Validity of Officer Self-Reported Citizen Complaints

• Author: Matthew J. Hickman
• Published date: 01 September 2007
• Date: 01 September 2007
• DOI: http://doi.org/10.1177/1098611106294760
• Subject Matter: Articles

Police Quarterly
Volume 10 Number 3
September 2007 332-341
© 2007 Sage Publications
10.1177/1098611106294760
Validity of Officer Self-Reported
http://pqx.sagepub.com
hosted at
Citizen Complaints
http://online.sagepub.com
A Research Note
Matthew J. Hickman
Seattle Universtiy
This research note explores the use of self-report surveys in police samples to study
citizen complaints. Officer self-reported citizen complaints are compared with official
departmental complaints data among a sample of 440 officers in a large municipal
police department. Findings reflect moderate correspondence between self-reported and
official data for ever having been the subject of a complaint (data correspond for 77% of
respondents; phi = .54). Validity appears stronger for dichotomous measures of com-
plaints, compared with frequency measures, though correspondence was high among
those officers self-reporting the frequency of complaints (data were within ±1 complaint
for 78.5% of these respondents; r = .81 after removing outliers). Demographic correlates
are similar for both official and self-report data. Results generally support the use of self-
report surveys in police samples to study citizen complaints.
Keywords:
self-reports; citizen complaints; police behavior
Our knowledge of the varieties of police problem behavior comes from a broad
range of data sources. Official departmental data are an obvious source. However,
like data concerning crimes known to the police, official departmental data concerning
officer behavior reflect behavior known to, and recognized by, the police administra-
tion. There remains a very sizeable “gray figure,” and the official data probably repre-
sent the proverbial tip of the iceberg.
Observational data are a useful source of information about police problem
behavior. Black and Reiss’s (1967) classic study of more than 5,000 police-citizen
interactions in Washington D.C., Chicago, and Boston is a good example: Some offi-
cers were verbally and physically abusive toward citizens, and some officers violated
the law including acceptance of bribes, use of “drop guns,” and receipt of stolen
property, among other offenses. The rate of all such criminal violations was reported
as an astonishing 23.7 per 100 in “City X,” 21.9 in “City Y,” and 16.5 in “City Z”
Author’s Note: Correspondence concerning this article should be addressed to Matthew J. Hickman, Seattle
University, Department of Criminal Justice, 901 12th Avenue, PO Box 222000, Seattle, WA 98122, e-mail:
matthew.hickman@usdoj.gov.
332

---

Hickman / Validity of Officer Self-Reports
333
(Reiss, 1971, p. 156). However, such studies place a large burden on officers and other
agency resources, are generally more expensive to fund, and may require extensive
training for observers and/or special equipment. Observational studies may also be
undesirable, given the relative infrequency of the events that one is trying to observe
and the potential of both observer bias and participant reactivity.
Another option is to survey citizens regarding their experiences with the police. One
example is the Police Public Contact Survey (PPCS), conducted as part of the National
Crime Victimization Survey (NCVS) in 1996, 1999, 2002, and 2005. This approach
seems to produce useful estimates of the overall prevalence of police-public contacts,
but with regard to specific types of contact, such as those involving force, the data are
limited by the relative infrequency of force incidents as well as the subjective inter-
pretations of the respondents. Also, the expense of hiring and training interviewers (or
contracting with professional interviewers) is an issue to consider.
A remaining and largely unexplored option is the self-report survey method. At
present, we do not know a great deal about the feasibility of such an approach. Some
scholars have argued that self-reports are inappropriate for this type of research (e.g.,
Goldstein, 1977; Klockars, Ivkovich, Harver, & Haberfeld, 1997), largely basing
their judgment on presumptions about the suspicious nature of police officers rather
than on empirical data. This highlights a significant measurement problem for social
scientists. The issue is whether the behaviors of interest can be studied in a valid,
reliable, and quantifiable manner using common social science techniques.
One approach would be to conduct an assessment of criterion-related (concurrent)
validity. That is, given a known accurate measure of some phenomenon, does the alter-
native measure yield similar results? An example would be to gauge the accuracy of
self-reports using official data as a standard. To the extent that the self-report data cor-
respond with the official data, self-reports may be used as a surrogate measure.
Unfortunately, researchers may not be free to assume that official data can serve as an
accurate criterion against which to compare self-report data. To the contrary, the offi-
cial data in some agencies may be less than accurate because of diversion in the course
of administrative processes, poor record-keeping practices, and other reasons. This
presents a kind of “measurement gap” insofar as the official departmental data do not
reflect actual incidents. Given this measurement gap, a multiple-measures approach is
more appropriate. Here, the criterion measure is not assumed to be perfectly accurate,
and the goal is to provide a more accurate description by using alternative measures
that may reveal more or less about the phenomenon of interest. By comparing alterna-
tive measures against each other, the researcher can see where measures correspond
and where each of the measures contributes additional information.
Given the societal costs of police...