Professional Documents
Culture Documents
net/publication/314271785
CITATIONS READS
5 861
2 authors:
Some of the authors of this publication are also working on these related projects:
Comprehensive Assessment of Mental, Physical, and Social Health in Cirque du Soleil & National Circus School View project
All content following this page was uploaded by Stephen D Benning on 03 August 2017.
Author Note
Las Vegas.
Funding for this work was provided through a Discovery Grant from Vanderbilt University. We
extend thanks to Emily A. Dowgwillo for coordinating data collection for this study along with Karen
Miller and Alan B. Storrow for giving us access to the emergency department for data collection.
Correspondence concerning this article (and requests for an extended version of this report)
should be addressed to Stephen D. Benning, 4505 S. Maryland Pkwy., Box 455030, Las Vegas, NV
©American Psychological Association, 2017. This paper is not the copy of record and may not exactly
replicate the authoritative document published in the APA journal. Please do not copy or cite without
https://doi.org/10.1037/pas0000452
VALIDITY OF MPQ-BF INCONSISTENCY SCALES 2
Abstract
Inconsistency scales represent a promising method for separating valid and invalid personality
profiles. In a sample of 1,258 participants in the waiting room of the emergency department of an urban
university hospital, we examined whether data from participants with profiles flagged as invalid (n = 132)
using the Variable Response Inconsistency (VRIN) or True Response Inconsistency (TRIN) scales of the
Multidimensional Personality Questionnaire’s brief form (MPQ-BF) differed from those that did not
exceed any validity cutoffs (n = 1,026). Invalid profiles’ scores on many scales were less internally
consistent and had less variability than those from valid profiles, especially for random and acquiescent
response styles. Scores on MPQ-BF primary trait scales from profiles featuring random responses
appeared more psychologically maladjusted than those on valid profiles. Compared to primary trait scores
on valid profiles, acquiescent profiles generally had higher scores, and counteracquiescent profiles had
lower scores. The higher-order component structure of invalid profiles was less consistent with published
MPQ-BF component structures than that of valid profiles, though negative emotionality was generally
reasonably well-preserved. Scores on primary traits associated with negative emotionality generally had
larger correlations with demographic criteria for valid profiles than invalid profiles. These results argue
that inconsistency scales meaningfully identify invalid profiles in normal-range personality assessment.
This study showed that normal-range personality trait scores from invalid profiles (due to random, yea-
saying, or nay-saying response styles) are distorted in many ways compared to valid personality profiles.
Furthermore, scores on invalid profiles are not related to demographic characteristics like scores on valid
profiles are. These results suggest that researchers and clinicians should not interpret personality profiles
The clinical utility of scales designed to separate invalid profiles from valid profiles is
scales often demonstrate reduced criterion validity compared to invalid profiles (Caldwell-
Andrews, Baer, & Berry, 2000; Rohling et al., 2011), though some researchers hold this is not
always true (McGrath, Mitchell, Kim, & Hough, 2010; Piedmont, McCrae, Riemann, &
Angleitner, 2000). Invalid profiles identified through scales measuring the consistency of
responding are also frequently associated with weaker validity coefficients (Handel, Ben-Porath,
Tellegen, & Archer, 2010; Wetter, Baer, Berry, Smith, & Larsen, 1992). However, solely
considering rank-order associations among scores does not address important aspects of
construct validity (Mash & Hunsley, 2005). Internal consistencies of scales, mean trait scores,
and the factorial structure of multi-scale inventories may differ between valid and invalid profiles
and are likely to affect clinical utility (Swets, 1986). These statistics can determine whether
2016). True Response Inconsistency (TRIN) scales on the MMPI-2 family of broadband
psychopathology instruments (e.g., Ben-Porath & Tellegen, 2008; Butcher et al., 2001), and the
detect fixed response patterns through endorsement of oppositely keyed item pairs in the same
direction. These inventories’ Variable Response Inconsistency (VRIN) scales assess random or
items that are keyed in the same direction (though the MMPI-2’s VRIN also has some same-
direction item pairs). Inconsistent responding can elevate scores on content-based validity scales,
such as those that assess either positive or negative self-presentation (Burchett et al., 2016).
Thus, profile inconsistency is associated with other forms of potentially deceptive responding.
In this study, we examined how profiles on the brief form of the MPQ (Patrick, Curtin, &
Tellegen, 2002) that were identified as invalid compared to those that were not. The MPQ-BF
consists of 11 primary trait scales that comprise three orthogonal higher-order components.
Positive Emotionality (PEM) comprises the primary traits of Well-being (feeling cheerful and
optimistic), Social Potency (enjoys dominant roles in the spotlight), Achievement (hard-working
and ambitious), and Social Closeness (interpersonally warm with many friends). Negative
Emotionality (NEM) comprises the primary traits of Stress Reaction (feeling anxious, moody,
and tense), Alienation (suspicious of and mistreated by others), and Aggression (enjoys violent
entertainment and willing to mistreat others for an advantage). Behavioral Constraint (CON)
comprises the primary traits of Control (preference for planning instead of acting), Harm
Avoidance (preference for dull experiences over thrilling and unsafe ones), and Traditionalism
(behavior influenced by parental, societal, and religious norms). The primary trait of Absorption
(proneness to cognitive involvement in sensory and imaginative experiences) does not load on
any specific higher-order component, but it is approximately equally related to PEM and NEM.
The MPQ-BF has three additional validity scales, two of which (VRIN and TRIN, as
noted above) detect inconsistent responding. It also has an Unlikely Virtues scale that assesses
moralistic (claiming social/intellectual status and managing agentic presentation) and egoistic
(denying socially deviant faults and managing communal presentation) self-deception (Paulhus
VALIDITY OF MPQ-BF INCONSISTENCY SCALES 5
& John, 1998). Judging profile validity with scales like Unlikely Virtues does not appreciably
alter criterion-related validity or clinical judgments about patients (Barthlow, Graham, Ben-
Porath, Tellegen, & Mcnulty, 2002). Rather, scales assessing self-deception are saturated with
content-related variance associated with psychological health; statistically controlling for them
often reduces criterion-related validity of personality scores (Paulhus, 1991). Thus, many
personality theorists have argued that scales like Unlikely Virtues should not be considered
genuine validity scales (Uziel, 2014). We omitted analyses involving Unlikely Virtues and
We used standard VRIN and TRIN cutoffs to identify invalid MPQ-BF profiles from
participants in an emergency department waiting room and compare invalid MPQ-BF profiles
with valid MPQ-BF profiles from participants in the same setting. This design allowed us to
examine invalid profiles occurring in an applied setting rather than with parametrically varied
differences between these profile groups with reference to internal characteristics of the MPQ-
BF (i.e., internal consistencies, descriptive statistics, and component structure) and to the
differences in correlations among primary trait scales and demographic variables. We considered
the ability of MPQ-BF inconsistency scales to detect invalid profiles using analyses of the test
Method
Participants were 1,258 adults present in the emergency department waiting area of a
private, urban, academic hospital. A total of 94 participants (7.5%) responded to fewer than 75%
of items and were excluded from further analysis, and 6 participants did not have sufficient data
VALIDITY OF MPQ-BF INCONSISTENCY SCALES 6
to compute scores on all 11 MPQ-BF primary trait scales. Vanderbilt University’s IRB approved
Of the 1,158 complete MPQ-BF profiles, 1,026 were valid; 132 profiles (11.5%) were
invalid due to extreme validity scale scores assessing inconsistent responding. Profiles were
invalid due to T scores > 78 on VRIN (n = 37; 28.0%), TRIN T scores > 75 (n = 32; 24.2%) or <
25 (n = 27; 20.5%), or a combination of both a T score on VRIN > 70 and a T score on TRIN
either > 70 (n = 20; 15.2% or < 30 (n = 16; 12.1%). Profiles identified as invalid solely for
having high VRIN scores were analyzed as “random” profiles, profiles invalid for having high
TRIN scores were classified as “acquiescent” profiles (n = 52), and profiles invalid for having
low TRIN scores were considered “counteracquiescent” profiles (n = 43). VRIN and TRIN
scores correlated -.04 (p = .148). Our data and analysis script are provided for reproducibility.
Participants were 35.5 years of age on average (SD = 12.7), and our sample was 57%
female. The racial and ethnic composition of the sample was 57.8% White, 35.8% Black, 2.8%
Hispanic, 2.8% Asian, 1.0% Pacific Islander, and 0.8% Native American, with 0.3% opting not
to report racial information (1% reported multiple races). Participants provided data using
netbook computers they checked out from a table that a research assistant staffed in the waiting
room. These netbooks administered demographic questions listed below before the MPQ-BF. A
two-second delay was imposed between responses to reduce the likelihood of terminating the
screening quickly to obtain the $5 cash compensation provided when the netbook was returned.
Participants inputted the largest number of alcoholic drinks they had consumed within
any 24 hour period in their lifetimes, the number of alcoholic drinks they consumed on average
per week, the number of caffeinated drinks they consumed on average each day, and their current
VALIDITY OF MPQ-BF INCONSISTENCY SCALES 7
and lifetime levels of nicotine use per day. Nicotine use was coded as follows: 0 = none, 1 = 10
cigarettes or less, 2 = 11-20 cigarettes, 3 = 21-30 cigarettes, 4 = 31 cigarettes or more. They also
reported on whether they had ever received treatment for a psychiatric diagnosis, their education
level along with that of their mother and father, whether they were currently employed, and their
yearly household income. Education level was coded as follows: 1 = never attended high school,
2 = attended high school but did not graduate, 3 = graduated from high school/received GED, 4 =
attended college but did not graduate, 5 = received 2-year college degree, 6 = received 4-year
conducted independent samples t tests for scores each primary trait to examine how mean levels
of primary traits differed between those with invalid and valid profiles. We adjusted the degrees
of freedom in those analyses because many scores in the invalid profile groups had reduced
variance relative to the valid profile group according to Levene’s test (see Table 1). These
analyses had 80% power to detect population effect size ds from .40 (acquiescent responding)
Following (Patrick et al., 2002), we examined the component structures of the MPQ-BF
primary traits with principal component analyses (albeit followed by promax rotation to examine
to compare the correspondence of each component in this sample with the cross-validation
sample in (Patrick et al., 2002). Across domains, experts judge congruence coefficients of .70-.84
to represent “poor” congruence, those between .85-.94 indicate “fair” congruence, and those of .
VALIDITY OF MPQ-BF INCONSISTENCY SCALES 8
95 and above provide evidence for “good” congruence (Lorenzo-Seva & ten Berge, 2006). In
addition to hewing closely to (Patrick et al., 2002) data analytic strategy, we also chose to use
PCA instead of CFA because CFAs make strong assumptions about all non-substantive loadings
being zero that often do not hold (Borkenau & Ostendorf, 1990). Finally, we tested the
differences between valid and invalid profiles in their criterion-related correlations between
MPQ-BF primary trait scales and relevant demographic features. These analyses had 80% power
to detect a population correlation difference of .30 between valid and invalid profiles.
Results
Table 1 displays the internal consistencies, descriptive statistics, and component structure
of the MPQ-BF primary trait scales for participants with valid and invalid profiles. Compared to
primary trait scores for valid profiles, internal consistencies were significantly lower for 10/11
random profile scores, 8/11 acquiescent profile scores, and 2/11 counteracquiescent profile
scores. Likewise, variability was reduced for 8/11 random profile scores, 6/11 acquiescent profile
scores, and 2/11 counteracquiescent profile scores. Invalid profiles had substantially distorted
Absorption differed from valid profiles across all three invalid profile groups. Random profiles
had lower scores on 3/11 primary traits (Cohen’s d range [-0.34, -0.53]) and higher scores on
Alienation (Cohen’s d = 0.36) and Aggression (Cohen’s d = 0.57). Acquiescent profiles had
higher scores on 6/11 primary traits (Cohen’s d range [0.24, 0.95]) and lower scores on Social
Closeness (Cohen’s d = -.43) and Control (Cohen’s d = -0.28). Counteracquiescent profiles had
lower scores on 7/11 primary traits (Cohen’s d range [-0.53, -1.43]). When data were pooled
across invalid profile groups, results largely mirrored those for the random response style group.
VALIDITY OF MPQ-BF INCONSISTENCY SCALES 9
The structure of the MPQ-BF for valid profiles largely paralleled that in (Patrick et al.,
2002); the first six eigenvalues in this analysis were 2.80, 1.98, 1.43, 0.87, 0.79, and 0.75. All
components showed fair congruence with normative MPQ-BF data. In contrast, the structure of
the MPQ for invalid profiles differed markedly from previous component structures. The first six
eigenvalues for random profiles were 2.67, 1.73, 1.32, 1.27, 1.08, and 0.87; those for acquiescent
profiles were 2.88, 1.75, 1.30, 1.04, 0.97, and 0.82; and those for counteracquiescent profiles
were 3.42, 2.70, 1.47, 0.77, 0.64, and 0.57. Random profiles’ PEM, acquiescent profiles’ CON,
and counteracquiescent profiles’ NEM components showed fair congruence with normative data.
However, the other component scores had poor or worse congruence to normative data. When all
three invalid profile groups were analyzed together, the resultant structure was remarkably
similar to the counteracquiescent group’s (PEM rc = .98, NEM rc = .98, CON rc = .89). These
components were also more strongly correlated (inter-component rs range [.19, .39]) than those
In addition to having the most coherent component structure across invalid profile
groups, scales comprising NEM had the strongest average correlations with demographic
variables (median |r| range [.15, .17]; median |r| for other primary trait scales = .09). Thus, Table
2 gives the correlations of Stress Reaction, Alienation, and Aggression scores from valid and
invalid MPQ-BF profiles with substance use, psychiatric history, and socioeconomic variables.
Many more correlations were statistically significant in the valid profile group due to its
substantially greater power to detect correlations different from zero. Nevertheless, using a
Sidak-corrected p value of .0170 to account for analyzing 3 primary trait scales at once,
correlations in the valid profile group were more often of larger magnitude than those in the
VALIDITY OF MPQ-BF INCONSISTENCY SCALES 10
invalid profile group. This pattern was significant for Stress Reaction (binomial sign test p = .
006) and trended toward significance for Alienation and Aggression (binomial sign test ps = .
032). Four out of 33 correlations differed significantly between the groups, which was more than
would be expected by chance alone, binomial sign test p = .002. Current employment was
correlated with Stress Reaction scores in the valid but not invalid profile group, a pattern that
was also evident for the relationships between participants’ education levels and both Alienation
and Aggression. Aggression also correlated positively with lifetime cigarette use in the valid
profile group, but this correlation was negative in the invalid profile group.
Discussion
In this study, invalid MPQ-BF profiles were less internally consistent, less coherence in
the components, had lower scale variance, and had different means than valid profiles. Their
correlations with nicotine use, education, and employment were also reduced or of the opposite
direction than those for valid profiles. Our results were consistent with prior work suggesting
that invalid profiles demonstrate less factor coherence and lower internal reliability (Kam &
Meyer, 2015; Sherman, Krug, & Birenbaum, 1979) with some reduction in correlations (Archer,
Fontaine, & McCrae, 1998). Thus, inconsistency-identified profiles evidenced multiple threats to
internal validity. Additionally, invalid profiles showed substantially reduced external validity,
even though low αs need not substantially suppress criterion-related validity (Schmitt, 1996).
Though the VRIN scale does not assess specific personality constructs, random profiles
had lower scores traits associated with PEM and CON and higher scores on NEM-related traits.
This pattern of scores represents the inverse of moralistic self-deceptive responding (Paulhus &
John, 1998), indicating that random responding distorted scores in potentially substantial ways.
VALIDITY OF MPQ-BF INCONSISTENCY SCALES 11
Acquiescent responding tended to increase primary trait scores on invalid profiles, but scores on
Social Closeness and Control were lower on invalid than valid profiles. The latter two primary
traits are among the most balanced in their keying on the MPQ-BF, as most primary traits have
no more than 2/12 reverse-coded items. In contrast, 6 Social Closeness items are reverse-coded,
as are 4 Control items. Thus, true responses on reverse-coded items may have lowered their
scores. However, Harm Avoidance also has 6 reverse-coded items, and its mean was not affected
by acquiescent responding, so coding alone does not explain the associations of Social Closeness
and Harm Avoidance with lower scores during acquiescence. Counteracquiescent responding
lowered primary trait scores while preserving most scores’ internal consistencies (with the
exceptions of Social Potency and Social Closeness). Thus, each type of response style had
unique effects on trait scores (Handel et al., 2010). Clinical assessments should include separate
scales that assess random and fixed responding, unlike the PAI (Morey, 2007) and the PPI-R
(Lilienfeld & Widows, 2005), which blend both response styles into single inconsistency scales.
These results indicate that validity scales measuring the consistency of responses in
normal-range personality assessment have utility in both group and individual level decision-
making. Inconsistently endorsed profiles still had sufficient true score variance to permit
reasonable inferences to be drawn about the relationships among scores. Even so, scores from
invalid profiles typically yielded no relationships between personality traits and demographics,
whereas valid profiles demonstrated small but significant relationships with demographic
characteristics. Thus, assuming similar component structures, group-level research may need to
concern itself with the apparent consistency of responses. In addition, clinicians must make
inferences about a specific individual’s personality based on the individual’s score. These
VALIDITY OF MPQ-BF INCONSISTENCY SCALES 12
comparisons are affected by changes in the mean and shape of score distributions (Youngstrom,
2013). Inconsistency scales may help separate data from which reasonably reliable inferences
can be made from those that are psychometric dross, especially when considering random or
acquiescent response styles. It is unlikely that inferences made about scales that are saturated
with less than half of true score variance and that do not cohere would survive Daubert tests of
evidential admissibility in forensic proceedings. Nor is it likely that invalid profiles, with
reduced internal consistency and varying mean scores, would have similar levels of diagnostic
This study was conducted in the waiting room of a busy emergency department with
people who were seeking treatment or associates of individuals seeking treatment. The increased
potential for distraction from using an atypical assessment environment could have increased the
rate of profile invalidity. However, our rate of profile invalidity was not above rates seen in
undergraduate samples (Sellbom & Ben-Porath, 2005), and our sampling methodology allowed
for a larger, more diverse sample than is typically explored with the MPQ. The number of
specific invalid profiles likely means that the PCAs for each invalid profile group were under-
powered. However, the overall invalid profile analyses were well-powered (MacCallum,
Widaman, Zhang, & Hong, 1999; Mundfrom, Shaw, & Ke, 2005). On the MMPI-2-RF, scores on
inconsistency scales are related to inconsistent reporting of symptoms on interview and reduced
intelligence; VRIN scores are also associated with education, and TRIN scores are additionally
associated with symptoms of depression and anxiety (Gu, Reddy, Green, Belfi, & Einzig, in
press). Similar work should be done with the MPQ’s validity scales to investigate reasons that
invalid profiles be generated. Furthermore, we only assessed profile invalidity with the MPQ’s
VALIDITY OF MPQ-BF INCONSISTENCY SCALES 13
inconsistency scales. It is unclear the degree to which different inventories’ validity scales – even
those that ostensibly measure the same types of invalid response styles – identify the same
individuals as producing invalid profiles (Nikolova, Hendry, Douglas, Edens, & Lilienfeld,
2012). Future work should examine whether specific inventories’ validity scales are particularly
good at identifying internally and externally invalid profiles, especially with respect to clinically
relevant criteria. Finally, additional studies should include criteria targeted to the constructs each
primary trait measures to assess the differential external validity of valid and invalid profiles.
alternate measures of deceptive responding. In cases where feigning healthy is a risk, future scale
development might build on the MMPI scales that evaluate superlative self-presentation (Butcher
& Han, 1995), which assesses moralistic bias more purely (Detrick & Chibnall, 2008). The
evolution of scales designed to assess negative self-presentation may point the way to further
developments in this area. In the more typical case of negative self-deception, presenting as more
ill, the MMPI family of scales offer two alternative frameworks for scale development. The Fp
scale (Arbisi & Ben-Porath, 1995) uses infrequently endorsed items in inpatient units to detect
potential malingering in the psychiatrically ill. Conversely, the RBS scale (Gervais, Ben-Porath,
Wygant, & Green, 2007) comprises items that predict failing at least one cognitive effort test to
detect potential malingering in neurocognitively impaired patients. Outside of the MMPI family
of scales, the PPI-R Deviant Responding scale represents another method for identifying
inaccurate responding using items that are logically guaranteed to be true or false
(Marcus, Church, O’Connell, & Lilienfeld, 2016). Nevertheless, our study highlights the need to
References
Arbisi, P. A., & Ben-Porath, Y. S. (1995). An MMPI-2 infrequent response scale for use with
Archer, R. P., Fontaine, J., & McCrae, R. R. (1998). Effects of two MMPI-2 validity scales on basic scale
https://doi.org/10.1207/s15327752jpa7001_6
Barthlow, D. L., Graham, J. R., Ben-Porath, Y. S., Tellegen, A., & Mcnulty, J. L. (2002). The
https://doi.org/10.1177/1073191102009003001
Ben-Porath, Y. S., & Tellegen, A. (2008). Minnesota Multiphasic Personality Inventory-2 (Restructured
Form) manual for administration, scoring, and interpretation. Minneapolis, MN: University of
Minnesota Press.
Borkenau, P., & Ostendorf, F. (1990). Comparing exploratory and confirmatory factor analysis: A study
on the 5-factor model of personality. Personality and Individual Differences, 11(5), 515–524.
https://doi.org/10.1016/0191-8869(90)90065-Y
Burchett, D., Dragon, W. R., Smith Holbert, A. M., Tarescavage, A. M., Mattson, C. A., Handel, R. W., &
https://doi.org/10.1037/pas0000205
Butcher, J. N., Graham, J. R., Ben-Porath, Y. S., Tellegen, A., Dahlstrom, W. G., & Kaemmer, B. (2001).
Press.
VALIDITY OF MPQ-BF INCONSISTENCY SCALES 15
Butcher, J. N., & Han, K. (1995). Development of an MMPI-2 scale to assess the presentation of self in a
personality assessment, Vol. 10 (pp. 25–50). Hillsdale, NJ, England: Lawrence Erlbaum
Associates, Inc.
Caldwell-Andrews, A., Baer, R. A., & Berry, D. T. R. (2000). Effects of response sets on NEO-PI-R
scores and their relations to external criteria. Journal of Personality Assessment, 74(3), 472–488.
https://doi.org/10.1207/S15327752JPA7403_10
Detrick, P., & Chibnall, J. T. (2008). Positive response distortion by police officer applicants: Association
of Paulhus Deception Scales with MMPI-2 and Inwald Personality Inventory validity scales.
Gervais, R. O., Ben-Porath, Y. S., Wygant, D. B., & Green, P. (2007). Development and validation of a
Response Bias Scale (RBS) for the MMPI-2. Assessment, 14(2), 196–208.
https://doi.org/10.1177/1073191106295861
Gu, W., Reddy, H. B., Green, D., Belfi, B., & Einzig, S. (in press). Inconsistent responding in a criminal
forensic setting: An evaluation of the VRIN-r and TRIN-r scales of the MMPI–2–RF. Journal of
Handel, R. W., Ben-Porath, Y. S., Tellegen, A., & Archer, R. P. (2010). Psychometric functioning of the
MMPI-2-RF VRIN-r and TRIN-r scales with varying degrees of randomness, acquiescence, and
Kam, C. C. S., & Meyer, J. P. (2015). How careless responding and acquiescence response bias can
Lilienfeld, S. O., & Widows, M. R. (2005). Psychopathic Personality Inventory – Revised. Lutz, FL:
PAR.
VALIDITY OF MPQ-BF INCONSISTENCY SCALES 16
Lorenzo-Seva, U., & ten Berge, J. M. F. (2006). Tucker’s congruence coefficient as a meaningful index of
MacCallum, R. C., Widaman, K. F., Zhang, S., & Hong, S. (1999). Sample size in factor analysis.
Marcus, D. K., Church, A. S., O’Connell, D., & Lilienfeld, S. O. (2016). Identifying Careless Responding
https://doi.org/10.1177/1073191116641507
Mash, E. J., & Hunsley, J. (2005). Evidence-based assessment of child and adolescent disorders: Issues
and challenges. Journal of Clinical Child & Adolescent Psychology, 34(3), 362–379.
https://doi.org/10.1207/s15374424jccp3403_1
McGrath, R. E., Mitchell, M., Kim, B. H., & Hough, L. (2010). Evidence for response bias as a source of
https://doi.org/10.1037/a0019216
Morey, L. C. (2007). Personality Assessment Inventory professional manual (2nd ed.). Lutz, FL: PAR.
Mundfrom, D. J., Shaw, D. G., & Ke, T. L. (2005). Minimum sample size recommendations for
https://doi.org/10.1207/s15327574ijt0502_4
Nikolova, N. L., Hendry, M. C., Douglas, K. S., Edens, J. F., & Lilienfeld, S. O. (2012). The
Patrick, C. J., Curtin, J. J., & Tellegen, A. (2002). Development and validation of a brief form of the
Paulhus, D. L. (1991). Measurement and control of response bias. In J. P. Robinson, P. R. Shaver, & L. S.
Wrightsman (Eds.), Measures of personality and social psychological attitudes (pp. 17–59). San
Paulhus, D. L., & John, O. P. (1998). Egoistic and moralistic biases in self-perception: The interplay of
self-deceptive styles with basic traits and motives. Journal of Personality, 66(6), 1025–1060.
https://doi.org/10.1111/1467-6494.00041
Piedmont, R. L., McCrae, R. R., Riemann, R., & Angleitner, A. (2000). On the invalidity of validity
scales: evidence from self-reports and observer ratings in volunteer samples. Journal of
Rohling, M. L., Larrabee, G. J., Greiffenstein, M. F., Ben-Porath, Y. S., Lees-Haley, P., Green, P., &
https://doi.org/10.1037/a0023327
Schmitt, N. (1996). Uses and abuses of coefficient alpha. Psychological Assessment, 8(4), 350–353.
https://doi.org/10.1037/1040-3590.8.4.350
Sellbom, M., & Ben-Porath, Y. S. (2005). Mapping the MMPI-2 Restructured Clinical scales onto normal
personality traits: evidence of construct validity. Journal of Personality Assessment, 85(2), 179–
187. https://doi.org/10.1207/s15327752jpa8502_10
Sherman, J. L., Krug, S. E., & Birenbaum, M. (1979). Checking the reliability and validity of HSPQ
https://doi.org/10.1207/s15327752jpa4306_15
Swets, J. A. (1986). Indices of discrimination or diagnostic accuracy: Their ROCs and implied models.
Tellegen, A., & Waller, N. G. (2008). Exploring personality through test construction: Development of the
(Eds.), The SAGE handbook of personality theory and assessment, Vol 2: Personality
measurement and testing (pp. 261–292). Thousand Oaks, CA, US: Sage Publications, Inc.
VALIDITY OF MPQ-BF INCONSISTENCY SCALES 18
Uziel, L. (2014). Impression management (“lie”) scales are associated with interpersonally oriented self-
https://doi.org/10.1111/jopy.12045
Wetter, M. W., Baer, R. A., Berry, D. T. R., Smith, G. T., & Larsen, L. H. (1992). Sensitivity of MMPI-2
validity scales to random responding and malingering. Psychological Assessment, 4(3), 369–374.
https://doi.org/10.1037/1040-3590.4.3.369
https://doi.org/10.1080/15374416.2012.736358
VALIDITY OF MPQ-BF INCONSISTENCY SCALES 19
Table 1
MPQ primary trait Valid profiles (n = 1,026) Random profiles (n = 37) Acquiescent profiles (n = 52) Counteracquiescent profiles (n = 43)
Component loadings Component loadings Component loadings Component loadings
a M (SD) 1 2 3 a M (SD) 1 2 3 a M (SD) 1 2 3 a M (SD) 1 2 3
Well-being .83 49.2 (10.6) -.30 .71 .05 .58 44.7 (7.3) .03 .78 .07 .71 53.4 (8.8) .69 .05 -.19 .83 38.7 (10.4) .87 .01 .04
Social Potency .72 54.1 (7.1) -.23 .53 -.39 .43 52.4 (5.3) -.11 .12 -.60 -.01 55.3 (2.9) .17 .02 -.76 .47 53.0 (4.1) .04 -.05 .92
Achievement .74 52.8 (8.7) .05 .71 .24 .32 49.9 (5.5) -.25 .20 .56 .32 54.8 (4.1) .77 -.16 .21 .61 44.8 (7.5) .66 -.11 .39
Social Closeness .82 47.2 (9.8) -.56 .40 -.04 .30 45.0 (5.0) .10 .75 -.11 .21 43.1 (3.7) .23 -.42 -.46 .55 47.0 (6.7) .42 -.61 .07
Stress Reaction .84 53.3 (10.1) .83 -.02 .01 .45 53.4 (5.3) .79 -.01 .26 .48 61.7 (5.8) -.03 .77 .21 .72 47.6 (7.5) .20 .91 -.31
Alienation .84 63.0 (9.2) .83 .01 -.01 .20 66.2 (3.2) .75 -.07 -.18 .64 71.6 (5.3) .04 .87 -.03 .81 58.1 (8.2) .11 .85 .11
Aggression .85 53.6 (11.4) .45 .07 -.54 .62 60.1 (7.7) .65 -.49 .07 .75 64.2 (8.5) .16 .63 -.34 .81 52.2 (8.8) -.12 .65 .49
Control .69 47.1 (9.2) -.12 .21 .70 .35 44.7 (7.0) -.04 -.11 .05 .11 44.5 (4.7) .51 .28 .17 .59 40.1 (7.5) .67 -.38 .05
Harm Avoidance .70 46.0 (8.5) -.10 -.19 .59 -.08 44.1 (4.8) -.10 -.10 -.36 .48 44.0 (6.7) .16 .00 .58 .66 44.6 (7.4) .61 -.10 -.33
Traditionalism .61 47.6 (7.2) .30 .19 .67 .34 45.2 (5.8) .71 .47 .08 .15 48.6 (5.1) .36 -.07 .61 .39 41.7 (6.1) .81 .27 -.16
Absorption .71 55.5 (8.7) .48 .65 -.01 .02 50.9 (5.0) .61 .16 -.13 .41 63.6 (6.1) .68 .09 .01 .68 43.1 (9.0) .65 .40 .15
Congruence coefficients .94 .90 .94 .79 .85 .34 .80 .78 .85 .69 .92 -.64
Note. For each primary trait scale, underlined alphas or italicized means and standard deviations
differ significantly from those of valid profiles (Sidak-corrected p <.00465 to control family-
wise error at .05 across 11 primary trait scales). Component loadings |.40| or greater are
displayed in bold.
VALIDITY OF MPQ-BF INCONSISTENCY SCALES 20
Table 2
Substance Use, Psychiatric History, and Socioeconomic Status Correlates of Valid and Invalid
SR AL AG
Maximum alcoholic drinks in 1 day .11* / .09 .10 / .15 .28**/ .10
Average alcoholic drinks per week .01 / .19 .07 / .10 .19**/ .07
Average caffeinated drinks per day .17**/ .04 .18**/ .07 .19**/ .03
Current cigarette use .23**/ .17 .15**/ .03 .13**/-.04
Maximum lifetime cigarette use .26**/ .06 .17**/ .00 .13**/-.13
History of psychiatric conditions .34**/ .16 .22**/ .12 .10* / .01
Education level: self -.18**/-.06 -.34**/-.08 -.27**/-.06
Education level: mother -.15**/ .02 -.16**/ .02 -.06 / .02
Education level: father -.13**/ .06 -.15**/-.03 -.09 /-.09
Currently employed? -.18**/ .05 -.23**/-.03 -.17**/-.03
Household income -.18**/ .06 -.26**/-.08 -.17**/-.05
Note. Correlations to the left of the slash represent those for valid profiles; ns range from 685
(maximum alcoholic drinks in one day) to 1,022 (history of psychiatric conditions). Correlations
to the right of the slash represent those for invalid profiles; ns range from 57 (maximum
alcoholic drinks in one day) to 131 (history of psychiatric conditions). SR = Stress Reaction, AL
error at .05 across 11 demographic variables), ** p < .001. Underlined correlations differ