Criminal Justice and Behavior 2004 Varela 649 75
-
Upload
corina-ica -
Category
Documents
-
view
12 -
download
0
description
Transcript of Criminal Justice and Behavior 2004 Varela 649 75
-
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
1/2
http://cjb.sagepub.com/Criminal Justice and Behavior
http://cjb.sagepub.com/content/31/6/649Theonline version of this article can be found at:
DOI: 10.1177/0093854804268746
2004 31: 649Criminal Justice and BehaviorJorge G. Varela, Marcus T. Boccaccini, Forrest Scogin, Jamie Stump and Alicia Caputo
Personality Testing in Law Enforcement Employment Settings: A Metaanalytic Review
Published by:
http://www.sagepublications.com
On behalf of:
International Association for Correctional and Forensic Psychology
can be found at:Criminal Justice and BehaviorAdditional services and information for
http://cjb.sagepub.com/cgi/alertsEmail Alerts:
http://cjb.sagepub.com/subscriptionsSubscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
http://cjb.sagepub.com/content/31/6/649.refs.htmlCitations:
What is This?
- Nov 10, 2004Version of Record>>
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/content/31/6/649http://cjb.sagepub.com/content/31/6/649http://www.sagepublications.com/http://www.ia4cfp.org/http://cjb.sagepub.com/cgi/alertshttp://cjb.sagepub.com/subscriptionshttp://cjb.sagepub.com/subscriptionshttp://www.sagepub.com/journalsReprints.navhttp://www.sagepub.com/journalsReprints.navhttp://www.sagepub.com/journalsPermissions.navhttp://cjb.sagepub.com/content/31/6/649.refs.htmlhttp://cjb.sagepub.com/content/31/6/649.refs.htmlhttp://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://cjb.sagepub.com/content/31/6/649.full.pdfhttp://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://cjb.sagepub.com/content/31/6/649.full.pdfhttp://cjb.sagepub.com/content/31/6/649.refs.htmlhttp://www.sagepub.com/journalsPermissions.navhttp://www.sagepub.com/journalsReprints.navhttp://cjb.sagepub.com/subscriptionshttp://cjb.sagepub.com/cgi/alertshttp://www.ia4cfp.org/http://www.sagepublications.com/http://cjb.sagepub.com/content/31/6/649http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
2/2
10.1177/0093854804268746CRIMINAL JUSTICE AND BEHAVIOR
Varela et al. / PERSONALITY TESTING IN LAW ENFORCEMENT
PERSONALITY TESTING IN
LAW ENFORCEMENT
EMPLOYMENT SETTINGS
A Meta-Analytic Review
JORGE G. VARELA
Wilford Hall USAF Medical Center
MARCUS T. BOCCACCINI
FORREST SCOGIN
JAMIE STUMP
ALICIA CAPUTO
The University of Alabama
Meta-analysis was usedto (a) assess theoverall validity of personality measures as predictors of
law enforcement officer job performance, (b) investigate the moderating effects of study design
characteristics on this relation, and (c) compare effects for commonly used instruments in this
setting. Results revealed a modest but statistically significant relation between personality test
scoresand officerperformance.Prediction wasstrongestfor theCalifornia PsychologicalInven-
tory and weaker for the Minnesota Multiphasic Personality Inventory and Inwald Personality
Inventory. Effect sizeswere larger for studies examiningcurrent jobperformance, as opposed to
future job performance. Implications for using personality tests in the law enforcement officer
hiring process are discussed, and recommendations for future research are provided.
Keywords: meta-analysis; police; law enforcement; personality assessment
It is standard practice in most major law enforcement agencies toemploy the services of mental health professionals to screen jobcandidates. Thegeneral approach followed in these evaluations is one
of screening out unfit candidates, rather than selecting in preferred
candidates. The typical psychological evaluation involves screening
for both major mental illness and personality traits that may interfere
649
CRIMINAL JUSTICE AND BEHAVIOR, Vol. 31 No. 6, December 2004 649-675
DOI: 10.1177/0093854804268746
2004 American Association for Correctional Psychology
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
3/2
with law enforcement officer job performance. Such traits may
include an inability to deal with stressful situations, being prone to
violent outbursts, and potential for substance abuse. Screening out
unfit job candidates is especially important in law enforcement set-
tings because law enforcement officers are entrusted with the respon-
sibility of protecting the public from harm. Often, this work is done in
anenvironment inwhichpublicopinion of thepolicesubculture is low
and where the demands and stress of police work may be unappreci-
ated. Psychological screening is one mechanism for identifying offi-
cer candidates that may be unable to uphold their responsibilities in
this environment.
Many law enforcement agencies use personality measures as partof their employee selection procedures. Ash, Slora, and Britton
(1990) surveyed 99 major metropolitan and U.S. state police depart-
ments. Of the 99 surveys distributed, 62 were returned. The research-
ers found that 42 (67.7%) of the departments that responded reported
using personality tests as part of their employee selection procedures.
Twenty-five (40.3%) of the departments reported using two or more
personality tests foreach candidate. Although the frequent useof per-
sonality testing in officer selection procedures suggests that those
involved in the officer selection process believe these tests contribute
useful information, theextent to which personality tests arepredictive
of officer performance is unclear. Numerous published and unpub-
lished studies have examined the relation between personality mea-sures and law enforcement officer performance. Most frequent in the
empirical literature are studies attempting to identify specific person-
ality test scales or groups of scales that arepredictive of objective per-
formance criteria, suchas termination, absenteeism, tardiness, citizen
complaints, and commendations, or subjective performance criteria,
650 CRIMINAL JUSTICE AND BEHAVIOR
AUTHORNOTE:Jorge G. Varelais with the U.S.Air Forceat WilfordHall Medical
Center in San Antonio, Texas. This research is based on Jorge G. Varelas doctoral
dissertation at The University of Alabama.Forrest Scogin and Jamie Stump, Depart-
mentof Psychology, TheUniversity of Alabama.Marcus T. Boccaccini, Ph.D., is now
at Sam Houston State University, Texas.Alicia Caputo is now with theDepartment of
Human Services in Alexandria, Virginia. Correspondence concerning this manu-script should be addressed to Forrest Scogin, Ph.D., Psychology Department, Box
870348, The University of Alabama, Tuscaloosa, AL 35487; telephone: 205-348-
1924; fax: 205-348-8648; e-mail: [email protected].
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
4/2
such as supervisor andpeer ratings of performance.Despite thegrow-
ing number of studies in this area, there is a lack of quantitative inte-
gration in this literature. Because researchers in this field have used
different personality assessment instruments, different outcome mea-
sures, and different study designs, there is no clear consensus about
what can be predicted from law enforcement officerspersonality test
scores. The present study uses meta-analysis to provide a clearer
picture of the validity of personality measures in law enforcement
settings.
VALIDITY OF PERSONALITY MEASURES IN EMPLOYMENT SETTINGS
Severalpublishedmeta-analyseshave examinedthevalidity of per-
sonality testing in employment settings. Findings from these meta-
analyses are reviewedhere for twopurposes. First, the research meth-
odologiesused in thesemeta-analyses provide a basis for thedesign of
this study. Second, three of the meta-analyses included some data
from law enforcementsettings,andtheir results provide anestimateof
theeffect sizes that might be expectedfrom a meta-analysis basedon a
more comprehensive review of the existing law enforcement officer
performance literature.
Schmitt, Gooding, Noe, and Kirsch (1984) conducted a meta-
analysis of 99 employee selection studies published in the Journal of
Applied Psychology or Personnel Psychology between 1964 and1982.1 These researchersexamined the effectiveness of several differ-
ent types of predictors (personality measures, aptitude assessments,
physical ability measures) across several occupational groups (pro-
fessional, managerial, clerical, sales, skilled, and unskilled). Perfor-
mance criteria included both subjective (performance ratings) and
objective (turnover, achievement/grades, status changes, and wages)
measures. Using only personality measures as predictors, an overall
mean correlation of .149 was observed. A mean correlation of .206
was observed for subjective performance criteria, and mean correla-
tions ranged from .121 (turnover) to .152 (achievement/grades) for
objective performance criteria. Schmitt et al. also found that effect
sizes varied depending on study design characteristics. Studies usinga concurrent design (data from incumbents) or purely predictive
design (recruit data not used for hiring decisions) produced larger
Varelaet al./ PERSONALITYTESTINGIN LAWENFORCEMENT 651
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
5/2
mean correlations than studies using an employee selection design
(recruit data used for hiring decisions). However, the extent to which
study design characteristics influenced the predictive validity of per-
sonality measures was not reported. In their analyses of study design
characteristics, the researchers combined correlation coefficients
across all available predictors (aptitude assessments, physical ability
measures, personality measures).
Tett, Jackson, Rothstein, and Reddon (1994) conducted a meta-
analysis of 86 studies examining the relation between personality
measures and job performance (see also Tett, Jackson, & Rothstein,
1991). Eleven of these studies were of law enforcement officers or
recruits. The overall corrected sample-weighted mean correlationbetween personality measures and job performance was .174. The
researchers compared correlations for studies that predicted specific
relations between personality measures and performance (confirma-
tory design) andstudies without a clear rationaleforexpecting signifi-
cant results (exploratory design) and found that prediction wassignif-
icantly stronger for confirmatory studies (.238) than exploratory
studies (.035). Mean correlations were also significantly larger for
studies using job recruits (.267) compared to studies using only job
incumbents (.120) and for published studies (.215) compared to dis-
sertations (.049). Although correlations were higher for subjective
performance measures (.186) than objective performance measures
(.103), this difference was not large enough to achieve statisticalsignificance.
Barrick and Mount (1991) conducted a meta-analysis to examine
thepredictivevalidityof theBigFivepersonality traitsacross fiveoccu-
pational groups (professionals, police, managers, sales, andskilled or
semiskilled) for three job performance criteria (job proficiency, train-
ing proficiency, and personnel data). Findings were reported for each
of the Big Five personality dimensions for police officers, despite the
fact this occupational group accounted for only 13% of the 162 sam-
ples included in the meta-analysis. Mean correlations for police offi-
cers, corrected for range restriction and measurement error in crite-
rion and predictor variables, were .09 (extraversion), .10 (emotional
stability), .10 (agreeableness), .22 (conscientiousness), and .00 (open-
ness to experience). Salgado (1997) conducted a similar meta-
analysis using samples from European countries and also reported
652 CRIMINAL JUSTICE AND BEHAVIOR
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
6/2
separate effects for law enforcement samples (k= 3 for most effects).
Mean correlations for police officers, corrected for range restriction
and measurement error in criterion and predictor variables, were .13
(extraversion), .15 (emotional stability), .09 (agreeableness), .24
(conscientiousness), and .12 (openness to experience).
In sum, meta-analyses of the validity of personality tests in employ-
ment settings have found that personality test scores have a consistent
but modest relation to job performance indices. Barrick and Mounts
(1991) and Salgados (1997) results suggest that similar effect sizes
might be observed in meta-analyses using only law enforcement
officer samples, but the small number of law enforcement studies
included in these meta-analyses precludes strong predictions. Themeta-analyses reviewed above also suggest that effect sizes can vary
as a result of study design characteristics. Tett et al. (1994) found that
prediction was better for confirmatory studies compared to explor-
atory studies and for job recruits compared to job incumbents. Find-
ings from the Schmitt et al. (1984) meta-analysis suggest that predic-
tion isstrongestforstudies usinga concurrent orpredictivedesignand
weakest for studies using an employee selection design. The question
remains as to whether these same relations exist in studies using per-
sonality measures to predict law enforcement officer performance.
UNPUBLISHED META-ANALYSIS OF PERSONALITY TESTS
IN LAW ENFORCEMENT SETTINGS
To our knowledge, there is at least one existing, but unpublished,
meta-analysis of thevalidityof personality testing in law enforcement
settings. OBrien (1996) conducted a meta-analysis of 29 published
studies (N= 4,466) examining the relation between personality mea-
sures and law enforcement officer performance. In this unpublished
review, an overall validity coefficient of .25 was reported. OBrien
examined several moderator variables, including the personality
instrument (Minnesota Multiphasic Personality Inventory [MMPI]
vs. California Psychological Inventory [CPI]) and type of perfor-
mance criteria that were used (subjective vs. objective; training vs.
incumbent). When comparing the overall predictive validity of theMMPI and CPI, OBrien used two different strategies. First, she
examined effect sizes from studies in which a clinicians interpreta-
Varelaet al./ PERSONALITYTESTINGIN LAWENFORCEMENT 653
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
7/2
tion ofpersonality test data (sometimes accompaniedbyclinical inter-
view data) were used to predict performance. Under these circum-
stances (k= 10), she found similar effect sizes for the MMPI (.46) and
theCPI (.32).Second, sheexaminedthepredictivevalidityof individ-
ual test scales. Based on the pattern of her results, she concluded that
prediction was stronger for CPI scales (17 of 22 mean validity coeffi-
cients were significantly different from 0) compared to MMPI scales
(1 of 13 mean validitycoefficients was significantly different from 0).
With respect to performance criteria, subjective and objective criteria
were predicted equally well (.20 and .25, respectively), as were train-
ing criteria and actual job performance criteria (.19 and .27,
respectively).Despite the existence of the OBrien (1996) meta-analysis, there
are several reasons whyfurther integrationof this literature is needed.
First, OBrien included only published findings in her meta-analysis.
As a result, it is likely that her effect sizes are inflated, because jour-
nals tend topublish studies with significant findings. Indeed, Tettet al.
(1994) found that correlations between personality measures and job
performance indices were significantly larger in published studies
compared to unpublished studies. Second, OBriens comparison
of personality tests did not include the Inwald Personality Inven-
tory (IPI; Inwald, Knatz, & Shushman, 1982), a measure specifically
designed for screening law enforcement applicants. According to the
publisher of the IPI, Hilson Research, Inc. (2000-2001), their instru-ment is used by more than 30% of the nations state police depart-
ments. Finally, OBriens meta-analysis has not been published and,
to our knowledge, has not been subjected to peer review.
PURPOSE OF THE CURRENT META-ANALYSIS
This article reports the results of a meta-analytic review of the
validity of personality testing in law enforcement settings. Although
previous meta-analyseshave included lawenforcement samples (e.g.,
Barrick & Mount, 1991; Salgado, 1997; Tett et al., 1994), the number
of samples included were small given the size of the law enforcement
officer performance research literature. The current study is designedto extend what is known about the relation between personality test
scores and law enforcement officer job performance by examining
654 CRIMINAL JUSTICE AND BEHAVIOR
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
8/28
effects from a substantially larger number of law enforcement sam-
ples (both published and unpublished) compared to previous meta-
analyses. We report effects for the overall validity of personality tests
in this setting and examine the impact of several moderator variables,
including predictor type (MMPI, CPI, IPI), study design characteris-
tics, sample characteristics, and publication status (published vs.
unpublished findings).
METHOD
CASE SELECTION
Data for this study were retrieved from scholarly journals, books,
conference presentations, dissertations, theses, and unpublished reports
from practitioners and test publishers. Several methods were used to
identify relevant studies. First, searches of PsycInfo, Dissertation
Abstracts International, and the National Criminal Justice Research
Service were conducted to identify all references to personality
assessment in law enforcement settings. Second, all volumes of the
following journals were hand searched:Journal of Applied Psychol-
ogy,Personnel Psychology,Professional Psychology: Research and
Practice, Journal of Police Science and Administration, Law and
Human Behavior,Behavioral Sciences and the Law,Criminal Justiceand Behavior,Journal of Police and Criminal Psychology, andJour-
nal of Personality Assessment. Third, reference lists of already identi-
fied sources were reviewed. Fourth, a request for data was placed in
the American Psychological Association Division 18 (Psychologists
in Public Service) newsletter, and requests for data were submitted to
the Internet discussion forums of the International Association of
Chiefs of PolicePsychological Services Section and Division 41 of
the American Psychological Association (American Psychology
Law Society). Fifth, leading researchers in the field were contacted,
including Robin Inwald (Hilson Research), Robert Hogan (Hogan
Assessment Systems), George Hargrave, Dierdre Hiatt, Curt Bartol,
Larry Beutler, Stanley Azen, and Mark Axelberd. Using these meth-ods, approximately 175 studies containing personality data were
identified.
Varelaet al./ PERSONALITYTESTINGIN LAWENFORCEMENT 655
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
9/2
Inclusionary and exclusionary criteria were established to identify
data that were appropriate for the current meta-analysis. The follow-
ing inclusionary criteria were used: (a) studies using police, correc-
tional, government security, or other law enforcement personnel, (b)
studies linking personality test data with job performance, (c) studies
using either training performance data or job performance data, and
(d) studies using either objective (e.g., absenteeism, tardiness, citizen
complaints) or subjective (e.g., supervisor ratings, peer ratings) indi-
ces of performance. In addition, the following exclusionary criteria
were used: (a) studies reporting only the results of multiple-predictor
analyses (e.g., multiple regression), and (b) studies failing to report
data necessary for the computation of effect sizes. Of the initial 175data sets that were identified, 78 met the study criteria and were
included in the meta-analysis. Many of the originally identified data
sets were excluded because they did not compare personality data to
job performance (e.g., reported personality data only). Others were
excluded because theyonly reported data for multiple-predictor anal-
yses or because thepersonality tests used were actually measures of
cognitive ability or vocational interest.
Almost half of the usable data sets came from journal articles (k=
36)2, whereas others came from theses or dissertations (k= 27), gov-
ernment reports (k= 4), conference presentations (k= 4), researchers
(k= 3), books or bookchapters (k= 2), and test manuals (k= 2). These
studies ranged in date from 1950 to 1999.
DATA CODING
Coders were used to extract relevant data and study design infor-
mation from the identified studies. Coders were asked to identify data
that were appropriate for inclusion in the meta-analysis and to report
thenumberand type or types of statistical analyses used in each study.
Data were coded as being published if they were obtained from jour-
nal articles, books or book chapters, test manuals, or government
reports, and unpublished if they were obtained from theses, disserta-
tions, conference presentations, or researchers. Officer characteris-
tics, such as the average age and educational level of the sample, werecoded. Job performance predictors were identified, including the
names of personality tests andscales that were used. Measures of offi-
656 CRIMINAL JUSTICE AND BEHAVIOR
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
10/
cer performance were identified and coded as either objective (repri-
mands, complaints, suspensions, days of work missed) or subjective
(supervisor or peer ratings). Performance criteria were also identified
as reflecting either training performance or on-the-job performance.
The amount of time that elapsed between personality testing and col-
lection of performance data (measurement interval) was recorded.
Finally, the design of each study was classified on each of the follow-
ing characteristics: (a) Were the personality tests administered and
used as part of the hiring process (screening) or as part of a research
project only (analogue)? (b) Were the personality tests used to predict
future job performance (predictive design) or were they administered
to incumbent officers to examine the relation between personalitydimensions and current performance (concurrent design)? (c) Did the
authors select tests and/or test subscales based on a priori hypotheses
about the relations between specific personality measures and perfor-
mance indices (confirmatory design), or did they select personality
measures without a clear rationale for expecting significant results
(exploratory design)?
The quality of each study was rated using a modified version of the
Instrument for Evaluating Experimental Research Reports (IEERR)
(Suydam, 1968). The IEERR was developed for rating the quality of
studies comparing the effectiveness of different educational pro-
grams. For this study, modifications to the IEERR were made to
accommodate the nature of the literature under review.3 Items on theIEERR are summed to provide a single study quality rating for each
study.
Coders initially received 2 hours of training and instruction con-
cerning theproper useof thecodingmanualanddata collection forms.
After the initial training was completed, each of the five coders was
given a practice set of studies to code. A second training session was
then conducted to clarify questions about the coding manual and data
collection forms. After the second training session, the coders col-
lected the data used in the meta-analysis.
CODER AGREEMENT
Each study was reviewed and coded by two coders. Discrepancies
about the proper coding of study design characteristics were resolved
Varelaet al./ PERSONALITYTESTINGIN LAWENFORCEMENT 657
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
11/
through critical discussion. Thecodersratings of study quality on the
IEERR were found to be consistent (r= .89), andsubsequent analyses
examining study quality were completed using the average rating for
each study.
STATISTICAL CONVERSIONS AND CORRECTIONS
Testsof significance wereconverted to rfor use as a commoneffect
size using computational formulas provided by Wolf (1986) and
Rosenthal (1994). In cases where only means andstandard deviations
were reported,rvalues were estimated by first generating test statis-
tics (e.g.,t,F) and then converting these values tor.
An important consideration in theaggregation of correlational data
is the sign of the reported coefficients. Because the purpose of this
study was to determine the magnitude of the effect sizes associated
with prediction of job performance, irrespective of directionality,
absolute values of correlation coefficients were used. Tett, Jackson,
Rothstein, andReddon (1999) have noted that theuse of absolute val-
ues leads to an upward bias in mean validity coefficients. However,
these authors have also reported that the amount of upward bias is
minimal when samplesizes and the value of rho are large (e.g., bias of
.01 when = .15 andN 100). Given such a small estimated bias in
findings, we used absolute values of correlations without correcting
for upward bias.For each variable examined in this meta-analytic review, observed
(uncorrected) and corrected validity coefficients were calculated.
Correlations were corrected for three types of study artifacts: attenua-
tion due to unreliability of measures, attenuation or enhancement due
to range restriction, and attenuation due to dichotomization of out-
come variables (discontinuity). Although these corrections typically
increase the size of correlation coefficients, they are undertaken to
provide estimates of outcome values if thestudies hadbeen conducted
without methodological flaws (Hunter & Schmidt, 1994).
The first correction made to the observed correlations was for
attenuation due to unreliability of predictor variables (Hunter &
Schmidt, 1994). Reliability estimates for predictor variables wereobtained from test manuals, published reliability studies, and study
authors. When more than one reliability estimate was located for a
658 CRIMINAL JUSTICE AND BEHAVIOR
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
12/
particular test scale, the largest estimate was used because it led to the
smallest amountofcorrection. Testscalesforwhich no reliability data
were available were left uncorrected.
Thesecondcorrection made to observed correlationswasforatten-
uation or enhancement due to range departure of predictor variables.
When the standard deviation of a sample differs from the population
standard deviation, the observed correlation is distorted. When the
standard deviation of the sample is smaller than the population stan-
dard deviation, the observed correlation is attenuated, and, con-
versely, when the samplestandard deviation is larger than the popula-
tion standard deviation, the observed correlation is inflated. Range
restriction is common inpersonnel selectionsettings because thesam-ple under investigation has typically passed an initial screening and
represents a small proportion of all applicants. In the current meta-
analysis, observed correlations were corrected for range departure
using the procedures recommended by Hunter and Schmidt (1994).
Thecrucial determinant of themagnitudeof this correction is theratio
of thesamplestandard deviation to thepopulation standard deviation.
Correction for range departure could only be completed for studies
that reported sample standard deviations. Estimates of population
standard deviation were found in test manuals, handbooks, and pub-
lished research.
The last correction made to observed single predictor correlations
was for attenuation due to dichotomization of performance variables.The correction formula recommended by Hunter and Schmidt (1994)
was applied to studies reporting dichotomous variables.
According to the Hunter and Schmidt (1994) modelof meta-analysis,
each of the three attenuating or enhancing factors is independent.
Thus, observed correlations were corrected for more than one factor
when applicable using the procedures recommended by Hunter and
Schmidt.
WITHIN-STUDY AGGREGATION
The number of validity coefficients reported in individual studies
ranged from 1 to 1,222. Prior to within-study aggregation, the meta-analytic data set consisted of 3,954 correlations. Validity coefficients
were aggregated within studies so that findings would not be unduly
Varelaet al./ PERSONALITYTESTINGIN LAWENFORCEMENT 659
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
13/
influenced by studies reporting numerous findings. Specifically, valid-
ity coefficients were aggregated according to categorical moderator
variables of interest. This aggregation procedure was used so that
the influence of moderator variables could be examined in the meta-
analysis. If we had simply computed a single mean across all of the
findings in each study, the influence of moderator variables would
have been lost in the aggregation. For example, if a study used both
turnover and supervisor ratings as performance criteria, aggregating
across these criteria would make it impossible to examine differences
in the prediction of subjective and objective performance criteria.
The effects of the following categorical moderator variables were
examined (see Data Coding section above for variable descriptions):predictor type (MMPI vs. CPI vs. IPI), nature of outcome measure
(subjective vs. objective, training vs. incumbent), study design (con-
firmatoryvs. exploratory, concurrentvs.predictive, screeningvs. ana-
logue), and data source (published vs. unpublished).
Data aggregationswithinstudies were made using sample-weighted
means so that coefficients based on larger samples were accorded
greater weight in the aggregation. Sample-weighted means were also
used to compute the average sampling error estimates (Hunter &
Schmidt, 1990).
TESTING THE SITUATIONAL SPECIFICITY HYPOTHESIS
The variance of a population correlation is used to test the situa-
tional specificity hypothesis. This test is conducted to determineif the
observedmeancorrelationgeneralizes across samples.If theobserved
mean correlation generalizes across samples, it is not situationally
specific. A common way of evaluating the situational specificity
hypothesis is to use thethree-fourths rule. This rule states that when
less than 75% of the population correlation variance is accounted for
by sampling error variance, the role of moderator variables should be
examined (Hunter & Schmidt, 1990). In these situations, the variance
not accounted for by sampling error maybe attributable to systematic
sources, such as study design characteristics or differences in the
characteristics of the samples being studied.
The amount of variance explained by sampling error is calculated
by dividing the sampling error of the population correlation by the
660 CRIMINAL JUSTICE AND BEHAVIOR
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
14/
variance of the observed sample-weighted mean correlation. This
value can exceed 100% because sampling error is estimated algebra-
ically, whereas the variance associated with each mean correlation is
determined from the observed effect sizes (Hunter & Schmidt, 1990).
Sampling error was computed using the formula provided by Hunter
and Schmidt (1990). Sampling error estimates were not corrected for
variance due to study artifacts. This approach has been referred toas a
bare-bones meta-analysis (Hunter & Schmidt, 1990, p. 156).
DETECTION OF MODERATOR VARIABLES
The influence of moderator variables was examined using Pearsoncorrelations for continuous moderator variables andz-tests for cate-
gorical moderator variables (see Hunter & Schmidt, 1990, p. 437-
438). Uncorrected validitycoefficients were used inallof themodera-
tor analyses.
FINAL DATA SET CHARACTERISTICS
The meta-analyses reported in this article were conducted using
validity coefficients from 78 studies with a combined total of 11,725
participants. After within-study aggregation, the meta-analytic data
set contained 168 validity coefficients.
RESULTS
PREDICTIVE VALIDITY ACROSS ALL PREDICTOR AND
OUTCOME VARIABLES
Meta-analysis results for all samples and all categorical moderator
subgroups are provided in Table 1. Themean sample-weighted corre-
lationacrossall predictors andoutcomes was.134, which increasedto
.218 when corrected for study artifacts (see Table 1, row 1). The lower
bound of the 95% confidence interval for the uncorrected coefficient
was greater than 0, indicating a statistically significant relation
between personality tests scores and performance criteria. However,
only 64% of the variance in the overall validity coefficient (uncor-
Varelaet al./ PERSONALITYTESTINGIN LAWENFORCEMENT 661
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
15/
662
TABLE1:
Meta-AnalysisResultsforAllSamples
andCategoricalModerators
O
bs.
Samp.
%Var.(r)Due
LowerCI
UpperCI
Corr.N-
z-Score
Moderator
ka
Nb
N-Wtd.r
c
Va
r.(r)d
ErrorVar.e
toSamp.Error
f
(9
5%)g
(95%)h
Wtd.r
i
Diff.j
Allsample
s
78
11,725
.134
.
011
.007
64
.122
.157
.218
CPI
13
2,049
.155
.
006
.008
100
.141
.169
.251
IPI
11
2,537
.100
.
001
.004
100
.090
.110
.196
MMPI
41
6,940
.108
.
010
.005
52
.096
.120
.206
k
Training
18
3,820
.112
.
006
.006
100
.100
.124
.188
Performan
ce
72
9,747
.143
.
013
.007
59
.129
.157
.231
1.37
Subjective
53
5,962
.134
.
009
.008
89
.118
.150
.192
Objective
41
7,115
.138
.
012
.006
49
.128
.148
.239
.18
Predictive
59
10,185
.125
.
009
.006
68
.115
.135
.216
Concurren
t
21
1,661
.199
.
020
.013
66
.181
.217
.228
2.23*
Screening
49
8,168
.118
.
009
.006
65
.106
.130
.213
Analogue
30
3,616
.162
.
013
.009
67
.148
.176
.226
1.77
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
16/
663
Explorator
y
64
10,373
.133
.
010
.007
63
.121
.145
.224
Confirmatory
15
1,447
.132
.
013
.009
68
.116
.148
.161
.03
Unpublished
34
5,203
.112
.
012
.007
59
.075
.149
.182
Published
44
6,522
.149
.
009
.007
75
.121
.177
.245
1.57
Note.CPI
=CaliforniaPsychologicalInventory;IPI=I
nwaldPersonalityInventory;MMPI=Minne
sotaMultiphasicPersonalityInventory.
a.Numberofstudiesprovidingdatatothegivenaggregation.
b.Number
ofindividualparticipantscontributingdatatothegivenaggregation.
c.Observe
dsample-weightedmeancorrelation.
d.Varianceintheobservedcorrelations.
e.Samplin
gerrorvariance.
f.Proportionofvarianceintheobservedcorrelationsduetosamplingerror.
g.Lowerlimitof95%confidenceinterval(CI)aroundtheobservedsample-weightedmeancorrelation.
h.Upperlimitof95%confidenceintervalaroundtheo
bservedsample-weightedmeancorrelation
.
i.Correcte
dsample-weightedmeancorrelation(meas
urementerror,rangerestriction,discontinuity).
j.Significa
ncetestofthedifferencebetweenuncorrec
tedvaliditycoefficientforthevariableinthisrowandtherowdirectlyaboveitusingaz
score.
k.SeeTab
le2forsignificancetestscomparingtheMM
PI,CPI,andIPI.
*p .05 for all correlations.
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
19/
unpublished meta-analysis, in which it was found that 17 of 22 CPI
scales were predictive of officer performance compared to only 1 of
13 MMPI scales. Onepossible reason for thesuperior performance of
the CPI in this setting is that the CPI is designed to be a measure of
normal personality traits (Gough, 1995), whereas the MMPI and IPI
are, for themost part,measures of psychopathology, deviant personal-
ity traits, and maladaptive behavior. In many law enforcement set-
tings, applicants must survive a rigorous initial screening process
before they are asked to complete a personality measure. The screen-
ingprocess often includes civil service testing, background investiga-
tions, criminal history investigations, and evaluative interviews. Most
pathological job candidates are eliminated during this process. Per-sonality measures designed to detect pathological personality traits
may be redundant when they are administered after the initial screen-
ing process. In contrast, personality measures that are designed to
assess normal personality traits, such as the CPI, may be more useful
in this context because they provide information that is not obtained
during the initial screening process. For instance, the CPI was
designed to provide information about consistent styles of interper-
sonal behavior. Because being a successful police officer requires
effective interpersonal skills (e.g., interacting with community
members, other officers, and supervisors), the CPI may be a useful
measure for predicting this important aspect of officer performance.
Second, mean correlations were larger for studies using a concur-rent design compared to those using a predictive design. This finding
suggests that personality measures are somewhat better at predicting
current job performance than future job performance. Studies using a
concurrent design examine the relation between personality test data
and officer performance at the same point in time, whereas studies
using a predictive design attempt to link personality test data with
future performance. If an officer is experiencing noticeable psycho-
logical problems, it makes sense that these problems would affect his
or hercurrent performance. However,current psychological problems
may not always affect future performance, which reduces the likeli-
hood that measures of psychological functioning can be used to
predict future job performance.
Finally, there was no significant difference in mean effect sizes
from published and unpublished studies. One reason for conducting
666 CRIMINAL JUSTICE AND BEHAVIOR
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
20/
this meta-analysis was that previous meta-analyses of personality
testing in law enforcement settings have focused on data from pub-
lished studies (e.g., Barrick & Mount, 1991; OBrien, 1996; Salgado,
1997). The current meta-analysis included data from both published
and unpublished studies. Because scholarly journals tend to publish
studies with significant findings, it is often expected that effects from
unpublished studies will besmaller than those from publishedstudies.
Indeed, Tettet al. (1994) found in their meta-analysis that correlations
between personality measures and job performance indices were sig-
nificantly largerinpublishedstudies. Onepossibleexplanation forthe
lack of a significant difference in the current meta-analysis is that
many of the unpublished studies may have gone unpublished for rea-sons other than significance of their findings. Indeed, many of the 34
unpublished studies included in the meta-analysis did contain signifi-
cant effects. Because many of the unpublished studies were masters
theses and doctoral dissertations, it is possible that they were never
publishedbecause graduate studentsconductingthe researchwerenot
interested in pursuing academic careers. It is also possible that they
may have been submitted for publication but were rejected because
they did not provide enough new information to warrant publication.
Irrespective of the reasons why these unpublished studies were never
published, the inclusionof so many unpublished studies in thecurrent
meta-analysis suggests that the effects reported here are not likely to
be inflated due to the inclusion of only published research.
IMPLICATIONS FOR PRACTICE
Although effect sizes observed in this meta-analysis are modest,
there are several reasons why it would be inappropriate at this point to
conclude that personality measures should not be used in the law
enforcement officer hiring process. First, the hiring process in law
enforcement settings is lengthy and complex, and job candidates are
typically evaluated on a number of different psychological and medi-
cal variables. Personality functioning is just one of these variables.
Each variable is intended to provide a unique piece of information
about a job candidate that can be combined with other pieces of infor-mation to provide an estimate of the candidates suitability for hire.
Support for using this type of multifaceted approach comes from
Varelaet al./ PERSONALITYTESTINGIN LAWENFORCEMENT 667
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
21/
research studies showing that predictive validity can be enhanced
when different types of psychological and medical variables are used
in combination to predict officer performance (e.g., Scogin,
Schumacher, Gardner, & Chaplin, 1995).
Second, it should not be expected that every personality test scale
can predict officer performance. It is likely that many personality test
scales are not predictive of officer performance and that practitioners
should only interpret the few scales that have a meaningful relation to
officer performance. This argument is supported by findings from
studies in which researchers have been able to use multiple-predictor
analyses (e.g., multiple regression, discriminant function analysis) to
identify optimally weighted combinations of personality test scalesthat can be used to successfully predict officer performance. We have
identified 41 studies (published andunpublished) that have used mul-
tiple predictor analyses to predict law enforcement officer perfor-
mance from personality test data. It would be inappropriate to make
conclusions about the predictive validity of personality tests in law
enforcement settings without considering the findings from these 41
studies. Table 4 contains a stem and leaf plot of the effect sizes from
the 128multiple predictoranalyses reported in these studies. Although
caution should be used when interpreting the values in Table 4
because they have not been corrected for capitalization on chance,
they clearly show that prediction of officer performance can be quite
good when anoptimallyweightedsubsetof personality scalesis used.Finally, the purpose of using personality measures in the law
enforcement officer hiring process is to reduce the likelihood that a
dangerous officer will be hired. Dangerous officers can be harmful to
members of the public, to fellow officers, and to the publics trust in
law enforcement agencies. Given the risks associated with allowing a
dangerous officer to be hired, the small overall predictive effects of
personality measures may be salient in some specific cases.
FUTURE RESEARCH
Findings from the current meta-analyses have implications for
future research, both at the meta-analytic and individual study levels.At the meta-analytic level, the existing research literature can be used
668 CRIMINAL JUSTICE AND BEHAVIOR
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
22/
to examine the validity of individual personality tests scales.Although we found that the overall validity coefficients associated
with personality measures were modest, there may be individual test
scales that are more strongly associated with officer performance.
Moreover, the existing research literature can be used to examine the
predictive validity of specific personality traits, such as those of the
five-factor model, in predicting specific components of law enforce-
ment office performance (e.g., Barrick & Mount, 1991; Salgado,
1997).
Finally, given therelatively superior performanceof theCPI, future
individual studies should examine the predictive validity of other
existing measures of normative personality traits, such as the 16 Per-
sonalityFactor Questionnaire,Personality Assessment Inventory, and
NEO Personality Inventory. There currently are an insufficient num-
Varelaet al./ PERSONALITYTESTINGIN LAWENFORCEMENT 669
TABLE 4: Stemand LeafDiagram of Effect Sizes FromMultiple-Predictor Analy-
ses Examining the Relation Between Personality Measures and Offi-
cer Performance
5 .0 : 02344
12 .0 : 555677778999
13 .1 : 0000012344444
18 .1 : 555566666677777888
9 .2 : 122224444
15 .2 : 556677778889999
10 .3 : 0000013334
12 .3 : 566667778899
8 .4 : 00112334
5 .4 : 55559
6 .5 : 111122
5 .5 : 556674 .6 : 0001
1 .6 : 6
2 .7 : 02
1 .7 : 7
0 .8 :
0 .8 :
1 .9 : 3
0 .9 :
1 1.0 : 0
Note. Diagram includes 128 coefficients from 41 studies (published and unpublished).Coefficients are multiple correlations (R) from regression analyses and phi coefficientscalculated from discriminant function analysis classification tables. A list of these stud-ies is available from the first author (JGV).
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
23/
ber of existing studies to provide reliable estimates of the predictive
validityof theseinstruments in lawenforcementemployment settings.
NOTES
1. Schmitt,Gooding,Noe,& Kirsch (1984)didnot reporthowmany, if any, ofthe 99 studies
included in their meta-analysis examined employee selection in law enforcement settings.
2. Data from some samples were reported inmultiplejournalarticles. Forthe purpose of this
study, effects from different articles reporting on the same sample were grouped together as
being from the same sample.
3. The version of the revised Instrument for Evaluating Experimental Research Reports
(IEERR) that was used in this study is available from the first author.
REFERENCES
References marked with an asterisk indicate studies included in the meta-analysis.
*Anson, R. A., Mann, J. D., & Sherman, D. (1986). Niederhoffers cynicism scale: Reliability
and beyond.Journal of Criminal Justice,14, 295-305.
Ash,P.,Slora,K. B.,& Britton,C. F. (1990).Police agency officerselectionpractices.Journalof
Police Science and Administration,17, 258-269.
*Azen, S.P., Snibbe,H. M.,& Montgomery, H. R. (1973).A longitudinalpredictive study of suc-
cess and performance of law enforcement officers.Journal of Applied Psychology,57, 190-
192.
*Azen, S. P., Snibbe, H. M., Montgomery, H. R., Fabricatore, J., & Earle, H. H. (1974). Predic-
torsof resignation and performanceof law enforcementofficers.American Journal of Com-
munity Psychology,2, 79-86.
*Baehr, M. E., Furcon, J. E., & Froemel, E. C. (1968).Psychological assessment of patrolmen
qualifications in relation to field performance: The identification of predictors for overall
performance of patrolmen and the relation between predictors and specific patterns of
exceptional and marginal performance. Washington, DC: Government Printing Office.
*Band, S. R.,& Manuele,C. A. (1987).Stressand police officerperformance:An examinationof
effective coping behavior.Journal of Police and Criminal Psychology,3, 30-42.
Barrick, M. R., & Mount, M. K. (1991). The Big Five personality dimensions and job perfor-
mance: A meta-analysis.Personnel Psychology,44, 1-26.
*Bartol, C. R. (1982). Psychological characteristics of small-town police officers.Journal of
Police Science and Administration,10, 58-63.
*Bartol,C. R. (1991).Predictive validation of theMMPIfor small-town police officerswho fail.
Professional Psychology: Research and Practice,22, 127-132.
*Bartol,C. R.,Bergen,G. T., Volckens,J. S.,Knoras, K. M. (1992).Womenin small-town polic-
ing: Job performance and stress. Criminal Justice and Behavior,19, 240-259.
*Benner, A. W. (1991).The changing cop: A longitudinal study of psychological testing within
law enforcement. Unpublished doctoral dissertation, Saybrook Institute, San Francisco.
670 CRIMINAL JUSTICE AND BEHAVIOR
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
24/
*Beutler, L. E., Storm,A., Kirkish,P., Scogin, F., & Gaines, J. A. (1985).Parameters in the pre-
diction of police officer performance. Professional Psychology: Research and Practice,16,
324-335.
*Boyce, T. N. (1988). Psychological screening for high risk police specialization. Unpublished
doctoral dissertation, Georgia State University, Atlanta.
*Bozza, C. M. (1990).Improving the prediction of police officer performance from screening
information. Unpublished doctoral dissertation, United States International University, San
Diego.
*Bradford, A. C. (1991). Psychological screening for narcotics officers and detectives. Unpub-
lished doctoral dissertation, Miami University, Oxford, OH.
*Cope, J.R. (1981).Personalitycharacteristicsof successfulversus unsuccessful policeofficers.
Unpublished doctoral dissertation, Florida Institute of Technology, Melbourne.
*Corey, D. M. (1988).The psychological suitability of police officer candidates. Unpublished
doctoral dissertation, Fielding Institute, Santa Barbara, CA.
*Cortina, J. M., Doherty, M. L., Schmitt, N., Kaufman, G., & Smith, R. G. (1992). The Big
Five personality factors in the MMPI and IPI: Predictors of police performance. PersonnelPsychology,45, 119-140.
*Costello, R. M.,Schoenfeld,L. S.,& Kobos,J. (1982).Policeapplicantscreening: Ananalogue
study.Journal of Clinical Psychology,38, 216-221.
*Daley, R. E. (1978).The relationship of personality variables to suitability for police work.
Unpublished doctoral dissertation, Florida Institute of Technology, Melbourne.
*Dean, D. (1974).The relationship betweenEysenckian personalityvariablesand ratings of job
performanceandpromotionpotential of a groupof police officers. Unpublisheddoctoraldis-
sertation, Ball State University, Muncie, IN.
*DuBois, P. H.,& Watson,R. I. (1950). Theselectionof patrolmen.Journalof Applied Psychol-
ogy,34, 80-95.
*Eisenberg, T., & Dowdle,M. (1981).Officer selection & performancestudy,San Jose, Califor-
nia police department. Los Gatos, CA: Personnel Performance.
*Geraghty, M. F. X. (1986).TheCPI test as a predictor of law enforcementofficer performance.
Unpublished doctoral dissertation, Florida Institute of Technology, Melbourne.
*Gottlieb,M. C., & Baker, C. F. (1974, May). Predictingpoliceofficer effectiveness. Paper pre-sented at the annual meeting of the Southwestern Psychological Association, El Paso, TX.
Gough, H. G. (1995).California Psychological Inventory: Introduction to form 434. Palo Alto,
CA: Consulting Psychologists.
*Griffith, T. L. (1991). Correlates of police and correctional officer performance. Unpublished
doctoral dissertation, Florida State University, Tallahassee.
*Hargrave, G. E. (1985). Using the MMPI and CPI to screen law enforcement applicants: A
study of reliabilityandvalidityofcliniciansdecisions.Journalof PoliceScience andAdmin-
istration,13, 221-224.
*Hargrave, G. E., & Hiatt, D. (1989). Use of the California psychological inventory in law
enforcement officer selection.Journal of Personality Assessment,53, 267-277.
*Hargrave, G. E., Hiatt, D., & Gaffney, T. W. (1986).A comparison of MMPI and CPI test pro-
files fortraffic officersand deputysheriffs.Journalof Police Science andAdministration, 14,
250-258.
*Hargrave, G. E.,Hiatt, D.,& Gaffney, T. W. (1988). F+4+9+Cn:An MMPI measureof aggres-
sion in law enforcement officers and applicants.Journal of Police Science and Administra-tion,16, 268-273.
*Henderson, N. D. (1979).Criterion-related validity of personality and aptitude scales: A com-
parison of validation results under voluntary and actual conditions. In C. D. Spielberger
Varelaet al./ PERSONALITYTESTINGIN LAWENFORCEMENT 671
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
25/
(Ed.),Police selection and evaluation: Issues and techniques (pp. 179-195). New York:
Praeger.
*Hess, L. R. (1972).Police entry tests and their predictability of score in police academy and
subsequent job performance. Unpublished doctoral dissertation,Marquette University, Mil-
waukee, WI.
*Hiatt, D., & Hargrave, G. E. (1988). MMPI profiles of problem peace officers.Journal of Per-
sonality Assessment,52, 722-731.
*Hiatt, D., & Hargrave, G. E. (1988). Predicting job performance problems with psychological
screening.Journal of Police Science and Administration,16, 122-125.
Hilson Research, Inc. (2000-2001).Testing/assessment services for public safety & security
[Brochure]. Kew Gardens, NY: Author.
*Hogan,R. (1971).Personalitycharacteristics ofhighlyrated policemen.Personnel Psychology,
24, 679-686.
*Hogan R., & Hogan J. (1995). Sheriff deputies.Hogan Personality Inventory manual. Tulsa,
OK: Hogan Assessment Systems.
*Hogan R., & Hogan J. (1995). Validity of the Hogan Personality Inventory for selecting policeofficers in (anonymous). Tulsa, OK: Hogan Assessment Systems.
*Hogan R., & Hogan J. (1995). Validity of the Hogan Personality Inventory for selecting police
officers in an Ohio municipality. Tulsa, OK: Hogan Assessment Systems.
*Hooke, J. F., & Krauss, H. H. (1971). Personality characteristics of successful sergeant appli-
cants.Journal of Law, Criminology, and Police Science,62, 104-106.
Hunter, J. E., & Schmidt, F. L. (1990).Methods of meta-analysis: Correcting error and bias in
research findings. Newbury Park, CA: Sage.
Hunter,J. E.,& Schmidt,F.L. (1994).Correctingfor sources ofartificial variation acrossstudies.
InH. Cooper& L.V.Hedges (Eds.), Thehandbookof research synthesis (pp.323-336).New
York: Russell Sage.
*Hwang, G. S. (1988).Validity of the California Psychological Inventory for police selection.
Unpublished masters thesis, North Texas State University, Denton.
*Inwald, R. E., & Brockwell, A. L. (1991). Predicting the performance of government security
personnel with the IPI and MMPI.Journal of Personality Assessment,56, 522-535.
*Inwald, R. E., Flanagan, C. L., & Kaufman, J. C. (1991, August).Officer supervisory ratingsclassifications. Paper presented at the annual convention of the American Psychological
Association, San Francisco.
*Inwald, R. E., Kaufman,J. C., & Solomon, R. (1991,August).IPI and HPP/SQ predictions of
peer ratings and class standings. Paper presented at the annual convention of the American
Psychological Association, San Francisco.
Inwald, R.E., Knatz,H.,& Shusman,E. (1982).Inwald Personality Inventorymanual. Kew Gar-
dens, NY: Hilson Research.
*Inwald,R. E., & Patterson, T.(1990).Useof theIPI andHPP/SQ forpredictingtrainee perfor-
mance in a government law enforcement agency. Kew Gardens, NY: Hilson Research.
*Inwald, R. E., & Sakales, S. R. (1982,August).Role of two personality screening measures to
identify on-the-jobbehavior problemsof law enforcement officer recruits. Paper presentedat
the annual convention of the American Psychological Association, Washington, DC.
*Inwald, R. E., & Shusman, E. J. (1984). The IPI and MMPI as predictors of academy perfor-
mance for police recruits.Journal of Police Science and Administration,12, 1-11.
*Inwald, R. E., & Shusman, E. J. (1984). Personality and performance sex differences of lawenforcement officer recruits.Journal of Police Science and Administration,12, 339-347.
672 CRIMINAL JUSTICE AND BEHAVIOR
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
26/
*Kleiman, L. S., & Gordon, M. E. (1986). An examination of the relationship between police
trainingacademyperformance and jobperformance.Journalof Police Science and Adminis-
tration,14, 293-299.
*Levine, M. (1979).Development of an MMPI subscale as an aid in police officer selection .
Unpublisheddoctoral dissertation,California SchoolofProfessionalPsychology,Berkeley.
*Mandel, K. (1970). The predictive validity of on-the-job performance of policemen from
recruitment information. Unpublished doctoral dissertation, University of Utah, Ogden.
*Marsh, S. H. (1962, January). Validating the selection of deputy sheriffs. Public Personnel
Review, 41-44.
*Mass, G. (1979). Using judgment and personality measures to predict effectiveness in
policework: An exploratorystudy. Unpublisheddoctoraldissertation,Ohio State University,
Columbus.
*Matyas, G. S. (1980). The relationshipof MMPI and biographical data to police selection and
police performance. Unpublished doctoral dissertation, University of MissouriColumbia.
*McDonough, L. B., & Monahan, J. (1975). The quality of community caretakers: A study of
mental health screening in a sheriffs department.Community Mental Health Journal,11,33-43.
*Merian, E. M., Stefan, D., Schoenfeld,L. S., & Kobos, J. C. (1980). Screening of police appli-
cants: A 5-item MMPI index.Psychological Reports,47, 155-158.
*Mills,C. J., & Bohannon, W. E. (1980).Personalitycharacteristicsof effective state police offi-
cers.Journal of Applied Psychology,65, 680-684.
*Mills,M. C. (1980).TheMMPI andthe prediction of policejob performance. Unpublisheddoc-
toral dissertation, University of Southern California, Los Angeles.
*Mufson, D. W., & Mufson, M. A. (1998). Predicting police officer performance using the
Inwald personality inventory: An illustration from Appalachia.Professional Psychology:
Research and Practice,29, 59-62.
*Neal, B. (1986). The K Scale (MMPI) and job performance. In J. T. Reese & H. A. Goldstein
(Eds.),Psychological services for law enforcement(pp. 83-90). Washington, DC: Govern-
ment Printing Office.
OBrien, S. G. (1996). The predictive validity of personality testing in police selection: A meta-
analysis. Unpublished masters thesis, University of Guelph, Guelph, Canada.*Pugh, G. (1985).The California PsychologicalInventory andpoliceselection.Journalof Police
Science and Administration,13, 172-177.
*Rand, T. M., & Wagner, E. E. (1973). Correlations between hand test variables and patrolmen
performance.Perceptual and Motor Skills,37, 477-478.
*Reming,G. C. (1988).Personality characteristicsof supercopsand habitual criminals.Journal
of Police Science and Administration,16, 163-167.
*Roberg,R. R. (1978).An analysis of the relationshipsamong higher education, belief systems,
and jobperformance of patrol officers.Journalof PoliceScience and Administration, 6, 336-
344.
Rosenthal, R. (1994). Parametric measures of effect size. In H. Cooper & L. V. Hedges (Eds.),
The handbook of research synthesis(pp. 231-244). New York: Russell Sage.
*Rybicki,S. L.,& Hogan,J. C. (1997). Validity of the Hogan Personality Inventory for selecting
deputy sheriff correctional officers. Tulsa, OK: Hogan Assessment Systems.
Salgado, J. F. (1997).The five factor modelof personalityand job performance in the European
Community.Journal of Applied Psychology,82, 30-43.
Varelaet al./ PERSONALITYTESTINGIN LAWENFORCEMENT 673
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
27/
*Sanchione, C. D.,Cuttler, M. J.,Muchinsky, P. M.,& Nelson-Gray, R. O. (1998). Prediction of
dysfunctional job behaviors among law enforcement officers.Journal of Applied Psychol-
ogy,83, 904-912.
*Saxe, S. J., & Reiser, M. (1976). A comparison of three police applicant groups using the
MMPI.Journal of Police Science and Administration,4, 419-425.
Schmitt, N., Gooding, R. Z., Noe, R. A., & Kirsch, M. (1984). Metaanalyses of validity studies
published between 1964 and 1982 and the investigation of study characteristics.Personnel
Psychology,37, 407-422.
*Schoenfeld, L. S., & Kobos, J. C. (1980). Screening police applicants: A study of reliability
with the MMPI.Psychological Reports,47, 419-425.
*Schuerger, J. M., Kochevar, K. F., & Reinwald, J. E. (1982). Male and female corrections offi-
cers: Personality and rated performance. Psychological Reports,51, 223-228.
Scogin, F., Schumacher, J. E., Gardner, J., & Chaplin, W. (1995). Predictive validity of psycho-
logical testing in law enforcement settings.Professional Psychology: Research and Prac-
tice,26, 68-71.
*Sendo, J. A. (1972).A study of the potential use of the Mann Attitude Inventory in the selectionof police recruits. Unpublished doctoral dissertation, Michigan State University, East
Lansing.
*Serko, B. A. (1981).Police selection: A predictive study. Unpublished doctoral dissertation,
Florida School of Professional Psychology, Tampa.
*Shaw, J. H. (1986). Effectiveness of the MMPI in differentiating ideal from undesirable police
officer applicants. In J. T. Reese & H. A. Goldstein (Eds.),Psychological services for law
enforcement(pp. 91-95). Washington, DC: Government Printing Office.
*Shusman, E. J., Inwald, R. E., & Landa, B. (1984). Correction officer job performance as pre-
dicted by the IPI and MMPI: A validation and cross-validation study. Criminal Justice and
Behavior,11, 309-329.
*Sterrett, M.R. (1984).Theutility of theBipolarPsychologicalInventory forpredicting tenure of
law enforcement officers. Unpublished doctoral dissertation, Claremont Graduate Univer-
sity, Claremont, CA.
*Super, J. T. (1995). Psychological characteristics of successful SWAT/tactical response team
personnel.Journal of Police and Criminal Psychology,10, 60-63.Suydam, M. N. (1968).An instrument forevaluating experimental educational research reports.
The Journal of Educational Research,61(3), 200-203.
*Sweda, M. G. (1988). The Iowa Law Enforcement Personnel Study: Predictionof law enforce-
ment job performance from biographical and personality variables. Unpublished doctoral
dissertation, University of Iowa, Iowa City.
*Swope, M. R. (1989). Validating state policetroopercareer performance with the SixteenPer-
sonality Factor questionnaire. Unpublished doctoral dissertation, Wayne State University,
Detroit, MI.
*Talley, J.E., & Hinz,L. D.(1990).Performancepredictionof public safetyand lawenforcement
personnel: A study in race and gender differences and MMPI subscales. Springfield, IL:
Charles C Thomas.
Tett,R. P., Jackson,D. N.,& Rothstein, M. (1991).Personalitymeasures as predictors of jobper-
formance: A meta-analytic review.Personnel Psychology,44, 703-742.
Tett, R. P., Jackson, D. N.,Rothstein, M., & Reddon, J. R. (1994). Meta-analysis of personality-
job performance relations: A reply to Ones, Mount, Barrick, and Hunter (1994).PersonnelPsychology,47, 157-172.
Tett, R. P., Jackson, D. N.,Rothstein, M., & Reddon, J. R. (1999).Meta-analysisof bidirectional
relations in personality-job performance research.Human Performance,12, 1-29.
674 CRIMINAL JUSTICE AND BEHAVIOR
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/ -
5/19/2018 Criminal Justice and Behavior 2004 Varela 649 75
28/2
*Tomini,B.A. (1995).Theperson-jobfit: Implications of selecting policepersonnel on thebasis
of job dimensions, attitudes, and personality. Unpublished doctoral dissertation, University
of Windsor, Windsor, Canada.
*Topp, B. W., & Kardash, C. A. (1986). Personality, achievement, and attrition: Validation in a
multiple-jurisdictionpolice academy.Journalof Police Science andAdministration, 14, 234-
241.
*Uno, E.A. (1979).Thepredictionof jobfailure:A studyof policeusingtheMMPI. Unpublished
doctoral dissertation, California School of Professional Psychology, Berkeley.
*Ward, J. C. (1981).The predictive validity of personality and demographic variables in the
selection of lawenforcementofficers. Unpublisheddoctoraldissertation, Universityof South
Florida, Tampa.
*Weekes, E.M. (1994).Theinfluenceof personalitydimensionsandphysicalabilitieson a pistol
shooting task. Unpublished doctoral dissertation, University of Houston, Houston, TX.
*Weiss, W. U., & Beuhler, K. (1995). The Psychopathic Deviate scale of the MMPI in police
selection.Journal of Police and Criminal Psychology,10, 57-60.
*West, S. D. (1988).The validity of the MMPI in the selection of police officers. Unpublishedmasters thesis, North Texas State University, Denton.
Wolf, F. M. (1986).Meta-analysis: Quantitative methods for research synthesis. Beverly Hills,
CA: Sage.
Varelaet al./ PERSONALITYTESTINGIN LAWENFORCEMENT 675
at University of Bucharest on August 5, 2014cjb.sagepub.comDownloaded from
http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/http://cjb.sagepub.com/