): 714-725.

36: 4 12-427.

Block / COMMUNITY, ENVIRONMENT, AND C R I M E 57

LUNDSGARDE, H. P. (1977) Murder in Space City. New York: Oxford University

MILLER, W. (1958) "Lower class cultureasa generatingmilieuofgangdelinquency."

PYLE, G. (1974) The Spatial Dynamics of Crime. Chicago: University of Chicago

REPPETTO, D. ( 1974) Residential Crime. Cambridge, MA: Ballinger. SHAW. C. and H. D. McKAY (1972) Juvenile Delinquency and Urban Areas. Chi-

SHEVKY, E. and W. BELL (1955) Social Area Analyses. Standard, CA: Stanford

SILVER, A. (1967) "The demand of order in civil society," pp. 1-24 in David Bordua

WOLFGANG, M. E. (1958) Patterns in Criminal Homicide. Philadelphia: University

__- and F. FERACUTI (1967) The Subculture of Violence: Towards an Integrated

Press.

J. of Social Issues 1 4 5-19.

Press.

cago: University of Chicago Press.

University Press.

(ed.) The Police: Six Sociological Essays. New York: John Wiley.

of Pennsylvania Press.

Theory in Criminology. London: Tavistock.

Richard Block is an Associate Professor of Sociology at Loyola University of Chicago. He received his Ph. D. from the University of Chicago. Currently he is developing a technique for analyzing patterns of crime cross-nationally and is applying these proceedures in a comparison of patterns of crime in the Netherlands and the United States.