Description
Using various types of assessments in coaching provides an opportunity for the client to identify key elements that impact their personality when working with others.Read the article “Police Applicant Response Bias on the California Psychological Inventory,” found in the Topic 6 Resources.In 500-750 words:Describe possible ethical considerations when using this type of tool.Describe how to remedy these considerations with a client.
Unformatted Attachment Preview
Psychological Services
© 2021 American Psychological Association
ISSN: 1541-1559
2022, Vol. 19, No. 1, 176–182
https://doi.org/10.1037/ser0000524
Police Applicant Response Bias on the California Psychological Inventory
Paul Detrick1 and Ryan M. Roberts2
1
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2
Florissant Psychological Services, Florissant, MO, United States
Law Enforcement Psychological Services, Inc., Los Gatos, CA, United States
Positive response bias is common under the high demand conditions of personnel selection. This study
investigated positive response bias on the California Psychological Inventory (CPI)-434 by police officer
applicants as scored by the CPI-434-Police and Public Safety Report. Police officer applicants completed the
CPI-434 under high and then again under low demand conditions. The high demand condition was
represented by the CPI administration as part of the preemployment psychological evaluation. The low
demand condition was represented by a CPI administration with no contingencies approximately 4 months
later during the police academy. Demand effects were illustrated by comparing applicant scores (withinsubject) under high versus low demand conditions on CPI-Police and Public Safety Report (CPI-PPSR)
prediction equations and CPI scales. Significant demand effects were observed on 23 of 46 comparisons.
Modest support for the construct validity of the Good Impression (Gi) scale and somewhat less for the Fake
Good Index as measures of positive response bias were observed. Response bias potential impact on police
officer performance predictions was discussed.
Impact Statement
This study illustrates the effects of response bias, an issue of concern in personnel selection, on the
California Psychological Inventory (CPI) widely used in police officer selection. The study utilizes a
within-subject/repeated measure design uniquely capable of illustrating positive response bias. It also
provides modest support for the construct validity of the validity scales measuring positive response bias
and discusses the potential impact of response bias on interpretation.
Keywords: response bias, CPI, police officer selection, demand effects, CPI-PPSR
generates combined gender norms for police and public safety
applicants, a comparison group of entry level applicants that were
hired and demonstrated satisfactory work performance at 1 year, a
series of suitability risk statements, and a listing of scales and risk
statements thought to be unfavorable indicators of potential performance based on their content and deviation from the average of the
mean of the U.S. population norms or special group norms.
In addition to substantive scales, the CPI-434 contains two measures of response bias, the Good Impression (Gi) scale and the Fake
Good Index. The Gi scale was developed by Gough using a sample of
testing under normal conditions and a sample with instructions to
“imagine themselves as applying for a job they very much wanted.”
(Gough & Bradley, 1996). To provide a more precise classification of
test taker’s approach to the CPI, Gough refined Lanning’s (1989)
analyses to produce three validity equations which use weighted
combinations of CPI scales to identify Fake Good, Fake Bad, and
Random protocols (Gough & Bradley, 1996). Cut scores on the
indices were established to maximize the correct identification of
experimentally produced protocols and to minimize misclassification.
Notably, in his validation analyses, Gough identified the highest
incidence of protocols identified as Fake Good (7.5%) to occur in a
police officer applicant sample collected from a midsized city on the
West coast (Gough & Bradley, 1996).
Support for use of the CPI-434 in the prediction of police officer
performance has been demonstrated. Prediction of police officer
dysfunctional behavior by use of the CPI-434 was reported by
Sarchione et al. (1998). Responsibility (d = .60), Socialization
The California Psychological Inventory (CPI; Gough & Bradley,
1996) is the most widely used measure of normal personality
functioning in police officer selection (Corey, 2016). The CPI is a
self-report inventory with most recent versions consisting of either
434 or 260 items. From the CPI-434, Roberts et al. (2018) developed
the CPI Police and Public Safety Report (CPI-434-PPSR). This report
is based on a normative sample of more than 50,000 public safety
applicants and supplements the basic CPI interpretive report. The
items of the CPI were generated to represent folk scales reflecting
important aspects of interpersonal functioning (Groth-Marnat, 2009).
In addition to the standard CPI scales, the CPI-434-PPSR report
This article was published Online First February 4, 2021.
https://orcid.org/0000-0002-7720-922X
Paul Detrick
Ryan M. Roberts
https://orcid.org/0000-0001-5611-3520
Paul Detrick provides psychological assessment services to numerous law
enforcement agencies in the St. Louis area. He is board certified in Police &
Public Safety Psychology.
Ryan Roberts provides psychological assessment services to numerous
law enforcement agencies throughout the U.S. and authors specialized test
reports for public safety selection. He is a co-owner of Johnson, Roberts, &
Associates, Inc., a licensee of the Myers-Briggs Company, and receives
royalties on sales of the California Psychological Inventory Police and Public
Safety Report.
Correspondence concerning this article should be addressed to Paul
Detrick, Florissant Psychological Services, 701 St. Francois, Florissant,
MO 63031, United States. Email: [email protected]
176
177
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
POLICE APPLICANT RESPONSE BIAS ON THE CPI
(d = .47), and Self-Control (d = .40) scales from the CPI were
predictive of disciplinary actions. A meta-analysis undertaken by
Varela et al. (2004) yielded modest but statistically significant
relationships between personality test scores, (California Psychological Inventory, Minnesota Multiphasic Personality Inventory,
and Inwald Personality Inventory) and job performance, with the
strongest association demonstrated for the CPI with an observed
uncorrected correlation coefficient of .16 reported. Finally, associations between pre-hire CPI and prorated Minnesota Multiphasic
Personality Inventory-2-Restructured Form (MMPI-2-RF) scale
scores and supervisor ratings were reported by Roberts et al.
(2019). For the CPI-434-PPSR prediction equations, Probability
of Involuntary Departure was most robustly associated with a
variety of performance-related problems. Dominance (Do) and
Independence (In) were positively associated with conflict management and excessive use of force issues. Tolerance (To), Hostility
(Hos), Integrity (Itg), and Probability of Involuntary Departure were
associated with problematic relationships with citizens. Capacity
for Status (Cs) was associated with communication and problemsolving. Socialization (So), To, Itg, Amicability (Ami), and Hos all
exhibited significant associations with various additional performance issues. CPI and MMPI-2-RF scores appeared to complement one another and provide incremental validity over either
scale used alone.
Despite this support for the construct validity of the CPI for use in
police officer selection, one potential threat, in the form of positive
response bias, is present under the high demand conditions of police
officer selection. Notably, a nationwide sample conducted by Corey
(2016) revealed that approximately 15% of police applicants are
failed by psychological screeners. Under such high demand conditions, applicants have an incentive to create a good impression by
selecting responses that highlight positive attributes, matching
responses to the perceived job demands, and exhibiting attributes
of an “ideal” employee (Leary & Kowalski, 1990; Schmit & Ryan,
1993). Such response bias appears common among job applicants
and is facilitated by the transparency of items on personality
inventories and the non-verifiability of responses (Donovan et al.,
2003; Rosse et al., 1998). Some findings support the concern that
applicants that engaged in response bias have a disproportionately
higher probability of being hired (Donovan et al., 2003). Other
evidence suggests that response bias does not significantly attenuate
the associations between personality measures and job performance
(Barrick & Mount, 1996; Hough & Ones, 2002; Ones et al., 1996).
Nevertheless, response bias continues to be a topic of concern in the
context of personnel selection in particular and appears to be more
complex than initially anticipated (Dilchert et al., 2006).
We know that police officer applicants engage in positive response
bias (Griffith et al., 2007). Detrick et al. (2010), using a within-subject
repeated measure design, found that when police officer applicants
were administered the NEO Personality Inventory-Revised (NEO
PI-R) under the high demand conditions of personnel selection they
reported significantly lower scales scores on the domain Neuroticism
(plus associated facets Anxiety, Angry Hostility, Depression, and
Self-Consciousness), higher Agreeableness domain scores (and facet
scores for Trust, Straightforwardness, and Compliance), higher scores
on the three facets of the Conscientiousness domain generally
associated with dependability (Dutifulness, Self-Discipline, and
Deliberation), higher scores on two facets of Extraversion (Warmth
and Assertiveness), and lower scores on Fantasy (Openness
Domain) than those scores generated under low demand conditions
(no contingencies) approximately 6 months later.
Similarly, response bias by police officer applicants in the form of
underreporting on a measure of psychopathology, the MMPI-2-RF,
has been illustrated using the same repeated measure/within-subject
design (Detrick & Chibnall, 2014). Police officer applicants exhibited underreporting (response bias) on higher-order scales
Emotional/Internalizing Dysfunction and Behavioral/Externalizing Dysfunction scales; 5 of Restructured Clinical scales; 6 of 9
Internalizing scales; 3 of 4 Externalizing scales; and 3 of 5
Personality Psychopathology 5 scales. L−r was predictive of
demand-related underreporting of Behavioral/Externalizing scales
while K−r predicted underreporting on Emotional/Internalizing
scales. The effect sizes for differences between many of the RC
scales as a function of demand (ranging from R2 = .16 to .34)
previously found to be associated with a wide array of performance
problems (Sellbom et al., 2007) suggested a potential underestimation of performance problems.
The present study illustrates the impact of response bias by police
officer applicants on the CPI-434-PPSR using a repeated measures/
within-subject design that naturalistically modifies demand conditions. This design allows for a direct measurement of response bias
as opposed to inference of response bias from a proxy measure such
as a validity scale (Griffith et al., 2007). This design also allows for
investigation of the construct validity of both measures of response
bias on the CPI, the Gi scale and the Fake Good Index.
Method
Participants
Police officer recruits (N = 50) from a large mid-western urban
police department and statewide law enforcement agency participated in this study. There were 42 men (84%) and 8 women (16%)
with a mean age of 26 years. With respect to race-ethnicity, there
were 37 self-identified non-Hispanic Caucasians (74%), 10 African Americans (20%), and 3 other race-ethnicity (6%). In order to
maximize sample size for the race-ethnicity variable the African
American and other groups were collapsed into a single group of
13 minority recruits (26%) of which the majority (77%) was
African American. All participants had successfully passed a
background investigation and written exam and received a conditional offer of employment contingent on passing physical and
psychological examinations. All participants then successfully
passed a physical examination and preemployment psychological
evaluation and completed approximately 4 months of police
academy training.
Measures, Design, and Procedures
Participants completed the CPI-434 (Gough & Bradley, 1996)
which was scored using the CPI Police and Public Safety Selection
Report (CPI-434-PPSR; Roberts et al., 2018). Extensive test and
report data are provided in the respective manuals. Non-gendered T
scores based on the general U.S. population normative sample were
utilized as computed by the scoring program. This was a naturalistic
repeated measures design in which two sets of CPI data were
collected from the same group of participants representing two
police academy classes. The first administration was a group
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
178
DETRICK AND ROBERTS
administration that was completed in a classroom setting as one
element in a comprehensive preemployment evaluation process that
included administration of the NEO-PI-R, (Costa & McCrae, 1992),
MMPI-2-RF (Ben-Porath & Tellegen, 2008/11), CPI, an alcohol use
checklist, personal history questionnaire, and semi-structured interview. Applicants were unaware that the CPI was not actually being
used as part of the evaluation. This administration was considered
the high demand condition. Participants had previously received a
conditional offer of employment but were required to pass the
preemployment psychological evaluation for admission to the police
academy, thus were highly incentivized to present themselves in a
positive light. Additional background, interview, and test data were
collected at this time and were considered as part of the psychological preemployment evaluation process. Study participants consisted
of those applicants successfully completing the preemployment
evaluation as well as approximately 4 months of the police academy
program. After approximately 4 months in the police academy, these
same recruits were asked to again complete the CPI-434 in a
classroom group setting. This second administration was considered
the low demand condition as the recruits were informed that the
assessment results were to be used for research purposes only, would
have no consequences of any kind, and would not be revealed to any
police department personnel. The time period between the high
demand and low demand administrations was approximately 4
months. Paired sample t-tests were used to determine whether there
was any significant difference in means scale scores between the low
and high demand samples. For those scales found to exhibit
significant differences in the different demand conditions, correlational analyses of scale change scores were run to assess the
relationship between change in Gi and Fake Good Index and change
in substantive scale scores. Because the study sample exhibits
significant range restriction in both the low and high demand
conditions as compared to the U.S. general population norms for
the CPI (see Table 1), corrected Cohen’s d statistics for attenuation
due to range restriction were also calculated to provide the best
estimate of the true effect of response bias on the instrument.
deviations for the CPI-434-PPSR scales and risk statements under
high and low demand conditions (paired t tests with Cohen’s d).
Cohen’s d statistics corrected for attenuation due to range restriction were also calculated based on methods used in Sarchione et al.
(1998) and Hunter and Schmidt (1990). Significant demand effects
were noted for the Good Impression and Fake Good Index validity
scales/indices; Risk Statements Involuntary Departure, Poorly
Suited, Anger Problems, and Job Problems; Folk Scales Dominance Do, Responsibility (Re), So, Self-Control (Sc), Well-Being
(Wb), To, Achievement-via-Conformance (Ac), Intellectual Efficiency (Ie), Psychological Mindedness (Py); Vector Scales Rule
and Norm Violating/Following and Ego Integration (v2); and Special
Purpose and Research Scales Optimism (So1), Work Orientation
(Wo), Managerial Potential (Mp), Leadership (Lp), Law Enforcement Orientation (Leo), Amicability (Ami), Narcissism (Nar), and
Hostility (Hos.) Corrected effect sizes ranged from |.52| to |1.76|.
Change scores were computed by subtracting low demand scores
form high demand scores. Change scores between the two validity
measures were statistically significant at r = .87. The change scores
for Gi and the Fake Good Index were correlated with the change
scores for substantive scales and risk statements shown to have
significant demand effects to investigate their association. It was not
possible to calculate corrections for attenuation due to range restriction for the correlational analyses due to the lack of a reference
population of scale change scores. Of the 23 correlational analyses
concerning Gi, eight were significant ( p < .05), ranging from |.31| to
|.72|. Of the 23 correlational analyses concerning Fake Good, four
were significant. As shown in Table 3, demand-related change in Gi
was significantly correlated with only a subset of scales and risk
statements shown to have demand-related effects. Notably, among
the Folk Scales only Socialization (Sc) and Tolerance (To) and
among Special Purpose and Research Scales, only Ami showed
statistically significant associations with changes in Gi. For Fake
Good Index, a smaller subset showed statistically significant associations, though change in Do was significantly associated with
change in Fake Good Index whereas its correlation with Gi was not.
Results
Discussion
Table 1 displays means and standard deviations for the CPI-434PPSR Police Officer Applicant Comparison Group (a high demand
sample), and this study’s high demand and low demand conditions. A
comparison of the high and low demand tables for this sample with
that of the CPI-434-PPSR comparison group shows that both the
comparison group sample and high demand sample shows constricted
standard deviations and means shifted towards a more favorable
representation. The Good Impression and Fake Good Index scales
were higher under high demand conditions than low demand conditions. All of the Risk Statement scales were higher under Low than
High demand conditions. All of the Folk Scales were higher under
High as opposed to Low demand conditions. All of the Special
Purpose and Research scales were higher under High as opposed to
Low demand condition, except Narcissism, Hostility, and Anxiety;
these scales were higher under Low demand condition.
Paired t-tests were used to compare CPI-PPSR prediction
equations and each CPI scale under high versus low demand
condition. A Bonferroni-adjusted significance level of p ≤ .001
was used due to the number of analyses. Of the 46 comparisons, 23
were significant ( p ≤ .001). Table 2 displays means and standard
These findings are consistent with studies on police officer applicants conducted using the NEO and MMPI-2-RF instruments (Detrick
et al., 2010; Detrick & Chibnall, 2014). The paired sampled t-tests
clearly support the hypothesis that police officer applicants engage in
significant levels of response bias in response to the high stakes testing
environment of a job application. In particular, the effects of response
bias were particularly notable among the risk statements, as well as in
the scales associated with self-management (Re, So, Sc, Wb, To) and
work-related constructs (Wo, Mp, Lp, Leo, Ami). Notably, in correcting for restriction in range, nearly all effect sizes were affected to a
meaningful degree, moving from a small to medium or medium to
large effect size. Of statistically significant mean differences, 21 of 25
corrected effect sizes were large with the remaining four were medium
in magnitude. This pattern is consistent with that noted by Detrick and
Chibnall (2014) regarding the MMPI-2-RF where significant underreporting was noted on the higher-order scales Behavioral/Externalizing Dysfunction (BXD) and Emotional/Internalizing Dysfunction
(EID). This pattern also matches the expectation that applicants to
police officer positions would seek to maximize traits they believe
police officers should possess (self-control and work ethic) and
179
POLICE APPLICANT RESPONSE BIAS ON THE CPI
Table 1
Descriptive Statistics of CPI-434-PPSR Scales and Risk Statements in the Police Officer Comparison Group, High Demand Condition, and
Low Demand Condition
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Scale/risk statement
CPI-434-PPSR Police Officer
Applicant Comparison Group
Low demand (LD)a
High demand (HD)a
69 (9)
57 (4)
66.1 (7.2)
56.5 (3.0)
74.2 (6.3)
59.7 (2.4)
13 (7)
27 (18)
33 (13)
13 (8)
17 (10)
38 (16)
26 (16)
39 (13)
16.2 (10.2)
38.9 (21.1)
35.1 (10.6)
11.4 (5.4)
18.6 (7.3)
44.8 (11.9)
33.1 (8.8)
38.9 (10.5)
11.2 (5.5)
24.3 (13.3)
30.1 (11.9)
9.6 (5)
16.1 (8.1)
37.5 (12.3)
31.3 (9.4)
32.3 (9.2)
62 (7)
58 (6)
59 (7)
56 (7)
56 (6)
61 (5)
58 (8)
58 (6)
56 (6)
63 (8)
54 (7)
62 (4)
60 (7)
62 (6)
61 (7)
59 (6)
60 (6)
50 (9)
60.1 (6.9)
57.1 (6.3)
56.5 (6.9)
53 (6.5)
57.1 (6.4)
59.7 (6.6)
55.5 (8)
53.6 (6.8)
53.3 (6.4)
60.8 (7.8)
47.5 (15.5)
57.8 (8.4)
56.8 (6.7)
58 (6.7)
59.1 (6.2)
55.7 (6.4)
57 (5)
49.1 (8.4)
64.5 (5.3)
58.8 (4.4)
58.2 (5.3)
53.6 (5.5)
57.3 (5.3)
60.3 (5)
57.9 (6.6)
58.6 (6.7)
56.6 (5.6)
66.6 (5.6)
52.5 (7.6)
62.3 (3)
59.8 (6.9)
63.5 (5.4)
61.2 (5.6)
59 (5.1)
60.2 (5.7)
48.4 (6.8)
15 (5)
27 (4)
46 (7)
14.6 (4.5)
26.9 (4.2)
42.3 (6.8)
14.1 (3.8)
28.8 (3.2)
47.2 (5.7)
58 (5)
54 (7)
55 (7)
42 (8)
57 (8)
63 (6)
65 (7)
64 (5)
70 (7)
62 (7)
44 (9)
37 (8)
41 (5)
54.2 (5.9)
54 (6.5)
57 (7.7)
38.1 (7.1)
57 (8.8)
58.6 (7.3)
58.6 (7.3)
61.7 (6.6)
70.4 (9.4)
57.2 (8.5)
46.3 (8.6)
41.8 (8.3)
41.5 (6.4)
57.6 (4.1)
54.8 (6.9)
59.9 (5.8)
39.2 (7)
59.4 (7.2)
64.9 (4.5)
65.8 (6.1)
66.4 (3.6)
76 (5.3)
62.8 (5.6)
42.6 (8.2)
37.4 (7.5)
40 (3.8)
Validity
Good Impression
Fake Good Index
Risk statements
Involuntary Departure
Poorly Suited
Drug Abuse
Drug Problems
Alcohol Problems
Anger Problems
Integrity Problems
Job Problems
Folk Scales
Dominance
Capacity for Status
Sociability
Social Presence
Self-Acceptance
Independence
Empathy
Responsibility
Socialization
Self-Control
Communality
Well-Being
Tolerance
Achievement-via-Conformance
Achievement-via-Independence
Intellectual Efficiency
Psychological Mindedness
Flexibility
Vector scales
Internality/Externality
Rule and Norm Violating/Following
Ego Integration
Special Purpose and Research Scales
Optimism (So1)
Self-Discipline (So2)
Favorable Memories of Family and Childhood (So3)
Interpersonal Awareness and Situational Sensitivity (So4)
Integrity
Work Orientation
Managerial Potential
Leadership
Law Enforcement Orientation
Amicability
Narcissism
Hostility
Anxiety
Note. CPI-434-Police Officer Comparison Group for Applicants is displayed. The Fake Good Index, risk statements, v2, and v3 scores displayed are raw.
a
Values are M (SD).
minimize those associated with problematic behaviors (risk
statements).
Specifically, within the risk statements, the largest effect sizes were
noted on Poorly Suited, Involuntary Departure, and Job Problems.
Notably, among the scales falling under the broad domain of “dealing
with others,” the only scale that evidenced significant underreporting
was Do. This suggests that while applicants generally do not expect
that candidates will be screened out for attributes such as timidity or
poor social skills, they do believe that appearing to be more
dominant and authoritative will benefit their chances of passing the
psychological evaluation. Within the domain of “self-management,”
half of the scales showed significant underreporting including
Responsibility, Self-Control, and Well-Being. In addition to these
scales, both Achievement-via-Conformance and Ego Integration
were also significantly affected by response bias. As expected, Special
Purpose and Research Scales in the “work-related” domain including
180
DETRICK AND ROBERTS
Table 2
Demand Condition Comparisons of CPI-434-PPSR Scales and Risk Statements
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Scale/risk statement
Validity
Good Impression
Fake Good Index
Risk statements
Involuntary Departure
Poorly Suited
Drug Abuse
Drug Problems
Alcohol Problems
Anger Problems
Integrity Problems
Job Problems
Folk Scales
Dominance
Capacity for Status
Sociability
Social Presence
Self-Acceptance
Independence
Empathy
Responsibility
Socialization
Self-Control
Communality
Well-Being
Tolerance
Achievement-via-Conformance
Achievement-via-Independence
Intellectual Efficiency
Psychological Mindedness
Flexibility
Vector Scales
Internality/Externality
Rule and Norm Violating/Following
Ego Integration
Special Purpose and Research Scales
Optimism (So1)
Self-Discipline (So2)
Favorable Memories of Family and Childhood (So3)
Interpersonal Awareness and Situational Sensitivity (So4)
Integrity
Work Orientation
Managerial Potential
Leadership
Law Enforcement Orientation
Amicability
Narcissism
Hostility
Anxiety
Low demand (LD)a
High demand (HD)a
t (Cohen’s d/corrected d) HD versus LD
66.1 (7.2)
56.5 (3.0)
66.1 (7.2)
59.7 (2.4)
−9.9 (−1.2/−1.76)*
−8.2 (−1.2/−1.65)*
16.2 (10.2)
38.9 (21.1)
35.1 (10.6)
11.4 (5.4)
18.6 (7.3)
44.8 (11.9)
33.1 (8.8)
38.9 (10.5)
11.2 (5.5)
24.3 (13.3)
30.1 (11.9)
9.6 (5)
16.1 (8.1)
37.5 (12.3)
31.3 (9.4)
32.3 (9.2)
4.2 (0.6/1.07)*
6.4 (0.83/1.21)*
2.8 (0.44/0.69)
2.1 (0.34/0.94)
2.1 (0.32/0.59)
4 (0.6/0.97)*
1.2 (0.2/0.36)
4.6 (0.67/1.16)*
60.1 (6.9)
57.1 (6.3)
56.5 (6.9)
53 (6.5)
57.1 (6.4)
59.7 (6.6)
55.5 (8)
53.6 (6.8)
53.3 (6.4)
60.8 (7.8)
47.5 (15.5)
57.8 (8.4)
56.8 (6.7)
58 (6.7)
59.1 (6.2)
55.7 (6.4)
57 (5)
49.1 (8.4)
64.5 (5.3)
58.8 (4.4)
58.2 (5.3)
53.6 (5.5)
57.3 (5.3)
60.3 (5)
57.9 (6.6)
58.6 (6.7)
56.6 (5.6)
66.6 (5.6)
52.5 (7.6)
62.3 (3)
59.8 (6.9)
63.5 (5.4)
61.2 (5.6)
59 (5.1)
60.2 (5.7)
48.4 (6.8)
−5.4 (−0.72/−1.17)*
−2.2 (−0.32/−0.58)
−2 (−0.29/−0.47)
−0.5 (−0.09/−0.14)
−0.2 (−0.03/−0.05)
−0.6 (−0.1/−0.16)
−2.1 (−0.33/−0.45)
−5.7 (−0.73/−1.08)*
−3.8 (−0.55/−0.91)*
−7.7 (−0.85/−1.25)*
−2.2 (−0.41/–)
−4.3 (−0.71/−1.13)*
−3.7 (−0.44/−0.64)*
−7.3 (−0.89/−1.46)*
−2.9 (−0.36/−0.61)
−5.2 (−0.57/−0.99)*
−4.3 (−0.59/−1.11)*
0.6 (0.09/0.12)
14.6 (4.5)
26.9 (4.2)
42.3 (6.8)
14.1 (3.8)
28.8 (3.2)
47.2 (5.7)
1.1 (0.13/0.19)
−3.8 (−0.52/−0.77)*
−6.8 (−0.77/−1.19)*
54.2 (5.9)
54 (6.5)
57 (7.7)
38.1 (7.1)
57 (8.8)
58.6 (7.3)
58.6 (7.3)
61.7 (6.6)
70.4 (9.4)
57.2 (8.5)
46.3 (8.6)
41.8 (8.3)
41.5 (6.4)
57.6 (4.1)
54.8 (6.9)
59.9 (5.8)
39.2 (7)
59.4 (7.2)
64.9 (4.5)
65.8 (6.1)
66.4 (3.6)
76 (5.3)
62.8 (5.6)
42.6 (8.2)
37.4 (7.5)
40 (3.8)
−4.3 (−0.68/−1.33)*
−0.9 (−0.12/−0.18)
−3.1 (−0.42/−0.62)
−1 (−0.16/−0.23)
−2.3 (−0.3/−0.37)
−7.7 (−1.03/−1.68)*
−6.2 (−0.7/v1.01)*
−6 (−0.9/−1.68)*
−4.5 (−0.73/−0.96)*
−5.7 (−0.79/−1.09)*
3.6 (0.44/0.52)*
4.3 (0.55/0.69)*
1.6 (0.28/0.54)
Note. The Fake Good Index, risk statements, v2, and v3 scores displayed are raw. A corrected effect size was not calculated for Communality ( Cm) due to lack
of range restriction evidenced by the pooled standard deviation of the sample.
a
Values are M (SD).
* Denotes comparisons significant at the p < .001 level.
Work Orientation, Managerial Potential, Leadership, Law Enforcement Orientation, and Amicability show significant response bias in
the positive direction.
The implication of these results is that applicants appear to believe
that they stand a better chance of passing the psychological evaluation if they are dominant, rule-following, self-controlled, psychologically fit, team-oriented, motivated, and a leader. While these are
all positive attributes that would not hurt their chances of passing
a psychological evaluation, not all of these traits are typically
important to the screening psychologist’s recommendation. The
mismatch between patterns of typical response bias and the
evaluation criteria has concrete implications for test interpretation
in the preemployment screening context.
Psychologists conducting suitability determinations for police
applicants are typically providing a screen-out function. In this
role, they look for indicators of future negative work performance,
POLICE APPLICANT RESPONSE BIAS ON THE CPI
Table 3
Demand Condition Correlations of CPI-434-PPSR Scales and Risk
Statements
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Scale/risk statement
Validity
Good Impression
Fake Good Index
Risk statements
Involuntary Departure
Poorly Suited
Integrity Problems
Job Problems
Folk Scales
Dominance
Responsibility
Socialization
Self-Control
Well-Being
Tolerance
Achievement-via-Conformance
Intellectual Efficiency
Psychological Mindedness
Vector Scales
Rule and Norm Violating/Following
Ego Integration
Special Purpose and Research Scales
Optimism (So1)
Work Orientation
Managerial Potential
Leadership
Law Enforcement Orientation
Amicability
Narcissism
Hostility
r
1.0*
.87*
.87*
1.0*
−.03
−.34*
−.57*
−.72*
.06
−.22
−.20
−.57*
.22
.18
.30
.50*
−.07
.31*
.22
.16
−.07
.41*
.23
.23
.28
.18
.23
.25
.15
−.11
.32*
.38*
.38*
.31*
−.03
.16
.26
.14
−.14
.38*
−.27
−.25
−.16
.03
.17
.17
−.07
.24
−.14
−.18
* denotes significance at the p < .05 level.
rather than the presence of positive attributes. Holding other variables
constant, an applicant with merely average scores on Leadership and
Work Orientation scales, for example, should fare no better in this
screen-out evaluation than one with exceptionally high scores on
these scales. CPI scales addressing interactions with other people,
on the contrary, are apparently not seen by applicants as affecting
the evaluation outcome by applicants. In reality, suitability determinations by psychologists specifically address these concerns.
For example, police officer essential job functions reported by the
California POST job analysis include assertiveness, social competence, and teamwork (Spilberg & Corey, 2019). Additionally,
low scores on these scales are also associated with negative on the job
outcomes such as obtaining poor ratings related to decision-making,
controlling conflict, reliability, relationships with co-workers, and
excessive use of force (Roberts et al., 2018). The practical implication
of the lack of influence from response bias on these scales (with the
exception of Do) is that positive scores on these scales may be
interpreted by the screening psychologist with less concern that
situational demand makes the scores appear more favorable than
they otherwise might be.
While a number of CPI risk statements and scales showed
significant response bias in this study, a smaller subset was shown
to have change scores that significantly correlated with the change
score of Gi, and yet smaller subset with the change score of Fake
Good Index. Of the risk statements and scales showing significant
demand effects, only Poorly Suited, Integrity Problems, Job
181
Problems, Sc, To, Rule and Norm Violating/Following (v2),
Ego Integration (v3), and Ami change scores were significantly
correlated with Gi change scores. Of these, only Job Problems, v2,
and v3 change scores were significantly correlated with Fake Good
Index change scores. Change in Fake Good was however significantly associated with change in Do, whereas the relationship with
Gi was not significant. Thus, Gi within the context of police officer
applicants has modest construct validity and Fake Good Index has
somewhat limited construct validity. In practice, assessors must be
aware of the patterns of CPI risk statements and scales showing
demand-induced effects. Scales and risk statements showing no
significant demand affects can be interpreted with less concern
about the potential effects of response bias. For those risk statements and scales showing significant demand-related effects, the
subset whose change scores showed to be significantly correlated
with change in Gi or Fake Good Index can be interpreted with
additional insight based on the elevation of Gi or the Fake Good
Index. For all other scales
