Download as pdf or txt
Download as pdf or txt
You are on page 1of 28

Police Quarterly

http://pqx.sagepub.com/

Environmental Constraints and Organizational Outcomes: Modeling Communities of Municipal Police


Departments
Douglas R. Davenport
Police Quarterly 1999 2: 174
DOI: 10.1177/109861119900200203

The online version of this article can be found at:


http://pqx.sagepub.com/content/2/2/174

Published by:

http://www.sagepublications.com

On behalf of:
Police Executive Research Forum
Police Section of the Academy of Criminal Justice Sciences

Additional services and information for Police Quarterly can be found at:

Email Alerts: http://pqx.sagepub.com/cgi/alerts

Subscriptions: http://pqx.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://pqx.sagepub.com/content/2/2/174.refs.html

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


Dav en port / MU NIC I PAL PO LICE
PO LICE
DE PART
QUARMENTS
TERLY (Vol. 2, No. 2, June 1999)

ENVIRONMENTAL CONSTRAINTS
AND ORGANIZATIONAL OUTCOMES:
MODELING COMMUNITIES OF
MUNICIPAL POLICE DEPARTMENTS
DOUGLAS R. DAVENPORT
Truman State University

Strategic management argues that administrators can significantly affect


agency performance. This article examines the role of the organizational
environment in determining agency effectiveness, arguing that performance
is in part constrained by external conditions that cannot be altered by manag-
ers. Factor analysis is used to identify relevant community characteristics
and create a multidimensional model of the public sector environment that is
parsimonious and theoretically grounded. This model is used to examine the
performance of 170 Texas municipal police departments over an extended
time period. As expected, environmental characteristics produced significant
negative effects on performance and demonstrates the value of modeling the
agency environment in this way.

The drive to improve the performance of public agencies is an enduring


theme at all levels of government. In 1993, Vice President Gore (1993)
spearheaded the National Performance Review, which sought to guarantee
“effective, efficient, and responsive government,” moving to “make govern-
ment work better and cost less” (p. i). Behind this effort is the push to “rein-
vent government” (Osborne & Gaebler, 1992), which has become a mantra
for many municipal administrators.
Police agencies have long been a focal point of performance review,
ostensibly to maintain accountability and to ascertain effectiveness and/or
efficiency. This has largely resulted from the confluence of three forces: (a)
the significant, visible, and often controversial nature of the work; (b) the

POLICE QUARTERLY Vol. 2 No. 2, June 1999 174–200


© 2000 Sage Publications, Inc.

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


Davenport / MUNICIPAL POLICE DEPARTMENTS 175

emphasis on crime fighting by both the public and the police themselves;
and (c) the collection of vast quantities of crime-related data on a regular
basis. Together, these factors have made police departments obvious targets
for performance evaluations. Of course, this has not automatically resulted
in appropriate or accurate evaluations. However, this preoccupation with
performance demonstrates the salience of review from both an internal and
external perspective.
Of course, all this presumes that performance (no matter how it is mea-
sured) is amenable to agency manipulation. This is a foundational presump-
tion on the part of strategic management initiatives (broadly conceived),
which suggest that organizational rationality results in goal maximization
and efficiency. Strategic management of this sort is not new and encom-
passes the notions of management by objectives (Drucker, 1954), total per-
formance management (National Center for Productivity and Quality of
Working Life, 1978), transformational leadership (Tichy & Ulrich, 1984),
teamwork (Orsburn, Moran, Musselwhite, Zenger, & Perrin, 1996), and the
most current variation—total quality management (Denhardt, 1995).
Productivity improvement efforts for policing agencies have often
proven to be unfruitful (Wycoff, 1982). Previous studies have failed to con-
sistently demonstrate the ability of police departments to lower crime rates
(Allison, 1972; Decker, 1979; Jacob, Lineberry, Heinz, Rich, & Swank,
1982; Levine, 1975), raise clearance rates (Greenwood, Chaiken, &
Petersilia, 1977), or even improve public feelings of safety (Kelling, Pate,
Dieckman, & Brown, 1974). Indeed, publicized rhetoric in recent years
suggests that criminal justice leaders have recognized this, leading to a call
for partnerships between citizens and the police because the latter cannot
complete the task of crime prevention alone (Brown, 1992; Sessions, 1990).
Of course, this acknowledgment has not prevented public officials (includ-
ing law enforcement agency heads) from claiming credit for recent drops in
reported crime (Metzgar, 1996; Vader, 1996; Weiss, 1996).
If police agency efforts fail to provide adequate explanation for perfor-
mance, what alternatives exist? A viable avenue of research considers the
role of environmental characteristics in determining performance. Such
studies are founded on the tenets of contingency theory, whereby external
factors are critically important to the formation, structure, and success of
firms (for a broad overview, see Lawrence & Lorsch, 1967).

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


176 POLICE QUARTERLY (Vol. 2, No. 2, June 1999)

ORGANIZATIONAL ENVIRONMENTS
The importance of the environment has led various organizational theo-
rists and management researchers to conceptualize and measure the envi-
ronment. The environment has generally been assessed along three dimen-
sions: complexity, dynamism, and resource availability (Sharfman & Dean,
1991). These dimensions are common threads running through the re-
search, as Sharfman and Dean (1991) note,

Throughout this literature we have found discussion of (a) the degree to which the
number and sophistication of elements in the environment make understanding it
more difficult; (b) the stability/predictability of an environment; and (c) the level of
resources available in an environment, relative to the number of firms competing for
those resources. (p. 684)

This focus led Sharfman and Dean to create a model that incorporated a
number of variables that captured these three dimensions. However, this
work was applied to private sector firms, and the variables used by
Sharfman and Dean in factor analysis are not applicable to the public sector.
Of course, the importance of the environment to public agencies has been
recognized, and applications of the open systems perspective to public sec-
tor organizations have become increasingly numerous. However, studies
generally conceptualize it in a political sense (Denhardt, 1995; Fried, 1976;
Rosen, 1993). Some prescriptive works have pointed out other aspects but
provide little direction for measuring them.
It is important to recognize that most bureaucracies operate in an envi-
ronment that supercedes strictly political aspects. Recent developments in
the literature note that the openness of the environment requires large
degrees of interorganizational cooperation (Alter & Hage, 1993). At the
local level, this system includes significant numbers of private entities con-
nected vis-à-vis measures such as privatization (Savas, 1982) and the cre-
ation of enterprise zones.
Relevant to this study is the observation that researchers have noted the
salience of community complexity for organizational structures (Lang-
worthy, 1985; Weisheit, Falcone, & Wells, 1994), degrees of agency profes-
sionalization (Henderson, 1975), and organizational effectiveness (Cord-
ner, 1989). The influence of environmental turbulence has also been
observed and measured with some validity (Culbreth, 1993).

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


Davenport / MUNICIPAL POLICE DEPARTMENTS 177

Criminology literature has examined the role of the environment in crim-


inal activity via social ecology models. The historic study conducted by
Shaw and McKay (1942, 1969) in Chicago examined the role of social dis-
organization (via the dynamic nature of the urban setting) in juvenile delin-
quency patterns. Shaw and McKay noted three broad forces at work that are
relevant for our discussion: heterogeneity, socioeconomic condition, and
population mobility.
Further studies have also concluded that environmental or ecological
factors are related to crime. Smith and Jarjoura (1988) found that population
density and racial heterogeneity were significant predictors of violent crime
rates in neighborhoods. They also found that there was an interactive effect
where low income and mobility occurred together. In those neighborhoods,
crime rates were slightly (but significantly) higher.
Sampson (1985) found that population mobility had a significant effect
on victimization rates for violent crime. Sampson and Groves (1989) exam-
ined data from British communities and found that victimization rates for
mugging/street robbery were significantly affected by ethnic heterogeneity
and urbanization, along with several indicators of social disorganization
(local friendship networks, unsupervised peer groups, and lack of organiza-
tional participation). In addition, Sampson and Groves (1989) found socio-
economic status to be significantly related to incidents of burglary, auto
theft, and vandalism (p. 791).
From an organizational perspective, Zhao (1996) examined the relation-
ship between external (environmental) characteristics and the implementa-
tion of community-oriented policing. In support of a contingency perspec-
tive, Zhao found that several characteristics were significantly correlated
with organizational change. Internally focused change involved manage-
ment innovations, and Zhao found such change correlated with racial heter-
ogeneity (operationalized as non-White population), mobility, population
change, and city size. Externally focused change was defined to include
“the reorientation of police operations and crime prevention activities”
(Zhao, 1996, p. 43). This type of change was correlated with racial hetero-
geneity, city size, mobility, and two measures of socioeconomic status (pov-
erty and unemployment).
Thus, the importance of the environment is widely accepted for both pri-
vate sector firms and public agencies. Managerial studies have attempted to
conceptualize and measure the environment along these three primary
dimensions, although researchers have not reached an agreement on the

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


178 POLICE QUARTERLY (Vol. 2, No. 2, June 1999)

operationalization or even the concepts behind those measures (Dess &


Rasheed, 1991; Sharfman & Dean, 1991). Criminological and police orga-
nization studies have also found these concepts (operationalized through
numerous variables) to be significant.

POLICE EFFECTIVENESS
A number of studies of police agencies have considered variables that are
external to the organization yet appear to play an important role in effective-
ness. Wellford (1974) found that crime rates are significantly affected by
socioeconomic variables. In contrast, social control variables (number and
rate of police, ratio of civilian personnel to sworn, and two budgetary mea-
sures) failed to explain a significant proportion of the variation.
Pogue (1975) conducted a statistical analysis to assess the effects of cer-
tain environmental characteristics and police expenditures on crime rates
and clearance rates. Pogue used 1968 data from 66 standard metropolitan
statistical areas (SMSAs) to assess clearance rates and used both linear and
log-linear equations. His research determined that both local- and state-
level police expenditures had no significant effect on clearance rates. One
environmental variable, population density, was significantly related to
clearance rate, showing a negative effect.
Pogue’s (1975) work is important to this research because he recognized
the influence of environmental factors and attempted to assess their effects.
Furthermore, he observed some degree of influence of community com-
plexity on effectiveness but did not comment on this finding in his conclu-
sion. Unfortunately, Pogue aggregated departmental statistics at the SMSA
level, making local variation impossible to assess.
A study conducted by Cordner (1989) is also important to this research
agenda. Cordner assessed the relationship between agency size, environ-
mental characteristics, and clearance rates. Cordner used the number of
sworn officers to represent agency size. Environmental characteristics were
region (area of the state), number of index crimes per sworn officer, and the
proportion of property crimes to total index crimes.
In conducting regression analysis on 1986 data for 84 police agencies in
the state of Maryland, Cordner (1989) found that region accounted for
almost all of the explained variance (p. 151). Nonmetropolitan status was
positively associated with clearance rates and was statistically significant
(beta = .54, p < .01). In addition, the portion of property crimes variable was

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


Davenport / MUNICIPAL POLICE DEPARTMENTS 179

statistically significant, although the additional explanatory power was


small (beta = –.19, p < .05). Agency size was positively associated with
clearance rates but was not a significant predictor. Likewise, workload
(number of index crimes per officer) was in the expected (negative) direc-
tion but was not significant.
The weakness of Cordner’s (1989) analysis is the failure to adequately
explore the environmental factors captured by the region variable. In
essence, the region variable allowed Cordner to isolate the Washington, DC
and Baltimore metropolitan areas through dummy variables, where clear-
ance rates were significantly lower and most likely subject to greater varia-
tion at the department level.
Overall, these results beg further exploration. The salience of the organi-
zational environment is relatively obvious. However, studies either have
failed to adequately measure multiple factors or have not used them in a
comprehensive multivariate analysis. This problem is exacerbated by the
failure to consider the interrelationships between the variables.1 Rather, the
studies focus on one or two variables that are ostensibly related to police
performance, using them to contrast with the lack of explanatory power of
departmental characteristics. Of importance to this work is the fact that both
Pogue (1975) and Cordner (1989) found environmental variables to be sig-
nificantly related to performance, measured as clearance rates.
Thus, it appears that two interrelated gaps remain to be filled, despite
attention on the part of researchers. First, the public sector environment has
not been adequately modeled in a parsimonious way. Studies that examine
the relevant variables fail to consider their interactive relationship. Second,
it remains to be seen whether a more inclusive model of the police agency
environment can account for any variance in performance. This work
attempts to address these gaps through the creation of a multidimensional
model of the environment and the application of that model to a critical
police performance measure—the clearance rate.

RESEARCH HYPOTHESES
Three general questions are considered in this study and are examined
through two hypotheses. First, can the public sector environment be
reduced to a manageable number of theoretically sound measures? The first
research hypothesis asserts that it can be: Environmental variables salient to
the public sector represent underlying dimensions that can be discovered.

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


180 POLICE QUARTERLY (Vol. 2, No. 2, June 1999)

This hypothesis is largely untested in previous work but is theoretically sup-


ported by both the business and criminology literature noted above.
The second and third general questions are interrelated. The second
question asks, Does the environment matter for police agency perfor-
mance? If so, the third question is important: In what ways does it matter? A
second research hypothesis seeks to answer both questions by stating the
following: Environmental characteristics are significantly and negatively
related to clearance rates.
This second hypothesis is generally supported by previous work. Cord-
ner (1989) and Weisheit, Falcone, and Wells (1994) noted that increased
community complexity led to decreased clearance rates. Population density
and racial heterogeneity have also been identified as having negative effects
on performance (Crank, 1990; Pogue, 1975). Furthermore, city size has
been noted in studies of investigative effectiveness, which has bearing on
clearance rates (Willmer, 1970).
Community turbulence appears to be a negative influence on agency per-
formance. When viewed from the perspective of police agencies, turbu-
lence means that social integration within the community is lower. This
affects the “density of acquaintanceship” (Freudenburg, 1986; Weisheit,
Falcone, & Wells, 1995, p. 26), or the extent to which people in the commu-
nity know one another. Thus, criminal activity may be ignored when it
involves strangers. This desire to avoid involvement exacerbates the prob-
lem facing the police of locating and identifying witnesses. Furthermore,
citizens may not recognize criminal activity because they do not know who
belongs in the neighborhood and who does not. This delays the reporting of
crime and decreases the likelihood of apprehension.
Finally, lower levels of resources may result in decreased agency capac-
ity. Of course, the relationship is indirect at best. The fact of this indirect
relationship may prove to be a difficult obstacle to overcome in the statisti-
cal analysis, particularly in terms of finding a significant relationship. How-
ever, it is intuitive that decreased resource capacity will have a negative
influence on performance.

METHODOLOGY AND DATA


This study uses data on Texas municipal police agencies in communities
of varying size and geographic distribution. The research uses a cross-sec-
tional design in a multivariate regression analysis, whereby the importance
of environmental characteristics with regard to performance can be

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


Davenport / MUNICIPAL POLICE DEPARTMENTS 181

assessed. Measures of environmental characteristics are factor analyzed to


determine whether the dimensions of complexity, resource availability, and
dynamism can be captured and measured in a parsimonious fashion. Mea-
sures of crime, which are important to this organizational set, are also
included in the factor analysis and are expected to produce an additional
environmental dimension. Factor scores are then used as independent vari-
ables in a multiple regression analysis of performance.
The performance of these departments is measured via their clearance
rates, calculated as the average rate for the period from 1980 to 1989. This
measure is designed to capture the general ability of departments to manage
crime during a 10-year period. Year-to-year variation has been observed in
pooled cross-sectional analyses of clearance data for Texas agencies (Dav-
enport, 1995). This variation appears to be due to stochastic processes rather
than any departmental efforts.2 Although the variation is not large in the
aggregate, it may be significant for any number of agencies and may signifi-
cantly affect the results of regression analyses for this study. Therefore, a
10-year average is computed to help ensure a more accurate portrait of
crime management efforts for each agency.
This time frame was also selected to coincide with the available data on
the independent variables. Because most of this data was compiled from
Census Bureau sources, information is not available for all variables beyond
1990. Thus, the question of temporal priority requires that the dependent
variable be measured in concert with the independent variables.3
I also separate the clearance rate into the components of violent crimes
versus property crimes for the time frame. Because clearance rates vary sig-
nificantly for these components, a separate analysis will provide additional
perspectives with regard to environmental effects, which may be distinct for
the two variables. This multidimensional portrait will therefore strengthen
any conclusions about the model.
Ideally, one should examine clearance rates for each of the seven catego-
ries of index offenses. However, smaller cities exhibit great variability in
clearance statistics due to low levels of crime. It is not uncommon for such
communities to report less than 100 index crimes for an entire year; when
this is the case, clearance rates can become quite volatile. If only one rape
were reported, the clearance rate for that crime might be 0% or 100%,
depending on whether the case was solved or not. In such instances, ratios
become meaningless and cannot be appropriately compared with other sta-
tistics. Because the majority of agencies in this data set operate in communi-
ties with populations less than 25,000, I have chosen to limit the breakdown

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


182 POLICE QUARTERLY (Vol. 2, No. 2, June 1999)

of clearance data into the two general categories of property and violent
offenses.4 The property crime category includes the offenses classified as
burglary, larceny theft, and motor vehicle theft. The violent crime category
includes murder and nonnegligent manslaughter, forcible rape, robbery,
and aggravated assault.
This study uses general purpose police departments–report uniform
crime report (UCR) data sent to the Texas Department of Public Safety.
Thus, it does not include county sheriff’s agencies or special purpose agen-
cies (such as school district or housing authority law enforcement entities)
that report UCR data. The study is also limited to municipalities with a pop-
ulation of 10,000 or greater in the 1990 census.
The selection of agencies serving municipalities of 10,000 or greater
may seem to be an unfortunate limitation. However, these 170 departments
accounted for 87.2% of all full-time law enforcement personnel working for
Texas municipal agencies in 1993 (Federal Bureau of Investigation, 1994).5
Furthermore, these departments reported 80.2% of all Part 1 index offenses
reported in Texas for the year 1994 (compiled from UCR data, FBI, 1995).

INDEPENDENT VARIABLES
A number of environmental variables are included in this study so that
the multiple dimensions of complexity, turbulence, and resource capacity
may be tapped via factor analysis. First of all, patent indicators of complex-
ity include urbanism, density, population, number of law enforcement
agencies in the county, foreign-born population, and minority population.
Urbanism is measured as the percentage of the county that is classified as
urban by the Census Bureau (Crank & Wells, 1991). Density is the popula-
tion per square mile (Pogue, 1975). Population is the municipal population
as measured by the Census Bureau. Agency is operationalized as the number
of municipal and county law enforcement agencies in the county. This mea-
sure is directly related to the complexity of the organizational environment
and is particularly relevant for police departments.
Minority status and foreign-born population tap the cultural and racial
diversity of a community. The former concept is operationalized as percent-
age African American and percentage Hispanic.6 The latter concept is
operationalized as the percentage of the population that was foreign born.
This appears to be quite relevant for Texas, where immigration from Central
America has been a significant issue. All of the demographic variables
taken from census data reflect 1990 information.

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


Davenport / MUNICIPAL POLICE DEPARTMENTS 183

The dimension of resource availability will be measured with a number


of variables (Pogue, 1975). Unemployment is the percentage reported as
unemployed for the municipality for the year 1990. Per capita income is the
personal income reported for 1989. Poverty is the percentage of the 1990
population that is at or below the poverty line.
The dimension of dynamism will be assessed through four variables.
Population change is the percentage change in population from 1980 to
1990. Density change is the percentage change in the population density of
the municipality from 1980 to 1990. Urbanism change is the percentage
change in the level of urbanism from 1980 to 1990. Mobility is the percent-
age of the population that has moved during the previous 5 years. All of
these measures attempt to capture the notion of change over time in sum-
mary fashion for the communities examined.
In addition, two measures of crime were included with the environmental
variables in this research: Crime rate is the amount of Part 1 index crime per
100,000 population, measured as the mean rate for the period from 1980 to
1989. Property is the proportion of Part 1 index crimes that are classified as
burglary, larceny, or auto theft. Both variables appear to be important to
departmental effectiveness (Cordner, 1989).

DEPENDENT VARIABLE
As discussed previously, a number of variables are available that assess
some aspect of the performance of police departments. During the past two
decades, researchers have moved away from examinations of crime rates or
clearance rates. As for reported crime rates, scholars have argued that the re-
lationship between the inputs (police) and the ultimate goal (reduction of
crime) is indirect (Wycoff, 1982). Instead, Wycoff (1982, p. 25) suggests
that intermediate goals be considered, such as crime management. Wycoff
(1982) describes it the following way:

The concept of crime management acknowledges that crime will occur despite the
most vigorous and effective police efforts at crime prevention. Given the existence of
crime, police can act to reduce its social costs by working to contain it within certain
geographic areas, by working to reduce the amount of physical injury or property loss
or damage, by providing assistance to victims (to include the recover of property or
assistance in acquiring compensation for damage or loss), and by providing reassur-
ance to the public.

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


184 POLICE QUARTERLY (Vol. 2, No. 2, June 1999)

Thus, although reduction of crime is still assessed at many levels, re-


search has also focused on public perceptions of police performance
(Brandl & Horvath, 1991), reduction of fear of crime, internal measures of
efficiency (Skogan, 1976), and the behavior of individual officers (Allen &
Maxfield, 1983).
Based on Wycoff’s input-output problem, one might consider clearance
data as a viable crime management measure, one which is more reflective of
an intermediate goal directly affected by police efforts. However, because
the RAND study of investigatory activities appeared (Greenwood et al.,
1977), researchers have generally ignored clearance statistics as a perfor-
mance measure. Several problems are noted, with the following issues
being most critical.
At the very heart of this matter is measurement reliability, or the consis-
tency of the measuring device (Spector, 1981). In the case of clearance sta-
tistics, this has been discussed as a cross-sectional issue. Departmental
clearance rates, it is argued, are not comparable due to variations in report-
ing practices and the possibility of reporting error on the part of agencies (in
general, see Greenwood, 1970; Greenwood et al., 1977; Hatry, 1975;
Schneider & Wiersema, 1990; Wycoff, 1982).
On a more serious note, clearance statistics have been critiqued as being
subject to politically motivated manipulation (distortion). Because UCR
data is self-reported, there is always the opportunity for departments to
intentionally overstate or misstate clearances. The lack of consistent exter-
nal monitoring of all data submitted to the FBI also allows for error to bias
the measures.7 In general, questions of reliability revolve around excep-
tional clearances, which use some exception to the arrest rule. Such clear-
ances must be determined by the department; it is possible that an agency
might simply indicate more offenses cleared than actually met the FBI
guidelines.
Whereas these concerns are valid, they are not conclusive. Several argu-
ments can be made to support the use of clearance statistics for evaluation
purposes. First, the role orientation of individual officers generally focuses
on crime fighting (Reuss-Ianni, 1983). Felony arrests (for patrol officers)
and solving cases (for investigators) represent key goals in daily operations.
Second, UCR data is available that covers an extended time period and has
been collected in a similar fashion. Thus, clearance statistics represent one
of the few data sources available for extended analysis of law enforcement
agencies and their performance. Concomitantly, trend analysis and aggre-

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


Davenport / MUNICIPAL POLICE DEPARTMENTS 185

gated over time measures serve to smooth out year-to-year variations.


Finally, the nonrandom variation in reporting practices discussed above
may not be as significant as suggested. More sustained attention is in order
with regard to this final issue.
As part of a larger study, the 170 departments examined here were mailed
a survey requesting information on organizational structure, mission, func-
tions, internal performance measures, and expenditures. Of the 170 surveys
mailed to police chiefs, 78 were returned, for a 45.9% response rate.8 Sev-
eral questions about clearance statistics were included and are important for
this discussion.
One question asked, “What are your departmental requirements for UCR
clearance of Part One crimes (other than published FBI guidelines)?”
Another question then asked, “Have these requirements changed during the
period 1980-1993 (yes/no)? If so, how have they changed?” In all, 69 agen-
cies provided a response to the departmental requirement question. Over-
whelmingly, these departments stated that they strictly followed the FBI
guidelines; none of the agencies indicated a requirement at odds with these
guidelines. Three of the departments indicated their desire to increase clear-
ance rates in responding to this question. None stated how they planned to
accomplish this, nor did they indicate a more liberal exceptional clearance
criterion.
As for the change in requirement question, 70 agencies responded. Only
two indicated that the methodology had changed; one stated that there had
been a “personnel change in records,” whereas another stated that the
department had been “improperly reporting sexual assaults.” This latter
response does not appear related to clearance data but rather to coding of
criminal incidents. In addition, 3 of the 70 agencies answering the question
indicated that they did not know whether any changes had occurred during
the period.
These responses suggest that departments attempt to follow the pub-
lished guidelines on the clearing of crimes, without additional criteria. Of
course, this does not rule out the possibility of error, either in coding or in
judgment. Such errors may not be random (if clerical staff routinely fail to
follow guidelines), but the likelihood of this appears to be minimal. The
stated desire to follow FBI guidelines suggests that administrative person-
nel will likely check the monthly reports for errors and provide adequate
training in the correct procedure to minimize such errors.

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


186 POLICE QUARTERLY (Vol. 2, No. 2, June 1999)

The alternative to erroneous data is that departments deliberately manip-


ulate clearance statistics to appear more effective in managing crime.
Although individual examples of this do occur, it does not imply that all or
even a large number of departments do so. Furthermore, the following ques-
tion remains to be answered: If clearance data are subject to frequent manip-
ulation, why do the rates remain so stable? As Cordner (1989) correctly
noted, “Despite these sources of error, aggregate clearance rates for the
entire country have been remarkably consistent from year to year” (p. 147).
He goes on to point out that the rate has been roughly 20% for many years.
In support of this observation, the overall clearance rate in the United States
for 1993 was 21.1% (FBI, 1994, p. 208).
One additional concern must be addressed here. This study of organiza-
tional performance focuses on the attainment of goals. It is arguable that
clearance rates may not reflect an organizational commitment, but only an
institutional pattern. If police departments are not actively pursuing
improved clearance rates as an organizational goal, then empirical findings
may not be conceptually valid (Spector, 1981). Thus, researchers may be
using the wrong measure to ask an appropriate question about performance.
I tested whether clearance statistics are of importance to police departments
through two questions included on the survey.
One question asked, “What criteria are used to evaluate your depart-
ment’s overall performance?” Respondents were presented with four op-
tions: number of crimes (or rate) committed, community satisfaction, num-
ber of crimes cleared (or rate) for UCR, and other. Additional space was
provided to describe the other criteria used. Of the 78 responding depart-
ments, 66 (84.6%) indicated that clearance data were evaluation criteria.
A second question asked respondents to rate the importance of UCR
clearances in terms of their departmental mission on a scale ranging from 0
(not important) to 4 (most important). Of the respondents, 50% indicated
that clearances were “important” to their mission, whereas another 33%
characterized them as “very important” (26.3%) or “most important”
(6.6%). Together, the responses to these two questions clearly indicate that
clearance statistics are quite important to municipal police agencies. It is
reasonable to conclude that clearance data are both important to individual
agencies and probably more reliable than previously thought. I have
attempted to address the methodological concerns raised by scholars and
found that the survey results support this more benign perspective. I turn
now to the statistical analysis of environmental characteristics.

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


Davenport / MUNICIPAL POLICE DEPARTMENTS 187

MODEL DEVELOPMENT
The purpose of factor analysis is to discover the underlying dimensions
that connect interrelated variables (Kim & Mueller, 1978; Norusis, 1992).
Factor analysis is useful as a data reduction technique (Miller & Whitehead,
1996), particularly when source variables are highly correlated. An exami-
nation of the zero-order correlations suggests that factor analysis is an
appropriate methodology for this data set.
Table 1 presents the results of factor analysis using the environmental
characteristics discussed above.9 The results in Table 1 were obtained using
a varimax (orthogonal) rotation with principal components analysis for
extraction (Norusis, 1992, p. 59). This particular algorithm is appropriate
when seeking to obtain factors that are uncorrelated, and it provides a sim-
ple structure (Norusis, 1994).
Four factors were extracted from these variables, which generally corre-
spond to the three dimensions discussed by Sharfman and Dean (1991),
along with the added dimension of violent crime. First of all, environmental
complexity is tapped via the variables of urbanism, number of law enforce-
ment agencies in the county, and population density. Population, which is
often used as a surrogate for this concept, shows only modest loading on this
factor. Thus, it seems clear that complexity is more than a function of popu-
lation. Per capita income also loads to some degree on this factor, suggest-
ing that economic and class distinctions are evident in heterogeneous
communities.
The dimension of resource capacity is observed through the variables
Hispanic population, unemployment, poverty, per capita income, and per-
cent foreign born.10 The fact that two of the minority variables load heavily
on this factor is not particularly surprising, based on the linkages between
poverty, unemployment, Hispanic population, and foreign-born popula-
tion. In Texas, this appears to be something of a regional characteristic, one
observed most often in the border counties near Mexico.11 Clearly, Hispan-
ics in these cities are disproportionately poor and unemployed.
The variables, mobility, population change, density change, and urban-
ism, change load appropriately on the environmental turbulence factor. As
Table 1 also shows, the resource variables that suggest a lack of capacity
(primarily unemployment and poverty) have negative loadings on the turbu-
lence factor. This indicates that turbulence may not have a particularly nega-
tive influence on communities.

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


188 POLICE QUARTERLY (Vol. 2, No. 2, June 1999)

TABLE 1. Rotated Factor Matrix for Environmental Variables

Variable Complexity Resources Turbulence Violent Crime

Density .773 .193 –.246 –.186


Law enforcement agencies .765 –.219 .303 .076
Urbanism .748 –.061 .271 .047
Per capita income .493 –.650 .206 –.189
Population .428 .104 .017 .634
Foreign born .344 .820 .065 –.006
Hispanic .092 .939 –.114 –.105
Unemployment –.188 .849 –.281 .110
Poverty –.297 .822 –.308 .170
Population change .087 –.158 .900 –.171
Mobility –.018 –.390 .631 .165
Density change .281 –.217 .740 –.140
Urbanism change .043 .016 .571 –.040
African American –.287 –.261 –.161 .655
Property crime percentage .314 –.203 .175 –.549
Crime rate .086 .080 .023 .807
Eigenvalue 2.695 5.088 1.873 1.442
Percentage of variance 16.8 31.8 11.7 9.0
Cumulative percentage 16.8 48.6 60.4 69.4

Where are these changes in environmental conditions most obvious?


Based on our understanding of historic municipal growth patterns (Ross,
Levine, & Stedman, 1991), one might suspect that suburban communities
would be most significantly affected by change. To test this expectation,
zero-order correlations were run for these turbulence variables in conjunc-
tion with suburban status.12 Three of the change variables (population, den-
sity, and urbanism change) are positively associated with suburban status
(.40, .35, and .35, respectively). All of these correlations were significant at
the .001 level.
Further correlations with suburban status are also illustrative at this
point. Per capita income is significantly correlated with suburban status
(.39), whereas unemployment and poverty are negatively correlated with
the suburbs (–.37 and –.52, respectively, significant at .001 level). Suburban
status was also found to be negatively correlated with the average crime rate
(–.30, significant at .001 level).
It seems appropriate to suggest that turbulence (as related to suburban
status) is a reflection of an influx of upwardly mobile residents. For those
individuals seeking a higher quality of life, suburban communities are espe-
cially attractive. This quality-of-life concern is characterized by the search

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


Davenport / MUNICIPAL POLICE DEPARTMENTS 189

for decreased fear of crime, expanded housing options, relatively newer


public sector infrastructures, and fewer nuisance industries.
Not coincidentally, the suburbs are portrayed as providing these very
things. Indeed, Danielson and Doig (1982) note that suburbs have used zon-
ing to prevent the construction of high-density housing and limit the influx
of heavy industries. Such moves effectively prevent the addition of many
lower income residents and ensure increased property values for those who
can afford to purchase housing. Thus, turbulence may be characterized as a
relatively positive force in the suburbs, where it has been most influential.
A final factor was extracted by the analysis, which is labeled violent
crime. The average crime rate, percentage African American population,
population, and percentage of index crimes that are property related load
heavily on this factor. Several observations are in order with regard to this
final factor.
First, the property crime percentage variable loads negatively. This indi-
cates that the factor predominately captures the aspects of violent crime. To
put it another way, cities that score high on this factor face higher levels of
crimes against persons, such as murder, rape, robbery, and assault.
Second, where violent crime is greater, the overall crime rate is higher.
Based on the factor loadings, it is clear that per capita crime is also higher in
the presence of greater violence. This may suggest a corollary to the broken
windows syndrome (Wilson & Kelling, 1982). Not only does disorder pro-
duce higher levels of crime indirectly but greater levels of violent crime are
also associated with increased rates of offenses generally. The relationship
is no doubt more complicated than this, yet it is evident from these findings.
Third, the relationship between crime rates and population indicated by
the factor loadings is intuitive. However, one must be careful not to see vio-
lent crime as simply the effect of big-city population. Descriptive analysis
suggests that crime is not merely a central city phenomenon but is signifi-
cant in communities of various sizes. A plot of the violent crime factor by
city showed that both Dallas and Houston scored high (2.68 and 2.79,
respectively), which is as expected. However, a number of other (smaller)
cities also scored relatively high, including Greenville (1.82), Galveston
(1.73), Paris (1.60), and Marshall (1.51). As the reader may know, these cit-
ies are concentrated in the East Texas region, suggesting that there is some-
thing of a regional effect at work.
Furthermore, some larger cities scored lower on the violent crime factor,
including San Antonio (.68), Austin (.52), El Paso (.14), and Corpus Christi
(–.01). Thus, although crime remains a salient issue in large cities, it is clear

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


190 POLICE QUARTERLY (Vol. 2, No. 2, June 1999)

that higher crime rates are not exclusively associated with large population
centers.
Finally, the linkage between relatively higher levels of violent crime and
African American population is consistent with other data. Walker (1994, p. 7)
notes that virtually all murders during a one-year period in Washington, DC
occurred in the poor African American neighborhoods on the eastside. Only
a few took place in the northwest sector, which is predominately White.
Although murder is only a small proportion of all violent crime that occurs,
this data indicate that the Texas statistics are not anomalous.13
The extraction of this fourth factor is not surprising because crime vari-
ables are included in the analysis. Based on the connection between the
crime factor and population, it may be appropriate to view the violent crime
factor as an additional component of environmental complexity. It is clear
that crime is interwoven with these other characteristics, significantly influ-
encing life within these communities.
As expected, these results illustrate the multifaceted nature of the
responding communities. They support the findings of Crank and Wells
(1991) with regard to the linkage between urbanism and heterogeneity, as
well as the more general observations of Cordner (1989) with regard to the
distinctions between metropolitan and rural communities.
The close parallel between these findings and the Sharfman and Dean
(1991) model is particularly noteworthy, considering the use of general
demographic and socioeconomic data rather than private sector business
and industry variables. Clearly, the results demonstrate the utility of a more
general model of the public sector environment for analyses. Based on these
results, the first hypothesis is supported. In the next section, this model is
used in an attempt to explain mean clearance rates of the 170 cities.14

ENVIRONMENTAL FACTORS AND PERFORMANCE


The second hypothesis asserts that environmental characteristics are sig-
nificantly and negatively related to clearance rates. Tests for this hypothesis
were conducted via multiple regression analysis. Four environmental factor
variables were created, and scores for each department were added to the
data set. Table 2 presents the results of equations for each of the three perfor-
mance measures. As this table shows, the second research hypothesis is
clearly supported for three of the four variables.
Several observations are salient to the discussion of performance. First,
community complexity produces a statistically significant negative influ-

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


Davenport / MUNICIPAL POLICE DEPARTMENTS 191

TABLE 2. Regression Results for 10-Year Performance Averages

Clearance Rate

Overall Violent Crime Property Crime


Variable Standardized Beta t Standardized Beta t Standardized Beta t

Complexity –.31 –4.17* –.38 –5.27* –.25 –3.28*


Resources –.04 .57 –.03 .35 –.02 .26
Turbulence –.18 –2.50** –.06 –.77 –.17 –2.30**
Violent crime .03 –.41 –.18 –2.52** –.08 .30
2
Adjusted R .109 .158 .074
F 6.03* 8.73* 4.31*

Note: Number of cases = 165.


*p < .01. **p < .05.

ence on performance. Where complexity is greater, clearance rates are


lower. This effect is observable for all three equations and provides most
of the explanatory power. These results suggest that environmental com-
plexity constrains the ability of departments to positively affect their
performance.
Where is complexity most observable? When cities are categorized by
population, those over 500,000 averaged 2.13 on this factor score. Because
this is a standardized variable, it means that the four cities in this population
group averaged just over 2 standard deviations above the mean. In contrast,
cities with populations between 10,000 and 25,000 (98 cities) averaged –.23
on this factor.
Of course, it would be tempting to suggest that complexity is little more
than a surrogate measure for population. However, the reader is reminded
that population actually loaded more heavily on the violent crime factor.
Furthermore, population does not adequately explain performance. Substi-
tuting population for the complexity variable in the overall clearance rate
equation produced a nonsignificant beta and a much poorer performing
model (adjusted R 2 = .03). Because nonlinear effects are possible when
using a measure such as population, the model was also tested through the
use of a categorical variable for city size. This variable (and the overall
model) exhibited properties similar to the population variable. Thus, com-
plexity is something distinct from population and is a significant predictor
of performance.
Second, turbulence also has a statistically significant negative influence
in two of the three equations. It is significant for property crime and for the

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


192 POLICE QUARTERLY (Vol. 2, No. 2, June 1999)

overall clearance rate model. The fact that the influence is negative seems
logical but somewhat at odds with the previous discussion about suburban
communities. However, these results are theoretically sound.
In his discussion of investigative work, Willmer (1970) indicated that
mobility of the population will have a negative effect on the availability of
information coming from the community. As residents move, the level of
social integration declines. Thus, individuals no longer know their neigh-
bors and do not generally forge close relationships with these strangers.
Likewise, those who are new to the community may be less likely to cooper-
ate with the police, particularly if they have not developed a sense of trust
with regard to the local agency.
The end result of this condition is that lower profile cases (such as most
property offenses) are less likely to be solved. Potential witnesses are less
likely to get involved with such cases, particularly when they do not know
their neighbors or do not consider the crime to be either personally threaten-
ing or serious. This also provides qualified support for the circumstance-
result hypothesis (Eck, 1984), which states that most crimes are solved by
virtue of readily identifiable witnesses rather than investigative effort. I
would qualify that hypothesis to state that routine property crimes are gen-
erally solved because of circumstances present at the time of the first con-
tact with officers rather than follow-up investigations.15 The fact that this
variable is not significant for violent crimes bears out this conclusion.
Because property crimes account for the vast majority of all index offenses,
mobility of the population also has a significant, negative influence on per-
formance in general.
Turbulence may also be a negative factor due to the dramatic increases in
population, resulting in significant shifts (growth) in service demands.
Police departments are generally unable to respond in a timely fashion to
increases in population. Inability to secure qualified applicants, coupled
with the retrenchment and cutbacks of the late 1970s, means that many
departments in the 1980s were unable to keep up with service demands.
Third, the violent crime factor is statistically significant for violent
crimes, producing a slight negative effect on performance. Thus, where a
higher proportion of the crimes are violent, solving violent crimes is ham-
pered. These results appear to be somewhat at odds with the findings of
Cordner (1989), who observed that a higher proportion of property crime
was a negative influence on performance. Cordner’s finding is intuitive
because property crimes are typically solved at a much lower rate, but the
present results suggest a more complex picture.

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


Davenport / MUNICIPAL POLICE DEPARTMENTS 193

The violent crime factor is characterized by both increased levels of vio-


lent crime and increased rates of overall crime. Thus, this finding can be dis-
tinguished from that of Cordner’s (1989), yet it is not completely contradic-
tory. All other things being equal, increased property crime proportions
reduce investigative effectiveness. However, in the presence of increased
crime rates, increased levels of violent crimes have a similar effect. This
finding complements previous studies, as noted below.
Qualitative studies of investigative efforts have noted that increased case-
loads for detectives result in their having less time to spend on each case
(Waegel, 1981). Those cases that are categorized as routine receive less
effort, whereas nonroutine cases will receive a disproportionately greater
share of the detective’s time (Greenwood et al., 1977; Waegel, 1996). Under
the circumstances created by higher levels of violent crime, even more solv-
able cases (i.e., violent crimes) may receive less investigative effort.
Clearly, increased caseloads become more problematic when making case
management decisions. It seems that these are instances where strategic
management initiatives might prove beneficial. However, the empirical evi-
dence observed in this study does not reveal the positive effects of any such
initiatives.
Finally, the resource capacity variable does not add any explanatory
power to the overall model and is not statistically significant. Clearly, the
relationship between capacity and clearance rates is not direct. Capacity
must be translated into departmental expenditures, which will be influenced
by internal spending priorities (such as capital versus operating expendi-
tures) and any number of additional variables. Thus, the failure to demon-
strate a significant relationship is not unanticipated.
Despite the statistically significant results, these findings must be inter-
preted with caution. Overall, none of the models explain clearance rates to a
high degree. Of course, it is possible that the relationship is nonlinear or that
foundational assumptions of least squares regression were violated. How-
ever, regression diagnostics failed to reveal substantive methodological
problems.16 Although the overall performance of these models is quite mod-
est, the research goal is not merely to explain clearance rates. Rather, it is to
identify and explicate the role that contextual factors play in constraining
performance. From that perspective, these results are promising. Of course,
it is important that additional work be done to replicate the study in other
regions so as to determine the ultimate utility of similar models. Because
regression results based on this number of cases may be unstable, it is

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


194 POLICE QUARTERLY (Vol. 2, No. 2, June 1999)

recommended that future efforts attempting to replicate these findings use a


larger sample size.

CONCLUSION
This study has demonstrated that the public sector environment, particu-
larly that of municipal police departments, can be captured with a theoreti-
cally based, empirically supported model. Furthermore, this study has
shown that the environment has a statistically significant and negative influ-
ence on the performance of those police agencies. Several additional com-
ments are important to this discussion.
First, these findings indicate that public agencies are constrained in their
quest to significantly alter their performance. Despite the value placed on
clearance statistics by agencies (and individual officers), environmental
conditions will hamper efforts at performance improvement. Therefore,
managers are encouraged to find the areas where marginal (yet significant)
gains can be made and capitalize on them. As noted above, effective man-
agement of more solvable cases may yield valuable results. This may also
include motivating patrol officers to be as thorough as possible at the initial
stages of the investigation, when the most critical pieces of information are
often most readily available. Procuring cooperation between patrol and
investigative bureaus is (ironically) more difficult than might be suspected.
However, it is an essential key to increasing performance in this vital area.
Second, police managers must continue to recognize the role of the
community in their performance. The philosophical tenets of community-
oriented policing suggest the creation of partnerships with the community
and identification of underlying problems (Manning, 1984; Trojanowicz &
Bucqueroux, 1990). Not only should managers see this as a supplement to
traditional police practices, but they must also recognize that effective part-
nerships are critical to long-term goals. This requires that they become
change agents, assisting the community in problem identification and solu-
tion creation. As this occurs, performance will be positively affected—
regardless of how it is measured.
Because of the nature of community complexity, it is important that
police agencies foster stronger ties with other area law enforcement entities
from all levels of government. Integrating data, networking, creating proj-
ect teams, and focusing on area repeat offenders are examples of this.
Although task forces are often created for narcotics operations and for high-

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


Davenport / MUNICIPAL POLICE DEPARTMENTS 195

profile murder cases, such cooperative ventures should be extended to other


crimes and should include more broad-based linkages. Such teams should
not be considered elite units; they should be common place and strategically
formed to focus on a variety of areawide problems.
Finally, although agency survey data suggest that clearance statistics are
very important internal indicators of success, police managers should not
view clearance of crimes as an ultimate goal. Although clearances are a nec-
essary condition for success, they are not sufficient. Among other things,
the issue of case attrition (Petersilia, Abrahamse, & Wilson, 1990) should
also be addressed. Removing agency tunnel vision and reconnecting the
institutions and bureaucracies that animate the justice system are important.
It is critical that police managers foster a more integrative perspective, forg-
ing effective ties with prosecutors to ensure that cases are adequately pre-
pared for trial. Although such actions rarely affect clearance rates, they do
affect the perceptions of the public, who view the justice system as lacking
in cooperation and responsible for such breakdowns.
What has been presented is designed to be illustrative rather than exhaus-
tive. Clearance rates are only one aspect of agency performance, and munic-
ipal police departments are only one form of public bureaucracy. However,
this study makes it clear that police performance is hampered by numerous
obstacles. One obstacle that deserves more adequate focus is the nature of
the community environment. A realistic, larger perspective will help man-
agers determine the most effective ways of improving performance, a
long-standing and elusive goal.

NOTES
1. Zhao (1996) noted that extreme levels of multicollineary existed between the various
environmental variables. Rather than using a data reduction technique, he chose to conduct
an analysis of variance and compare three categories of organizations.
2. The study cited did not assess the effects of environmental characteristics but focused
on the relationship between clearance data, crime data, and sworn personnel data. Based on
this limited set of explanatory variables, the observed year-to-year variation appears to be
chance variation (a “random walk” as discussed in Pindyck & Rubinfeld, 1991) rather than
shifts due to trending.
3. One might argue for lagged effects and examine clearance rates during the period after
1989. However, appropriate time-series analysis (via Auto Regressive Integrated Moving
Average [ARIMA] or Generalized Least Squares-Auto Regressive Moving Average [GLS-
ARMA]) is best conducted following the disaggregation of data for both independent and

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


196 POLICE QUARTERLY (Vol. 2, No. 2, June 1999)

dependent variables. Unfortunately, only one data point was available for some independent
variables, making this impossible. The time frame limitation used here presents the most
straightforward relationship.
4. Of 170 departments, 102 (60%) are located in communities between populations of
10,000 and 25,000.
5. These data were compiled from statistical tables included in the 1993 published data.
They include all agencies submitting uniform crime report data to the Federal Bureau of
Investigation during that year.
6. The term Hispanic refers to origin and may include individuals of any race. Thus, use
of this variable (which is not mutually exclusive) represents a potential problem. However,
the Texas population has a significant Hispanic proportion that must be recognized for this
study.
7. The FBI does monitor the reporting practices of some agencies on a routine basis, as do
the state-level statistical centers that forward data to the FBI.
8. Independent samples t tests revealed no significant differences between the survey
respondents and the larger agency population with regard to clearance rates. Thus, it appears
that there is no performance bias among responding agencies.
9. Bivariate correlational analysis was conducted for all variables used in the factor analy-
sis. This procedure revealed no contradictory or unusual relationships. Due to the large num-
ber of variables considered, a thorough discussion of zero-order correlations may be consid-
ered tedious and would be repetitive of the points made about the factor analysis. Therefore,
zero-order correlational analysis is not presented here.
10. The factor loadings produced scores that tap the absence of resources, which was due
largely to the combined effects of poverty and unemployment. The resulting factor variable
for resources was then recomputed, subtracting the value for each case from zero. The new
variable produced the same values, but it was signed in the opposite direction. This did not
affect the statistical analyses presented in the study; rather, it ensures that the results are prop-
erly interpreted by the reader.
11. For example, responding cities in the border counties of Cameron, El Paso, Hidalgo,
and Webb, Texas had poverty rates of 35%, 28%, 39%, and 36%, respectively. This compares
with a mean of 17.4% for all 170 cities in the sampling frame.
12. Suburban status is defined here as any city within a metropolitan statistical area that is
not designated as a central city by the Census Bureau.
13. Clearly, this is not a discussion of causal relationships and should not be interpreted as
such. Rather, it is a note about an observed correlational pattern.
14. Five cities were dropped from subsequent analysis due to missing data.
15. Eck (1984) presents his own qualification of this hypothesis, which he labels the tri-
age system.
16. Analysis of the residuals’ plot indicated a potential problem with regard to the equal-
ity of variances in the error terms. A replication of the models using weighted least squares
with a power transformation produced similar error terms, coefficients, and overall adjusted
R2 values. The original ordinary least squares models are presented here due to the more
straightforward interpretation of results.

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


Davenport / MUNICIPAL POLICE DEPARTMENTS 197

REFERENCES
Allen, D. N., & Maxfield, M. G. (1983). Judging police performance: Views and behavior of
patrol officers. In R. R. Bennett (Ed.), Police at work: Policy issues and analysis (pp. 65-
86). Beverly Hills, CA: Sage.
Allison, J. P. (1972). Economic factors and the rate of crime. Land Economics, 48, 193.
Alter, C., & Hage, J. (1993). Organizations working together. Newbury Park, CA: Sage.
Brandl, S. G., & Horvath, F. (1991). Crime-victim evaluation of police investigative perfor-
mance. Journal of Criminal Justice, 19, 293-305.
Brown, L. (1992). Violent crime and community involvement. Vital Speeches, 58, 182-184.
Cordner, G. W. (1989). Police agency size and investigative effectiveness. Journal of Crimi-
nal Justice, 17, 145-155.
Crank, J. P. (1990). The influence of environmental and organizational factors on police style
in urban and rural environments. Journal of Research in Crime and Delinquency, 27,
166-189.
Crank, J. P., & Wells, L. E. (1991). The effects of size and urbanism on structure among Illi-
nois police departments. Justice Quarterly, 8, 169-185.
Culbreth, P. J. (1993). Local government adaptation to a turbulent environment. Doctoral
dissertation, University of La Verne, La Verne, CA.
Danielson, M. N., & Doig, J. (1982). New York: The politics of urban regional development.
Berkeley: University of California Press.
Davenport, D. R. (1995). Crime management in Texas: Police personnel, clearance rates
and random walks. Paper presented at the annual meeting of the Southwestern Political
Science Association, Dallas, TX.
Decker, S. (1979). Allocating police resources and fluctuating crime rates. In D. M. Petersen
(Ed.), Police work (pp. 97-112). Beverly Hills, CA: Sage.
Denhardt, R. (1995). Public administration: An action orientation (2nd ed). Belmont, CA:
Wadsworth.
Dess, G. G., & Rasheed, A.M.A. (1991). Conceptualizing and measuring environments: A
critique and suggestions. Journal of Management, 17, 701-710.
Drucker, P. F. (1954). The practice of management. New York: Harper & Row.
Eck, J. E. (1984). Solving crimes. Research in Action, NIJ Reports, 184, 4-8.
Federal Bureau of Investigation. (1994). Uniform crime reports for the United States 1993.
Washington, DC: Author.
Federal Bureau of Investigation. (1995). Uniform crime reports for the United States 1994.
Washington, DC: Author.
Freudenburg, W. R. (1986). The density of acquaintanceship: An overlooked variable in
community research. American Journal of Sociology, 92, 27-63.
Fried, R. C. (1976). Performance in American bureaucracy. Boston: Little, Brown.
Gore, A. (1993). Creating a government that works better and costs less: Report of the
National Performance Review. New York: Random House.
Greenwood, P. W. (1970). An analysis of the apprehension activities of the New York City
Police Department. Santa Monica, CA: RAND.
Greenwood, P. W., Chaiken J. M., & Petersilia, J. (1977). The criminal investigation process.
Lexington, MA: DC Heath.

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


198 POLICE QUARTERLY (Vol. 2, No. 2, June 1999)

Hatry, H. P. (1975). Wrestling with police crime control productivity measurement. In J. L.


Wolfle & J. F. Heaphy (Eds.), Readings on productivity in policing (pp. 86-128). Wash-
ington, DC: Police Foundation.
Henderson, T. A. (1975). The relative effects of community complexity and of sheriffs upon
the professionalism of sheriff departments. American Journal of Political Science, 19,
107-132.
Jacob, H., Lineberry, R. L., Heinz, A. M., Rich, M. J., & Swank, D. H. (1982). Governmental
responses to crime. Washington, DC: National Institute of Justice.
Kelling, G. L., Pate, T., Dieckman, D., & Brown, C. E. (1974). The Kansas City preventative
patrol experiment: A summary report. Washington, DC: Police Foundation.
Kim, J., & Mueller, C. W. (1978). Introduction to factor analysis: What it is and how to do it
(07-013). Sage university paper series on quantitative applications in the social sciences.
Beverly Hills, CA: Sage.
Langworthy, R. H. (1985). Police department size and agency structure. Journal of Criminal
Justice, 13, 15-27.
Lawrence, P. R., & Lorsch, J. W. (1967). Organization and environment. Boston: Harvard
University.
Levine, J. P. (1975). The ineffectiveness of adding police to prevent crime. Public Policy, 23,
523-545.
Manning, P. K. (1984). Community-based policing. American Journal of Police, 3, 205-227.
Metzgar, S. (1996, July 10). New York’s crime rate takes a dive. Times Union.
Miller, L. S., & Whitehead, J. T. (1996). Introduction to criminal justice research and statis-
tics. Cincinnati, OH: Anderson.
National Center for Productivity and Quality of Working Life. (1978). Total performance
management: Some pointers for action. Washington, DC: Government Printing Office.
Norusis, M. J. (1992). SPSS/PC+ professional statistics version 5.0. Chicago: SPSS.
Norusis, M.J. (1994). SPSS professional statistics 6.1. Chicago: SPSS.
Orsburn, J. D., Moran, L., Musselwhite, E., Zenger, J. H., & Perrin, C. (1996). Self-directed
work teams: The new American challenge. In J. S. Ott (Ed.), Classic readings in organi-
zational behavior (2nd ed., pp. 272-281). Belmont, CA: Wadsworth.
Osborne, D., & Gaebler, T. (1992). Reinventing government: How the entrepreneurial spirit
is transforming the public sector. Reading, MA: Addison-Wesley.
Petersilia, J., Abrahamse, A., & Wilson, J. Q. (1990). A summary of RAND’s research on
police performance, community characteristics, and case attrition. Journal of Police Sci-
ence and Administration, 17, 219-226.
Pindyck, R. S., & Rubinfeld, D. L. (1991). Econometric models and economic forecasts (3rd
ed.). New York: McGraw-Hill.
Pogue, T. F. (1975). Effect of police expenditures on crime rates: Some evidence. Public
Finance Quarterly, 3, 14-44.
Reuss-Ianni, E. (1983). Two cultures of policing. New Brunswick, NJ: Transaction Books.
Rosen, E. D. (1993). Improving public sector productivity: Concepts and practice. Newbury
Park, CA: Sage.
Ross, B. H., Levine, M. A., & Stedman, M. S. (1991). Urban politics: Power in metropolitan
America (4th ed.). Itasca, IL: F. E. Peacock.
Sampson, R. J. (1985). Neighborhood and crime: The structural determinants of personal
victimization. Journal of Research in Crime and Delinquency, 22, 7-40.

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


Davenport / MUNICIPAL POLICE DEPARTMENTS 199

Sampson, R. J., & Groves, W. B. (1989). Community structure and crime: Testing social-
disorganization theory. American Journal of Sociology, 94, 774-802.
Savas, E. S. (1982). Privatizing the public sector. Chatham, NJ: Chatham House.
Schneider, V. W., & Wiersema, B. (1990). Limits and use of the uniform crime reports. In D.
L. MacKenzie, P. J. Baunach, & R.R. Roberg (Eds.), Measuring crime: Large-scale,
long-range efforts (pp. 21-48). Albany: State University of New York Press.
Sessions, W. S. (1990). Police and citizens working together. FBI Law Enforcement Bulletin,
59(10), 1.
Sharfman, M. P., & Dean, J. W., Jr. (1991). Conceptualizing and measuring the organiza-
tional environment: A multidimensional approach. Journal of Management, 17,
681-700.
Shaw, C., & McKay, H. (1942). Juvenile delinquency and urban areas. Chicago: University
of Chicago Press.
Shaw C., & McKay, H. (1969). Juvenile delinquency and urban areas (rev. ed.). Chicago:
University of Chicago Press.
Skogan, W. G. (1976). Efficiency and effectiveness in big-city police departments. Public
Administration Review, 36, 278-286.
Smith, D. A., & Jarjoura, G. R. (1988). Social structure and criminal victimization. Journal
of Research in Crime and Delinquency, 25, 27-52.
Spector, P. E. (1981). Sage university paper series on quantitative applications in the social
sciences. Research designs (07-023). Beverly Hills, CA: Sage.
Tichy, N. M., & Ulrich, D. O. (1984). The leadership challenge—A call for the
transformational leader. Sloan Management Review, 26, 59-68.
Trojanowicz, R. C., & Bucqueroux, B. (1990). Community policing. Cincinnati, OH: Ander-
son.
Vader, J. E. (1996, August 21). Portland crime rate tumbles. The Oregonian, p. A01.
Waegel, W. B. (1981). Case routinization in investigative police work. Social Problems, 28,
263-275.
Waegel, W. B. (1996). Patterns of police investigation of urban crimes. In G. W. Cordner, L. K.
Gaines, & V. E. Kappelerin (Eds.), Police operations: Analysis and evaluation (pp. 189-
211). Cincinnati, OH: Anderson.
Walker, S. (1994). Sense and nonsense about crime and drugs: A policy guide (3rd ed.).
Belmont, CA: Wadsworth.
Weisheit, R. A., Falcone, D. N, & Wells, L. E. (1994). Rural crime and rural policing. Wash-
ington, DC: National Institute of Justice.
Weisheit, R. A., Falcone, D. N., & Wells, L. E. (1995). Crime and policing in rural and
small-town America: An overview of the issues. Washington, DC: National Institute of
Justice.
Weiss, M. (1996, July 3). Crime rate drops again. New York Post.
Wellford, C. R. (1974). Crime and the police: A multivariate analysis. Criminology, 12,
195-213.
Willmer, M.A.P. (1970). Crime and information theory. Edinburgh, Scotland: Edinburgh
University Press.
Wilson, J. Q., & Kelling, G. L. (1982, March). Broken windows: Police and neighborhood
safety. Atlantic Monthly, 249, 29-38.

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010


200 POLICE QUARTERLY (Vol. 2, No. 2, June 1999)

Wycoff, M. A. (1982). Evaluating the crime-effectiveness of municipal police. In J. R.


Greene (Ed.), Managing police work: Issues and analysis (pp. 15-35). Beverly Hills, CA:
Sage.
Zhao, J. (1996). Why police organizations change. Washington, DC: Police Executive
Research Forum.

Douglas R. Davenport received a B.A. in Bible from Central Bible College,


and an M.P.A and a Ph.D. in political science from Texas Tech University. He
served the City of Lubbock, Texas in various capacities, including police offi-
cer, detective, and communications center supervisor. He is now an associate
professor of justice systems at Truman State University in Kirksville, Mis-
souri. He is currently is researching dimensions of rurality and their effects
upon crime and public policy, the implementation of community policing,
and the effects of the organizational environment upon public agencies.

Downloaded from pqx.sagepub.com by Marta Pita on October 10, 2010

You might also like