Book Review: Neighborhood Structures and Crime: A Spatial Analysis

• Published date: 01 December 2011
• Author: John R. Hipp
• DOI: 10.1177/0734016811403573
• Date: 01 December 2011
• Subject Matter: Book Reviews

traditional justice systems, more effective forms of justice, and safer communities. Their book is also

highly readable and accessible to those new to restorative justice. It is essential reading for anyone

interested in restorative justice theory and practice.

G. Kikuchi

Neighborhood Structures and Crime: A Spatial Analysis El Paso,

TX: LFB Scholarly Publishing LLC., 2010. xiii, 200 pp. $65.00. ISBN: 9-781-593-32396-7

Reviewed by: John R. Hipp, University of California, Irvine, CA, USA

DOI: 10.1177/0734016811403573

In Neighborhood Structures and Crime: A Spatial Analysis, George Kikuchi explores the spatial and

temporal dimensions of crime in neighborhoods. The book is motivated by two key ecological

theories: social disorganization theory and routine activities theory. Nonetheless, the strength of the

book comes less as a study of theory-driven empirical research on the change in crime in neighbor-

hoods over time—as the large implication of the empirical findings for the theories under study is

often hard to discern—but instead as a demonstration of a panoply of quantitative techniques that

can be used to explore longitudinal and spatial data. To be sure, the book is not explicitly didactic, so one

will not walk away from the analysis with a deep understanding of how to use the techniques but instead

the focus is on providing a flavor of the techniques. As a consequence, the reader with a less quantitative

bent should be prepared for quantitative analyses that are not always distilled to a more digestible form.

Readers should also be forewarned that stylistically, the book reads more like a three-article dissertation.

The book contains three analytical chapters. The first analytical chapter uses data for census tracts

in Seattle, Washington from 1960 to 2005. Some preliminary exploratory spatial data analysis

techniques are employed. Following this, a latent trajectory model (LTM) is used to test whether

characteristics of the neighborhood in 1960 affect the trajectory of crime over the next 40 years.

Given that there are numerous demographic and socioeconomic changes occurring in these neigh-

borhoods over this same 40-year period, it is not clear why we would expect the static characteristics

in 1960 to affect this entire crime trajectory. Therefore, later in the chapter, he estimates trajectory

modelsthat include measurescapturing the changingdemographics over thedecade to test whether they

predictthe change in crime rates overthe same decade, which arguablyprovides a more appropriate test

of the social process. Kikuchi then turns to panel regression models, in which incorporating spatial

effectsfrom nearby neighborhoodsis more straightforwardthan in LTM’s. There is some evidencefrom

these models that variables inspired by the routine activities theorybetter explain crime when they are

treated as time varying—that is, a shorter timehorizon for their effects. And measurescapturing social

disorganizationseemed to be more important for explaining long-term trajectories of crime.

The second analysis chapter demonstrates geographically weighted regression (GWR) models.

The power of these models is the ability to estimate a different value for a coefficient in each tract,

rather than a single coefficient value for all tracts in the city, as is done in most estimation

techniques. Although this is a technique growing in popularity, detecting spatial variability in

coefficients does not provide insight into why such variability might exist. The technique

produces coefficient estimates that can be plotted on a map that show the geographic pattern of the

results, as Kikuchi does for the variables in his models. The question then is determining whether

tracts that containeither similarly positive, or similarly negative, coefficients are similar in somefash-

ion. That is, the task is then to explain these parameter values (i.e., these values are specified as the

outcome of a model). A successful identification of neighborhood characteristics that vary with such

estimated coefficients suggests theneed to include interaction terms in the model to account for these

528 Criminal Justice Review 36(4)