This article was downloaded by: [Laurentian University]

On: 07 October 2014, At: 09:51

Publisher: Routledge
Informa Ltd Registered in England and Wales Registered
Number: 1072954 Registered office: Mortimer House, 37-41
Mortimer Street, London W1T 3JH, UK

Journal of Crime and

Justice
Publication details, including
instructions for authors and
subscription information:
http://www.tandfonline.com/loi/
rjcj20

ROUTINE ACTIVITIES
AND MOTOR VEHICLE
THEFT: A CRIME
SPECIFIC APPROACH
a
Heith Copes
a
University of Tennessee
Published online: 10 Jan 2012.

To cite this article: Heith Copes (1999) ROUTINE ACTIVITIES

AND MOTOR VEHICLE THEFT: A CRIME SPECIFIC APPROACH,
Journal of Crime and Justice, 22:2, 125-146, DOI:
10.1080/0735648X.1999.9721097

To link to this article: http://

dx.doi.org/10.1080/0735648X.1999.9721097

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy

of all the information (the “Content”) contained in the
publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or
 warranties whatsoever as to the accuracy, completeness,
or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and
views of the authors, and are not the views of or endorsed
by Taylor & Francis. The accuracy of the Content should not
be relied upon and should be independently verified with
primary sources of information. Taylor and Francis shall not be
liable for any losses, actions, claims, proceedings, demands,
costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection
with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private

Downloaded by [Laurentian University] at 09:51 07 October 2014

study purposes. Any substantial or systematic reproduction,

redistribution, reselling, loan, sub-licensing, systematic
supply, or distribution in any form to anyone is expressly
forbidden. Terms & Conditions of access and use can be found
at http://www.tandfonline.com/page/terms-and-conditions
 Journal of Crime & Justice Volume 22 No. 2 1999 125

ROUTINE ACTIVITIES AND MOTOR VEHICLE

THEFT:
A CRIME SPECIFIC APPROACH
Downloaded by [Laurentian University] at 09:51 07 October 2014

Heith Copes
University of Tennessee

ABSTRACT

The routine activity approach to crime predicts that the routine daily
activities of populations affect the availability of targets and, thereby,
shape the crime rate. Routine activity researchers have realized the
benefits of using crime-specific models. The current study follows this
tradition by examining how the routine daily activities of populations
affect the supply ofpotential offenders, the availability of targets, and the
level of capable guardianship against crime, and thereby, shape the rate
of motor vehicle the$. Results from multiple regression analysis show
that measures of potential offenders, suitable targets, and guardianship
explain variation in the rate of motor vehicle the$. Findings suggest that
crime-specific measures are an appropriate method for examining the
merits of routine activity theory.

INTRODUCTION
Sociologists and lay people alike have long known that some urban
areas are more dangerous than others. Underlying the assumption that
some places are to be avoided, is the understanding that crime is not a
random event but occurs in a determinative social context. This is the
point of departure for an approach to crime known as opportunity theory.
Perhaps the best-known opportunity theory is the routine activity
approach (Cohen & Felson, 1979). Routine activity theory focuses on the
social circumstances that generate high rates of crime. It suggests that the
 126 Journal of Crime & Justice

daily routines characteristic of a population affect the quantity and

location of crime by determining how frequently potential offenders
encounter suitable targets.
With the gaining popularity of rational choice theories researchers
have begun examining the factors that contribute to the prevalence of
specific crimes. Traditionally, researchers have pooled various crime
types into a single measure. By doing so, they assume that the underlying
causes of crime are the same for different types of crime. Numerous
researchers, however, have found that routine activities explain one type
Downloaded by [Laurentian University] at 09:51 07 October 2014

of crime but not another (Maxfield, 1987; Sampson & Wooldredge, 1987;
Thompson & Fisher, 1996). Researchers using the routine activity
approach have followed suit in testing the theory crime-specifically. The
lnfluence of routine activities has been used to explain a variety of crimes
such as larceny (Mustaine & Tewksbury, 1998), burglary (Robinson &
Robinson, 1997), arson (Stahura & Hollinger, 1988), and homicide
(Kennedy & Silverman, 1990; Messner & Tardiff, 1985). The current
study follows this tradition by examining a previously neglected area of
crime using crime-specific measures. This study examines how the
routine daily activities of populations affect the supply of potential
offenders, the availabilityof targets, and the level of capable guardianship
against crime, and thereby, shape the rate of motor vehicle theft.
Specifically, it uses social and physical characteristics of census tracts to
explain motor vehicle theft rates in a southern parish (county).

ROUTINE ACTIVITIES
The routine activity approach posits that variation in crime rates is
produced by variation in the supply of motivated offenders, suitable
targets, and capable guardians. The structure of everyday life influences
the volume, location and type of crime by influencing how frequently the
previous three elements occur together. Whereas most criminological
theories explain crime trends by positing variation in the number of
persons predisposed to offend, Cohen & Felson (1979; Felson & Cohen,
1980) argue that the structure of a society’s work, leisure, and education
patterns increase criminal opportunities without necessarily increasing
criminal motivation. When applied to the problem of spatial variation in
crime, a neighborhood can have a hgher rate of crime even when the
supply of offenders remains constant simply by providing more criminal
opportunities. Cohen and Felson (1979) illustrated this idea by showing
 Journal of Crime and Justice 127

how social changes in society (increases in the number of females

entering into the labor force and increases in the number of nontraditional
households) led to routine activities taking place further away from the
homes, thus creating greater opportunities for crime.
Investigators have used routine activity theory to explain variation
in crime rates in an array of localities that range from nations to street
comers. Bennet (1991) examined 52 nations using macro-structural
measures of routine activities and found that the theory was crime-
Downloaded by [Laurentian University] at 09:51 07 October 2014

specific, working better for property crime than for personal crime. An
examination of the 124 largest urban areas in the United States found that
cities where people frequently participate in nonhousehold leisure
activities had hgher rates of property crime than in cities where they stay
home more (Messner & Blau, 1987). This relationship was expected
because non-household leisure activities draw people out of their homes;
thus, makmg people more ldcely to encounter potential offenders and to
leave their home and property unprotected.
The routine activity approach has also been used to explain
criminogenic “hot spots” - the concentration of crime in specific locations.
A study of residential city blocks in Cleveland, Ohio, found that crime
occurred more frequently on blocks with bars or taverns (Roncek &
Maier, 1991) and on blocks adjacent to public high schools (Roncek &
Faggiani, 1985). Also, Sherman, Gartin, & Buerger (1989) analyzed
more than 300,000 calls to the police in Minneapolis and found that 50
percent of the calls came from only 3 percent of the city’s addresses and
intersections. These “hot spots” included large department stores, bars,
convenience stores, and a public housing apartment building. Bars,
schools, and other public places facilitate routine activities that are
conducive to crime by bringing unacquainted people together, thereby
reducing social controls. In addition, they attract those who are most
predisposed to offend, namely, young-males.
While the routine activity theory is widely tested and underpins
many crime prevention initiatives, it does have critics. One criticism is
that there is a need to develop more specific models to test and further
develop the theory (Mustaine & Tewksbury, 1997). For instance, routine
activity theorists often aggregate crimes into broad categories, which
make it difficult to determine the effects that hypothesized variables have
on specific crimes. Previous research has shown that routine activity
variables explain property crimes better than they explain violent crimes
(Bennet, 1991; Meithe & Meier, 1987). Burglary and household larceny
are not influenced by the same routine activity variables because they
 128 Journal of Crime & Justice

have different opportunity structures (Maxfield, 1987; Sampson &

Wooldredge, 1987; Thompson & Fisher, 1996). Mustaine & Tewksbury
(1998) found that specific lifestyle measures, such as eating out and
studying out, are better predictors of victimization than general measures,
such as the number of evenings spent away from the home.
The way Cohen & Felson (1979) operationalize guardianship
illustrates the limitations of not measuring concepts crime-specifically.
They measured guardianship as the percentage of females who work
Downloaded by [Laurentian University] at 09:51 07 October 2014

outside of the home, much like numerous other researchers. This measure
assumes that the level of guardianshp decreases when no one is home to
protect property, which is logical if we are studying burglary because
vacant homes are unprotected. Using this measure makes less sense if the
target is an automobile because people usually bring their car with them
when they travel. Therefore, the number of persons home does not
measure adequately guardianship for all crimes.
Cantor & Land’s (1985) finding that guardianshp is related to theft
but not to violent crime supports this critique. Another example of this
problem is shown in an examination of the 124 largest urban areas in
1980 (Messner & Blau, 1987). Measures of nonhousehold activities were
significantly related to burglary and larceny but not to motor vehicle theft.
By assuming that leaving home reduces guardianship, they were unable
to explain variation in rates of motor vehicle theft. These findings suggest
that future research should avoid pooling crime types because the use of
general measures for concepts may confound the relative importance of
offenders, targets, and guardianship.

ROUTINE ACTIVITY AND MOTOR VEHICLE

THEFT
Motor vehicle theft has received little theoretical attention from
investigators. Most studies are descriptive; that is they describe types of
motor vehicle theft, types of models stolen, characteristics of offenders or
ways to prevent car theft without placing it in a theoretical framework.
While no study has used the routine activity approach to examine motor
vehicle theft specifically, many studies have provided findings that are
compatible with the theory. For example, one tenet of routine activity
theory states that increases in the availability of a product increases its risk
of becoming a target. Cohen & Felson (1979) illustrated this belief by
showing how the rise in small, portable goods coincided with an increase
 Journal of Crime and Justice 129

in household property crime. Similarly,the increase in the availability of

cars between 1921 and 1925 led to an increase in the rate of auto theft
during that same period (Gould, 1969; Mansfield, Gould, & Namenworth,
1974).
Routine activity theory also posits that proximity to motivated
offenders increases the likelihood of a target being victimized. Urban
areas that have a large pool of potential offenders tend to have high rates
of motor vehicle theft. City blocks with bars have nearly twice as many
auto thefts as city blocks without them (Roncek & Bell, 1981; Roncek &
Downloaded by [Laurentian University] at 09:51 07 October 2014

Maier, 1991). Further, blocks adjacent to high schools have higher levels
of auto theft than blocks that are not near high schools (Roncek &
Faggiani, 1985). Bars and schools are thought to attract those most ldcely
to be motivated to commit a crime, namely, males and the young, thereby,
increasing the llkelihood of crime.
In addition to the availability of automobiles and the size of the
offender pool, there must be a sufficiently low level of capable
guardianship for the offender to elude detection and arrest. This is
evidenced by the fact that nearly 74 percent of auto thefts are committed
during the evening hours (Hope, 1987). In England, Canada, and
Australia, parking lots with attendants have lower rates of auto theft than
similar lots with no attendant on duty (Clarke, 1983; Engstad, 1975;
NRMA, Ltd., 1990). Some cars are chosen not because they can be stolen
easily but because they can be concealed more effectively. Cars stolen to
be exported to Mexico tended to be the same types of cars that are made
and sold there (Field, Clarke & Harris, 1992; Miller, 1987). By choosing
cars that are common in Mexico, thieves cut their chances of being easily
recognized.
Note that guardianshipnot only refers to people protecting property,
but also to target hardening activities that make the vehicle harder to steal.
In 1988, heavy trucks, buses and motorcycles had a significantly higher
theft rate per 100,000 vehicles than did passenger cars. Harlow (1988)
accounts for this finding by suggesting that trucks, buses and motorcycles
are simply easier to steal. This may be shown by the higher number of
uncompleted thefts for cars than for trucks and motorcycles. Certain
passenger cars are more often selected as targets based on their level of
security. The poor ignition locks of the 1969-1974Ford models has been
attributed as a major cause for their disproportionately high rate of theft.
Further evidence for this is the fact that theft rates for Fords fell by a
quarter after locks were upgraded in 1975 (Karmen, 1981). Sirmlarly, the
introduction of steering column locks effectively reduced car theft rates
 130 Journal of Crime & Justice

in the United States (Mayhew, et al., 1976). These findings led Clarke
and Harris (1992) to suggest that the best way to prevent auto theft is to
manufacture more secure vehicles.
These findings suggest that a routine activity approach to motor
vehicle theft is a promising area of research. The current study builds on
previous routine activity research by examining a previously neglected
area of crime (motor vehicle theft) using crime-specific measures. It is
hypothesized that the rate of motor vehicle theft is a function of the
Downloaded by [Laurentian University] at 09:51 07 October 2014

number of offenders, available targets, and level of guardianship

characteristic of census tracts. Consequently, three bivariate hypotheses
about motor vehicle theft are suggested. First, the rate of motor vehicle
theft varies directly with the supply of potential offenders. Second, the
rate of motor vehicle theft varies directly with the availability of vehicles.
Finally, the rate of motor vehicle theft varies inversely with the level of
capable guardianship. Areas with large numbers of offenders, with large
numbers of targets, and with low levels of guardianship are expected to
have high rates of motor vehicle theft. In addition, it is expected that
there will be an interaction among the three variables that affects the rate
of motor vehicle theft beyond the additive effects.

DATA AND METHODS

Previous research has found a strong relationship between crime

rates and the characteristicsof urban areas. Complications arise, however,
when data are aggregated to cities or nations because they are less
homogeneous than neighborhoods (Blalock, 1982). Aggregation of
dissimilar units confounds the effects of the variables of interest in macro-
level studies. Standard methodological procedure calls for dlsaggregating
variables to the smallest theoretically appropriate level possible. The
units of analysis for this study are the 41 census tracts in a southern U.S.
parish. Census tracts are used because they are one of the smallest
aggregate areas that possess the environmental features that are thought
to increase the chance of a convergence among the three components of
predatory crime. While city blocks (Roncek & Faggiani, 1985; Roncek
& Maier, 1991) and street segments (Robinson, 1999) have been used by
previous researchers, they are too small to generate rates of motor vehicle
thefts that are appropriate for this analyses.
 Journal of Crime and Justice 131

Dependent Variable
The dependent variable for this study is the rate of motor vehicle
theft for each census tract. Theft rates were measured by using data made
available by the Lafayette, Louisiana, Police Department. Incident reports
were obtained on all motor vehicle thefts reported to the police for the
years 1994 through 1996. These reports included the time and date of
theft, the location of the theft, the make and model of the vehicle, the
Downloaded by [Laurentian University] at 09:51 07 October 2014

estimated value of the vehicle when it was stolen and when it was
recovered. These reports only included vehicles that were stolen and not
those that were vandalized or burglarized. The motor vehicle theft rate
for this study was determined for each census tract by dividing the
number of thefts by the number of registered vehicles in that tract and
then multiplying the quotient by 1,000.
The use of official statistics is problematic because they tend to
underestimate the extent of crime. Information from the UCR and the
NCVS indicate, however, that motor vehicle theft has the highest
reporting rate of all Type I offenses. The high rates of reporting are
attributable to several factors. For instance, a car is the most expensive
single item the majority of us have purchased. The loss of such a large
investment is a great economic strain. The hope that the police may be
able to relieve h s pressure by recovering the car is sufficient for victims
to report the crime, when they might otherwise not. Another reason for
the high levels of reporting is that insurance agencies require a police
report. For those fortunate enough to have purchased theft insurance
reporting is a necessary step in recovering some of their loss. A final
reason for the heightened rate of reporting is that many people believe
their vehicle may have been stolen for the commission of another crime.
Although this represents only a small part of all auto thefts (McCaghy et
al. 1977) the fear of being accused for someone else’s crime is strong
enough to warrant a call to the police.
To determine the number of motor vehicle thefts for each census
tract, theft locations were mapped using U.S. Census Bureau
Topologically Integrated Geographic Encoding and Referencing
(TIGEWLine) files and a software program entitled ArcView.
TIGEWLine files are computer readable maps that were created to aid
census and survey programs (U.S. Bureau of the Census, 1991). They
include information for all census features, such as, streets, blocks, tracts,
and counties. Often, however, TIGER/Line files do not contain complete
data. In some areas, for example, the full range of street numbers is not
 132 Journal of Crime & Justice

included, which creates problems in locating some addresses. Despite

these problems, the location of 1,279 of the 1,438 reported thefts (89
percent) were able to be plotted.
Independent Variable
The independent variables represent sociodemographic and physical
characteristics of the census tracts. Table 1 provides a list of the
variables, the manner in which they were measured, and the means and
Downloaded by [Laurentian University] at 09:51 07 October 2014

standard deviations for each. Measures of the independent variables used

in the analysis are taken from 1990 census documents.
Independent variables regarding motivated offenders include percent
young-male and percent poor. These variables are used because the UCR
(1997) and prior research (Clarke & Hams, 1992; McCaghy, Giordano,
& Henson, 1977) suggest that those most frequently arrested for motor
vehicle theft are young, male, and poor. The only variable that needs
further explanation is percent young-males. It is operationalized as the
percentage of the population that are males aged 15 to 24 for each tract.
It is hypothesized that both percent young-male and percent poverty will
be directly related to motor vehicle theft. That is, as the percentages of
these two variables increases, so will the rate of motor vehicle theft.
The supply of available motor vehicles is operationalized by two
indicators: car density and road density. These variables represent
physical environmental features of the census tract that affect the number
of cars that are located in or pass through the tract, and therefore
opportunities. Increased opportunities should increase the rate of theft,
therefore, these variables should have a direct influence on motor vehicle
theft. Car density is measured by dividing the total number of cars in a
census tract by that tract’s size, in square miles. This variable measures
the number of cars located in a tract. Automobile traffic also influences
the availability of vehicles in an area. The number of cars on the road is
positively related to the frequency of car thefts (Wilkins, 1964). Ideally,
car traffic should be measured by counting the number of vehicles that
pass a given point. For instance, Robinson (1999) used an electronic eye
counter to record the amount of automotive traffic at various street
locations. Unfortunately, a direct measure of car traffic was not available
at the tract level for this study because of cost and time constraints, thus
an indirect measure had to be used. One measure used is the number of
roads in an area (Boggs, 1965). Road density is measured by dividing the
number of roads that are in or passed through the tract by its size in square
miles.
 Journal of Crime and Justice 133

Table 1
VARIABLES, AND HYPOTHESIZED
MEASUREMENTS, RELATIONSHIPS

Variables Measurements Mean S.D.

Motor Vehicle (#Thefts / #Cars) X 1000 14.67 9.23

Downloaded by [Laurentian University] at 09:51 07 October 2014

Theft Rate

Offenders
Percent Poor # Below Poverty / Population 22.19 15.78
Percent Young-Males # Males Age 15-24 / Population 8.49 5.28

Targets
Road Density # Roads / Square Miles 32.97 31.79
of the Census Tract
Car Density # Cars / Square Miles 1013.49 858.99
of the Census Tract

Guardianship
Multi-Unit Housing # Non Multi - Unit Housing*/ 86.24 15.30
# Housing Units
Population Density # People / Square Miles 1912.00 1873.16
*Structures with five or more housing units.

The guardianship variable is operationalized using population

density and percent multiple-unit housing. According to Felson (1994),
densely populated areas create a “natural” surveillance, thereby helping
to deter crime. Felson notes that someone who has stake in the property
is the best guardian against crime. Nevertheless, ordinary citizens going
about their daily routines can act as significant deterrents to would be
offenders. More densely populated areas mean more opportunity for
informal guardianship and, consequently, a lower rate of motor vehicle
theft. Density is measured by dividing the total number of people in a
tract by its area, in square miles.
Because unoccupied cars are the principal targets of motor vehicle
theft, cars parked on the street or in large lots are at a higher risk of being
stolen than cars parked in garages with perimeter security (Clarke, 1983;
Hope, 1987; NRMA, 1990; Engstad, 1975). Multiple-unit housing
 I34 Journal of Crime & Justice

interferes with people’s ability to provide adequate protection for their

cars by forcing residents to leave them unattended in dangerous areas.
Therefore, the percentage of housing occupied by multiple-unit structures
is used as a measure of guardianship. Multiple-unit housing is
operationalized in this way so that both guardianship variables will
measure informal protection. That is, as both variables increase the rate
Downloaded by [Laurentian University] at 09:51 07 October 2014

of motor vehicle theft should decrease. A multiple-unit structure is

defined as a structure with five or more housing units.

Analysis

Hypotheses are tested using a two-stage strategy of ordinary least

squares regression. In the first stage, separate equations are run for each
construct (potential offenders, target availability, and guardianship) and
for all variables together. In this stage, we will be able to determine how
well the variables representing each construct account for the variance in
motor vehicle theft. In the second stage, single indicators for each
theoretical construct are created by standardizing the variables and
summing their z-scores. Next, the dependent variable was regressed on
the theoretical indicators. This stage will determine which theoretical
concept explains the greatest amount of variation in motor vehicle theft
while accounting for multicollinearity. It also allows the hypothesized
interactions among the independent variables to be tested.

RESULTS
Bivariate correlations indicate that all but two of the independent
variables are significantly correlated with motor vehicle theft (see Table
2). The relationship between the young-male variable and motor vehicle
theft is not significant, but it is in the predicted direction. The correlation
between motor vehicle theft and the percent multiple-unit housing is
neither significant nor in the predicted direction. Density was expected
to be inversely correlated with motor vehicle theft, but this was not the
case. In the current study the rate of motor vehicle theft increases with
densely populated areas.
 Downloaded by [Laurentian University] at 09:51 07 October 2014

Table 2
ZERO-ORDER CORRELATION MATRIX

Theft Percent Percent Car Road Population Percent

Rate Poor Young-Male Density Density Density Multi-Unit

Theft Rate 1.oo

Percent Poor .70* 1.oo

Percent
Young-Male .12 .48* 1 .oo
Car Density .32" .14 .I6 1.oo

Road Density .64' .63' .48. .72' 1.oo

Population
Density .45* .57* .61* -.71* .81' 1.oo
\
Percent E
Multi-Unit .09 -.02 -.58* .36** -.33** -.39** 1.oo 5
'p s .01
*.
p <_ .05
 136 Journal of Crime & Justice

Stage One
To examine the additive effects of the routine activity variables, four
separate regression equations were computed for motor vehicle theft.
Separate equations were run by construct (motivated offenders, target
availability, and guardianship) and for all variables together (Table 3).
With this technique it is possible to observe how well the variables
representing each construct account for the variance in motor vehicle
Downloaded by [Laurentian University] at 09:51 07 October 2014

theft, as well as the effect of the full equation.

In the first model, motor vehicle theft was regressed on the
independent variables representing potential offenders (see Columns 1 in
Table 3). As predicted, percent poor was directly related to motor vehicle
theft. Those areas with higher percentages of people below the poverty
line tended to have higher rates of motor vehicle theft than areas with
lower percentages of poor people. The percentage of young males in a
population was found to be significantly related to the rate of motor
vehicle theft, but not in the direction expected. These findings suggest
that areas with more young males have lower rates of motor vehicle theft
than areas that have fewer males.
In the second model, motor vehicle theft was regressed on the
variable representing target suitability (see Columns 2 in Table 3). Both
variables were expected to be positively correlated with motor vehicle
theft. These predictions were not fully supported. Road density was
significantly related to motor vehicle theft when controlling for car
density; that is, those areas with more roads running through them have
higher rates of motor vehicle theft than areas with fewer roads. Car
density, on the other hand, was significantly related to motor vehicle theft
in the bivariate statistics, but it is not significantly related to motor vehcle
theft when controlling for road density.
In the third model, motor vehicle theft is regressed on the variables
representing guardianship (see Column 3 in Table 3). It is hypothesized
that population density is negatively correlated with motor vehicle theft;
that is, more people in an area should provide higher levels of informal
guardianship by having more eyes on the streets, thereby deterring crime.
The current findings indicate that the rate of motor vehicle theft vanes
positively with population density. This finding suggests that more
densely populated areas have higher rates of motor vehicle theft than
areas that have fewer inhabitants per square mile, the opposite of what
was predicted. Similarly, multiple-unit housing is significantlyrelated to
motor vehicle theft in the opposite direction as was hypothesized. High
 Journal of Crime and Justice 137

concentrations of multiple-unit dwellings was expected to increase the

rate of motor vehicle theft because these types of homes are often unable
to provide adequate supervision of their vehicles. The analysis show that
areas with more multiple-unit dwellings tend to have lower rates of motor
vehicle theft than areas with fewer multiple-unit dwellings.

Table 3
Downloaded by [Laurentian University] at 09:51 07 October 2014

OLS REGRESSION ESTIMATES FOR PARTIAL AND FULLMODELS

PREDICTING THE RATE OF MOTOR VEHICLE THEFT

Offender Target Guardianship Full

Variab1e Model Model Model Model
Offender Pool
Percent Poor .84* .59*
Percent Young-Male -.28* -.32*

Suitable Targets
Road Density .84* .50*
Car Density -.28 .01

Guardianship
Percent Multiple-Units .32* .08
Population Density .58* -.05

R2 .55 .45 .29 .67

NOTES: Only the standardized coefficients are shown. All F values were significant at the
.05 level.
p I.05

In the final equation, motor vehicle theft is regressed on all of the

independent variables (see Column 4 in Table 3). The results of the full
equation indicate that, after controlling for the other variables, only
percent poor and road density are significantly related to motor vehicle
theft. The remaining variables were not significantly related to the rate
of motor vehlcle theft. Two of the variables (car density and population
density) switched signs in the full model. Car density was significant in
neither the reduced nor the full model. Population density was significant
in the reduced model but not in the full model. The fact that these two
 I38 Journal of Crime & Justice

variables switched signs after additional variables were included suggests

that multicollinearity may be a problem. The second stage of analysis
addresses the multicollinearity problem.

Stage Two

To account for multicollinearity and to check for the hypothesized

interactions a single indicator for each theoretical construct (offender
Downloaded by [Laurentian University] at 09:51 07 October 2014

pool, suitable targets, and guar&anshp) was created by standarduing and

summing their constituent variables. The offender pool was created by
adding together the standardized scores of percent poor and percent
young-males. The available targets indicator was created by summing the
z-scores for road density and car density. Guardianship was created by
standardizing population density and percent multiple-unit housing and
a d b g them together. Motor vehicle theft then was regressed on the three
indicators (see Table 4).

Table 4
OLS REGRESSION ESTIMATESFOR SINGLE INDICATORS PREDICTING
THE RATE OF MOTORVEHICLE THEFT

Variable D b S.E

Offender Pool .26 2.94* 1.58

Suitable Target .2s 5.82* 3.06

Guardianship .30 5.20* 2.43

*p I .01
Rz=.42

Routine activity theory predicts that the rate of motor vehicle theft
varies directly with the supply of motivated offenders, varies directly with
the number of available targets, and varies indirectly with the level of
capable guardianship; that is, areas with large numbers of offenders, with
large numbers of targets, and with low levels of guardianship are expected
to have higher rates of motor vehicle theft than areas without these
characteristics. We expect this because these areas provide more criminal
 Journal of Crime and Justice 139

opportunities. In addition, it is expected that there will be an interaction

among the three variables which affects the rate of motor vehicle theft
beyond the additive effects.
Table 4 displays the results of this analysis. Offender pool, suitable
targets, and guardianship are all marginally significant (p< .05) predictors
of motor vehicle theft. Offender pool and suitable targets were related to
the dependent variable in the hypothesized direction, but guardianship
was significantly related to motor vehicle theft in the opposite direction.
Downloaded by [Laurentian University] at 09:51 07 October 2014

An examination of the standardized betas reveals that all three variables

explain roughly the same amount of variation, guardianship explains
slightly more than the other two. This finding supports the contention that
crime is the result of a convergence of all three elements and that the
removal of one is sufficient to reduce the amount of crime in an area.
All possible two-way and three-way interactions also were tested
using backward stepwise regression. Only two of the interactions were
significant. The offender variable interacted with both the target variable
and the guardianship variable. The relationship between guardianship and
motor vehicle theft is a function of the supply of potential offenders; that
is, the level of capable guardianship has an effect on motor vehicle theft
only when the supply of offenders is high. In those areas with fewer
potential offenders, guardianship has no significant impact on the rate of
motor vehicle theft. Guardianship influences the rate of motor vehicle
only when there are many potential offenders. The second significant
interaction suggests that as the availability of targets increases, the
influence of offenders on the rate of motor vehicle theft also increases.
The relationship between potential offenders and motor vehicle theft is a
function of the availability of targets. The three-way interaction was not
significant.

CONCLUSION AND DISCUSSION

This paper presents an investigation on how the routine activities
characteristic of an area influence that area’s rate of motor vehicle theft.
While the results of the study provide moderate support for routine
activity theory, the reader should be aware of two caveats before
generalizing these findings to other situations. First, one of the main
criticisms of routine activity theory is its reliance on proxy measures to
test its main assumptions (Meithe & Meier, 1990; 1994; Mustaine &
Tewksbury, 1997; 1998). The reliance on proxy variables makes it
 140 Journal of Crime & Justice

difficult to rule out alternative explanations forcing the researcher to go

beyond the data to draw conclusions about their findings. This study was
able to address this issue for the suitable target variables, but not for the
other key concepts.
Second, the data come from a single parish (county) in Louisiana. It
is possible that the current results are unique to the area. Motor vehicle
theft rates vary across regions of the country. For example, the western
region of the country consistently has higher rates of motor vehicle theft
Downloaded by [Laurentian University] at 09:51 07 October 2014

than the other regions, while the south usually ranks in the middle (UCR,
1997). According to Ranking of U S . Cities, this city's motor vehicle theft
rate falls close to the mean, ranking 167thout of 300 cities (Morgan &
Morgan, 1999). The fact that Lafayette is close to the average reduces the
risks of generalizing to other areas, it does not, however, fully eliminate
that risk.
Despite these caveats the findings from this study are useful for a
better understanding of the routine activity approach. The results of the
regression analyses support the hypotheses that the rate of motor vehicle
theft varies directly with the supply of potential offenders. In fact, the
offender variables explain roughly the same amount of variation in motor
vehicle theft rates as does the available target variables. Note that routine
activity theorists often downplay the role of motivated offenders by
assuming their presence and overemphasize the importance of criminal
opportunities. This assumption does not appear to be justified in light of
the previous findings. Results from this study show that the size of
offender pools is as good a predictor of a real variation in the motor
vehicle theft rate as is the number of criminal opportunities.
In the first stage of this analysis, percent poor made the greatest
contribution toward explaining variation in the rate of motor vehicle theft.
Consistent with expectations, as the percentage of poor increases so does
the rate of motor vehicle theft, even after controlling for the other
variables. The young-males variable, however, did not significantly
explain variation in motor vehicle theft. This is not the first study to find
no relationship between young-males and crime (Stahura & Hollinger,
1988). One explanation for this finding is that areas with a high
proportion of young people tend to have fewer cars, thereby decreasing
opportunities. The correlation between percent young-males and the total
number of cars in an area is negative (r = - .27), however it is not
significant. It is also possible that young males leave their neighborhoods
to steal the cars they desire. The current data set, however, does not
contain the necessary information to test this hypothesis.
 Journal of Crime and Justice I41

The suitable target index significantly explained the rate of motor

vehicle theft as predicted. When the two constituent variables are
separated out, only road density was significantly related to the rates of
motor vehicle theft. This finding suggests that the number of available
targets is not sufficient to make that item susceptibleto high rates of theft.
The offender also must have a convenient way to remove the object that
they desire and a quick escape route. Rengert's (1996) study of burglary
in Greenwich, Connecticut, found that 37 percent of burglaries were
committed less than a half rmle from a freeway exit. A study in Canada
Downloaded by [Laurentian University] at 09:51 07 October 2014

found that houses with more street segments leading to it have a greater
chance of being burglarized than houses with fewer streets leading to it
(Beavon, Brantingham, & Brantingham, 1994). A house on a cul-de-sac
is at a lower risk of crime than a house on the a corner. Thus, more roads
provide more opportunities for thieves to get away with the stolen vehcle.
The data do not support the hypothesis that the level of capable
guarhanshp is inversely related to the rate of motor vehicle theft. In fact,
this study shows the opposite to be true. The second stage of analysis
showed that guardianship has a strong positive correlation with the rate
of motor vehicle theft. When the variables used to create the guardianshp
index are examined individually, the results are similar. Neither
population density nor multiple-unit dwelling density are significant
predictors of the rate of motor vehicle theft when controlling for the other
variables. The two guardianship variables are only significantly related
to motor vehicle theft when they are in a model alone. Felson (1994)
hypothesized that densely populated areas have the ability to provide
informal protection of property by having more eyes on the street. Our
results contradxt this hypothesis and are consistent with Shichor, Decker,
& 0' Brien (1980), who found that motor vehicle theft was the only non-
contact property crime that was not negatively correlated with density. It
appears that motor vehicle theft is influenced by a different set of factors
than other forms of property crime. The percentage of multiple-unit
housing has a similar effect on the rate of motor vehicle theft as did
population density. It was significant but not in the direction predicted.
A possible explanation for this finding is that a large proportion of
vehicles are stolen from non-residential, business areas, or places where
people generally do not live (Biles, 1974; Eck & Spelman, 1988).
Unfortunately, data about the number of businesses in a given tract were
unavailable so this hypotheses could not be tested. Robinson (1999)
found that the sporahc daily routines of students in apartment complexes
contributed to a lower rate of burglary. This same explanation, increased
 142 Journal of Crime & Justice

surveillance, may account for the lower rates of motor vehicle theft.
It is not uncommon for researchers to find strong or moderate
support for offender and target variables, while finding weak or no
support for guardianship variables (Bennet, 1991; Massey, Krohn, &
Bonati, 1989; Miethe & Meier, 1990; 1994 Stahura & Hollinger, 1988;
Stahura & Sloan, 1988). The weak support for the guardianship variables
can be explained in two ways, one theoretical and the other
methodological. Theoretically, it is possible that the level of guard~anshp
Downloaded by [Laurentian University] at 09:51 07 October 2014

is not as important to crime prevention as suggested by Cohen & Felson.

However, interviews with thieves suggests that guardianship is a
consideration in their decision-making (Wright & Decker, 1994).
Another, more likely, explanation is that researchers have not adequately
measured aggregate levels of guardianship. The measures used in macro-
level research may not be those factors that prevent crime at the street
level. In this case, car alarms, steering-wheel locks, and other target
hardening activities may be the best protection against car theft. As of
yet, no one has been able to obtain aggregate measures for these
protection measures. These types of behaviors may be more instrumental
in protecting car theft than the aggregate measures used in this study.
Therefore, more exact examinations of the relationship between
guardianship and crime rates should be conducted.
To summarize, this study demonstrates that the routine activities
associated with an area significantly affect the rate of motor vehicle theft.
As the number of potential offenders and the supply of available targets
increases in an area so does that area’s motor vehicle theft rate. The one
discrepancy between expectations and what was found is that the level of
guardianship had the opposite affect on motor vehicle theft rates than was
predicted. While this study produced results that are somewhat
inconsistent to Cohen & Felson’s (1979) original conception of the
routine activity approach, it nonetheless provided evidence in favor of the
theory, especially when using crime-specific measures.

REFERENCES
Beavon, D. J., Brantingham, P.L., Brantingham, P.J. (1994). “The Influence of
Street Networks on the Patterning of Property Offenses.” In Crime Prevention
Studies (Vol. 2), R.V. Clarke (Ed.), Monsey, NY: Criminal Justice Press.

Bennett, R. R. ( 1 99 1). “Routine Activities: A Cross-national Assessment of a

 Journal of Crime and Justice 143

Criminological Perspective.” Social Forces 70: 147-163.

Biles, D. (1974). “The Victims of Car Stealing.” Australian and New Zealand
Journal of Criminology 7: 99-109.

Blalock, H. M. (1989). Conceptualization and Measurement in the Social

Sciences. Beverly Hills, CA.: Sage.

Boggs, S. L. (1965). “Urban Crime Patterns.” American Sociological Review 30:

Downloaded by [Laurentian University] at 09:51 07 October 2014

899-908.

Cantor, D. & Land, K. (1985). “Unemployment and Crime Rates in Post-World

War I1 United States: A Theoretical and Empirical Analysis.” American
Sociological Review 50: 3 17-322.

Clarke, R. V. (1983). “Situational Crime Prevention: Its Theoretical Basis and

Practical Scope.” In Crime and Justice: An Annual Review of Research (Vol. 4),
M. Tonry and N. Morris (Eds). Chicago: University of Chicago Press.

Clarke, R.V. & Harris, P. M. (1 992). “Auto Theft and its Prevention.” In Crime
and Justice: A Review of Research (Vol. 16), M . Tonry (Ed.). Chicago:
University of Chicago Press.

Cohen, L. E. & Felson, M. (1 979). “Social Change and Crime Rate Trends: A
Routine Activity Approach.” American Sociological Review 44: 588-608.

Eck, J. & Spelman, W. (1 988). Problem Solving: Problem-Oriented Policing in

Newport News. Washington, DC: Police Executive research Forum and the
National Institute of Justice.

Engstad, Peter A. (1975). “Environmental Opportunities and the Ecology of

Crime.” In Crime in Canadian Society, R. A. Silverman and J. J. Teevan (Eds.).
Toronto: Butteworth.

Federal Bureau of Investigation, Department of Justice. (1 997). Uniform Crime

Report: Crime in the United States, Washington, DC: U. S. Government Printing
Office.

Felson, M. & Cohen, L. E. (1980). “Human Ecology and Crime: A Routine

Activity Approach.” Human Ecology 8: 389-406.

Felson, M. (1994). Crime and Everyday Life: Insights and Implications for
Society. Thousand Oaks, CA: Pine Forge Press.
 144 Journal of Crime & Justice

Field, S., Clarke, R.V., and Harris, P.M. (1992). “The Mexican Vehicle Market
and Auto Theft in Border Areas of the United States.” Security Journal 2: 210-
225.

Gould, L. C. (1969). “The Changing Structure of Property Crime in an AMuent

Society.” Social Forces 48: 50-59.

Harlow, C. W. (1988). Motor Vehicle Theft. Washington, D.C.: US. Department

Downloaded by [Laurentian University] at 09:51 07 October 2014

of Justice, Bureau of Justice Statistics.

Hope, T. (1987). “Residential Aspects of Auto Crime.” Research Bulletin 23.

London: Home Office Research and Planning Unit.

Hough, M. and Mayhew, P. (1 985). “Taking Account of Crime: Key Findings

from the British Crime Survey.” Home Oflce Research Study 85. London: H. M.
Stationary Office.

Karmen, A. (1981). “Auto Theft and Corporate Irresponsibility.” Contemporaty

Crisis 5: 63-81.

Kennedy, L. W. & Silverman, R. A. (1 990). “The Elderly Victim of Homicide:

An Application of the Routine Activities Approach.” Sociological Quarterly 3 1 :
307-3 19.

Mansfield, R., Gould, L. and Namenworth, J. (1 974). “A Socioeconomic Model

for the Prediction of Societal Rates of Property Theft.” Social Forces 52: 462-
472.

Massey, J. L., Krohn, M. D. & Bonati, L. M. (1 989). “Property Crime and the
Routine Activities of Individuals.” Journal of Research in Crime and
Delinquency 26: 378-400.

Mayhew, P., Clarke, R. V., & Hough, J. M. (1976). “Steering Column Locks and
Car Theft.” In Crime as Opportunity (Home Office Research Study No. 34), P.
Mayhew, R.V. Clarke, A. Struman, & J. M. Hough, (Eds.). London: HMSO.

Maxfield, M. (1987). “Lifestyles and Routine Activity Theories of Crime:

Empirical Studies of Victimization, Delinquency, and Offender Decision-
Making.” Journal of Qualitative Criminology 3:275-282.

McCaghy, C. H., Giordano, P. C. & Henson, T. K. (1 977). “Auto Theft Offender

and Offense Characteristics.” Criminology 15:367-385.
 Journal of Crime and Justice 145

Messner, S. F. & Blau, J. R. (1987). “Routine Leisure Activities and Rates of

Crime: A Macro-Level Analysis.” Social Forces 65: 1035-1052.

Messner, S.F. and Tardiff, K. (1985). “The Social Ecology of Urban Homicide:
An Application of the Routine Activities Approach.” Criminology 23:241-267.

Miethe, T. & Meier, R. (1990). “Opportunity, Choice, and Criminal

Victimization Rates: A Test of a Theoretical Model.” Journal of Research in
Downloaded by [Laurentian University] at 09:51 07 October 2014

Crime and Delinquency 27:243-266.

Miethe, T. & Meier, R. (1994). Crime and its Social Context: Toward an
Integrated Theory of Offenders, Victims, and Situations. Albany, NY: State
University of New York Press.

Miethe, T., Stafford, M. C. and Long, J. S. (1 987). “Social Differentiation in

Criminal Victimization: A Test of Routine ActivityflifestyleTheories.” American
Sociological Review 52: 184-194.

Miller, M. V. (1 987). “Vehicle Theft Along the Texas-Mexico Border.” Journal

of Borderland Studies 2: 12-32.

Morgan, K. 0. & Morgan, S. (1999). City Crime Rankings: Crime in

Metropolitan America 5th ed.. Lawrence, KS: Morgan Quint0 Press.

Mustaine, E. & Tewksbury, R. (1997). “Obstacles in the Assessment of Routine

Activities Theory.” Social Pathology 3: 177-194.

Mustaine, E. & Tewksbury, R. (1998). “Predicting Risks of Larceny Theft

Victimization: A Routine Activity Analysis Using Refined Lifestyle Measures.”
Criminology 36:829-857.

NRMA. (1990). Car Theft in New South Wales. Sydney: NRMA Insurance
Limited.

Rengert, G. (1 996). The Geography of Illegal Drugs. Boulder, CO: Westview.

Robinson, M. (1 999). “Lifestyles, Routine Activities, and Residential Burglary

Victimization.” Journal of Crime and Justice 22:27-56.

Robinson, M. & Robinson, C. (1 997). “Environmental Characteristics

Associated with Residential Burglaries of Student Apartment Complexes.”
Environment and Behavior 29:657-675.
 I46 Journal of Crime & Justice

Roncek, D. and Bell, R. (1981). “Bars, Blocks and Crime.” Journal of

Environmental Systems 1 1:35-47.

Roncek, D. & Faggiani, D. (1 985). “High Schools and Crime.” Sociological

Quarterly 64:491-505.

Roncek, D. W. & Maier, P. A. (1991). “Bars, Blocks, and Crime Revisited:

Linking the Theory of Routine Activities to the Empiricism of Hot Spots.”
Criminology 29: 725-753.
Downloaded by [Laurentian University] at 09:51 07 October 2014

Sampson, R. and Wooldredge, J. (1987). “Linking the Macro-and Micro-Level

Dimensions of Lifestyle-Routine Activity and Opportunity Models of Predatory
Victimization.” Journal of Quantitative Criminology 3:371-393.

Sherman, L. W., Gartin, P. R. & Buerger, M. E. (1989). “Hot Spots of Predatory

Crime: Routine Activities and the Criminology of Place.” Criminology 27:27-55.
Shichor, D., Decker, D., & O’Brien, R. (1980). “The Relationship of Criminal
Victimization, Police Per Capita and Population Density in Twenty-Six Cities.”
Journal of Criminal Justice 8:309-3 16.

Stahura, J. M. and Hollinger, R. C. (1988). “A Routine Activities Approach to

Suburban Arson Rates.” Sociological Spectrum 8:349-369.

Stahura, J. M. & Sloan, J. J. (1988). “Urban Stratification of Places, Routine

Activities and Suburban Crime Rates.” Social Forces 66:1102-1118.

U. S. Bureau of the Census. (1991). TZGER Questions and Answers. U.S.

Department of Commerce. Economic and Statistics Administration.

U S . Bureau of the Census. ( 1 990). Population and Housing Characteristics for

Census Tracts and Block Numbering Areas, Lafayette, Louisiana MSA.
Washington, DC: U.S. Government Printing Office.

Wilkins, L. (1964). Social Deviance. London: Tavistock.