See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/6538930

Malingering on the Personality Assessment Inventory: Identification of Specific

Feigned Disorders

Article  in  Journal of Personality Assessment · March 2007

DOI: 10.1080/00223890709336833 · Source: PubMed

CITATIONS READS
24 2,676

4 authors, including:

Christopher J Hopwood Leslie C Morey

Michigan State University Texas A&M University
168 PUBLICATIONS   4,047 CITATIONS    262 PUBLICATIONS   13,005 CITATIONS   

SEE PROFILE SEE PROFILE

Kenneth W Sewell
Oklahoma State University - Stillwater
102 PUBLICATIONS   3,544 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Development and Validation of a Self-Report Scale for Measuring Subjective Emptiness View project

CLPS Study View project

All content following this page was uploaded by Christopher J Hopwood on 21 July 2014.

The user has requested enhancement of the downloaded file.

 JOURNAL OF PERSONALITY ASSESSMENT, 88(1), 43–48
HOPWOOD, MOREY,
PAI FEIGNING
ROGERS, SEWELL Copyright © 2007, Lawrence Erlbaum Associates, Inc.

Malingering on the Personality Assessment Inventory:

Identification of Specific Feigned Disorders
Christopher J. Hopwood and Leslie C. Morey
Department of Psychology
Texas A&M University

Richard Rogers and Kenneth Sewell

Department of Psychology
University of North Texas

Profile validity is a primary consideration in the clinical assessment of psychopathology. Sev-

eral indicators of negative impression management have been developed for the Personality
Assessment Inventory (PAI; Morey, 1991) that can both indicate its presence and help differen-
tiate effortful negative distortion from distortion arising from cognitive sets associated with
psychopathology. In this study, we tested a method designed to delineate the specific Clinical
scales relevant for interpretation of deliberately feigned disorders. We used associations be-
tween the Negative Impression Management (NIM) scale and Clinical scales in the normative
standardization sample to derive NIM predicted scale scores in a regression framework. We
contrasted these predicted scores with observed scores on Clinical scales to yield NIM pre-
dicted discrepancies hypothesized to identify those Clinical scales most salient for the interpre-
tation of negative distortion. We found this method to be effective in identifying particular dis-
tortion on the relevant scales for individuals attempting to feign 3 specific diagnoses (major
depressive disorder, generalized anxiety disorder, and schizophrenia).

An important first step in the interpretation of clinical data in- major depressive disorder (MDD), present with cognitive dis-
volves the validity of reported information. Many multiscale tortions that lead them to overrepresent the negative aspects of
self-report instruments include scales and methods designed themselves, their environment, and their future. Second, indi-
to identify random, positive, and negative distortion that can viduals may also consciously overreport particular problem-
influence the accuracy of clinical inferences. The severity of atic aspects of themselves or their environment. For example,
distortion, typically indicated by comparisons of scores with individuals may feign a disorder for external incentives such
standardization samples, indicates the extent to which the cli- as attention, treatment services, or other considerations.
nician should consider the effects of distortion in their inter- The PAI includes several indexes of negative distortion,
pretations of other scale scores and if substantive scales the most direct of which is the Negative Impression Manage-
should be interpreted at all. Previous research on profile va- ment (NIM) scale. The association of NIM with a covert
lidity for multiscale inventories has focused mainly on differ- component of negative distortion that is associated with true
entiating honest from distorted responding. In this article, we psychopathology is suggested by two related facts: NIM
represent an examination of a procedure that may more pre- scores are higher in clinical than community normative sam-
cisely specify the effects of negative distortion on the Person- ples, and NIM correlates with many PAI Clinical scales.
ality Assessment Inventory (PAI; Morey, 1991) once it has However, the NIM scale has also been found effective in dis-
been established that such distortion is likely to be present. criminating feigning from honest responders, which indi-
cates that individuals who try to “fake bad” will produce
NEGATIVE DISTORTION ON THE PAI elevated NIM scores (Calhoun, Earnst, Tucker, Kirby, &
Beckham, 2000; Morey, 1991; Morey & Lanier, 1998; Rog-
Negative distortion on self-report measures can occur for two ers, Ornduff, & Sewell, 1993). Generally, extreme elevations
reasons. First, individuals with many forms of genuine on NIM scores are associated with feigning, although indi-
psychopathology, such as borderline personality disorder or viduals who attempt to feign relatively low-severity disor-
44 HOPWOOD, MOREY, ROGERS, SEWELL
ders, such as generalized anxiety disorder (GAD) or built into the PAI scoring and interpretation software as an
dysthymic disorder, often do not produce marked NIM ele- overlay to the observed profile (Morey, 1999).
vations (Bagby, Nicholson, Bacchiochi, Ryder, & Bury, When indicators such as the RDF suggest probable efforts
2002; Rogers, Sewell, Morey, & Ustad, 1996). to feign psychopathology, a discrepancy between an ob-
Several methods can augment NIM in disentangling served Clinical scale and the score predicted by NIM pre-
feigning from covert sources of negative distortion. Re- dicted score (when observed > NIM predicted) may be
gardless of the specific method being employed, consider- hypothesized to indicate specific areas that the individual is
ation of the assessment context is important, as base rates trying to feign. For example, an individual endeavoring to
of feigning are directly associated with the opportunity for feign schizophrenia is likely to endorse bizarre symptoms
unwarranted gains. Other indexes of effortful negative dis- such as those that occur on NIM and thus receive an elevated
simulation have also been developed for the PAI (Morey, NIM score as well as an elevation on RDF. They are also
1996; Rogers et al., 1996). One that holds particular prom- likely to report significant symptomatology on other Clinical
ise is the Rogers Discriminant Function (RDF; Rogers et scales, which would be more or less anticipated from the
al., 1996). This function uses several scales from the profile NIM elevation. However, given the motive to appear schizo-
to discriminate individuals who feign from genuine pa- phrenic, this person would be likely to endorse most of the
tients; it has demonstrated an impressive effectiveness items that they could identify as comprising that scale and
across several simulation samples (Bagby et al., 2002; thus achieve a score on the Schizophrenia (SCZ) scale much
Morey & Lanier, 1998; Rogers et al., 1996), although it ap- higher than would be predicted by NIM alone. Conversely,
pears to be less effective in criminal-forensic settings (Rog- scores on other scales would be anticipated to be at or below
ers, Sewell, Cruise, Wang, & Ustad, 1998) where the base levels predicted by the NIM score.
rate of pathology-free individuals may be lower than in In this study, we sought to test this hypothesis by contrast-
standard simulation studies. As the RDF is a function of ing samples of known feigners of particular disorders with
profile configuration rather than profile elevation, RDF (a) known feigners of other particular disorders and (b) sam-
mean scores are similar in clinical and community samples ples of patients manifesting the clinical diagnoses that the
(Morey, 1996). This suggests that unlike NIM, the RDF is former were attempting to feign. In particular, we hypothe-
not influenced by true psychopathology but is instead a sized that discrepancies between observed Clinical scale
more direct measure of effortful negative distortion. scores and NIM predicted Clinical scale scores in feigning
samples are likely to be greatest for the disorder the individ-
ual is attempting to feign. This discrepancy is anticipated to
be informative about the nature of the feigning over and
SPECIFIC IMPACT OF EFFORTFUL
above the actual scale score obtained by the feigner. For ex-
NEGATIVE DISTORTION
ample, in a person attempting to simulate schizophrenia, it is
not merely the fact that their PAI SCZ scale is elevated that is
This article represents an effort to examine a strategy that revealing; rather, it is that the SCZ score is elevated beyond
may allow a more precise determination of the specific im- that expected from the NIM score that is instructive about the
pact of deliberate efforts to negatively distort the PAI profile. nature of the feigning.
This strategy did not involve questions regarding the general
occurrence of negative distortion in a profile; instead, we
sought to determine what in particular about the profile is METHOD
likely to be distorted. Thus, given that the profile data (such
as the RDF) suggests that a respondent is feigning, what ar- The study involved the reanalysis of existing data sets (Rog-
eas of the profile is the respondent particularly attempting to ers et al., 1993, 1996) involving individuals with either true
distort? Morey (1999) developed a method for determining or feigned clinical diagnoses. The original researchers used a
the extent to which different scales are affected by negative simulation design and focused on three common disorders:
distortion involving a comparison of an individual’s ob- depression, generalized anxiety, and schizophrenia. System-
served scale scores with the scale scores that would be pre- atic comparisons were made between genuine patients with
dicted by their NIM elevation alone using simple linear re- these specific disorders and participants instructed to feign
gression (NIM predicted scores). We computed NIM these same disorders who were provided with incentives for
predicted scores using regression equations built with data successful dissimulation. In this investigation, we utilized
from the clinical standardization sample (Morey, 1991) in entirely original analyses to examine the usefulness of the
which each scale is predicted to be a function of the NIM ele- NIM prediction strategy in determining which particular dis-
vation. The NIM predicted method was designed to assist the order was likely to be feigned. The analyses involved dis-
evaluator in determining the extent to which scales are ele- crepancies between their observed and NIM predicted scores
vated above and beyond what would be expected given a gen- on the three relevant PAI scales: Depression (DEP), Anxiety
erally negative response set. The NIM predicted scores are (ANX), and SCZ.
 PAI FEIGNING 45
Participants that some form of feigning has occurred. With respect to the
validity of the RDF itself, the function has been cross-
We used two groups of participants for this study. The clini- validated successfully on multiple, diverse samples (e.g.,
cal sample consisted of patients from the PAI clinical stan- Bagby et al., 2002; Morey, 1996; Morey & Lanier, 1998).
dardization sample with diagnoses made by doctoral-level
clinicians who agreed to provide PAI data for the standard- Data Analyses
ization procedure (Morey, 1991). These diagnoses were
based on all available data, but clinicians were masked to PAI We computed means and standard deviations of the discrep-
results at the time the diagnoses were assigned. The feigning ancy between the observed and NIM predicted T scores on
sample was composed of individuals asked to purposefully the DEP, ANX, and SCZ scales for all participants in both
feign different forms of psychopathology (Rogers et al., samples. We tested the interaction between relevant scales
1996). (diagnostic scale for the clinical sample, feigned scale for the
feigning sample), and we tested discrepancy with a 2 × 3 × 3
Clinical sample. Morey (1991) asked practitioners mixed analysis of variance (ANOVA). The between-subject
who provided the clinical standardization data to include the factors were condition (feigned vs. clinical) and disorder
individual’s primary diagnoses. For clarity, Morey excluded (MDD, GAD, and schizophrenia) and the within-subjects
data if clients warranted more than one of the three desig- factor was PAI scales (DEP, ANX, and SCZ). We predicted a
nated disorders. As a result of this refinement, the diagnoses three-way interaction consistent with the hypothesis that the
were MDD (n = 145), GAD (n = 28), and schizophrenia (n = magnitude of discrepancy scores would be determined by the
50). The average age of clinical participants was 38.86 years disorder, the scale, and the group. We also predicted a two-
(SD = 11.54). Half of the sample (n = 111) were male and way interaction between scale and disorder after controlling
half female, and data were missing for 1 individual. A major- for the effect of the three-way interaction, as we anticipated
ity of the sample were Euro-American (n = 186); 24 were Af- that the magnitude would be highest for the scale represent-
rican American, and 13 represented other ethnic groups or ing its associated disorder (i.e., DEP–MDD, ANX–GAD,
had missing ethnicity data. The average score on NIM for the SCZ-schizophrenia) whether the individual was feigning or
clinical sample was 64.4T (SD = 16.2), and the average score experiencing a true clinical disorder. Finally, we constructed
on the RDF was –1.12 (SD = 1.10). As reported in Rogers et efficiency tables, and we computed hit rates and kappas for
al. (1996), 11.3% of the clinical sample were above the cut classifying feigning sample participants into disorder-
score for feigning on the RDF. specific subsamples using the highest discrepancy score as a
marker of simulated diagnosis.
Feigning sample. Participants in the feigning sample Given data consistent with our hypothesis, it still may be
were given incentive to successfully feign one of three spe- argued that a simpler explanation of our results is that rela-
cific disorders: MDD (n = 60), GAD (n = 57), or schizophre- tive scale elevation alone could be used to determine which
nia (n = 65). Participants in this sample varied in terms of disorder is being feigned. In other words, if the SCZ scale is
knowledge of psychopathology. Of the participants, 60 were high when the PAI indicators suggest feigning, the clinician
graduate students who had taken or were taking an advanced might simply infer that the respondent is attempting to feign
course in psychopathology, and 122 were undergraduates. Of schizophrenia rather than consult NIM-predicted scores. To
the participants, 63 were male, 115 were female, and 4 did test the incremental validity of discrepancy over scale eleva-
not report their gender. Individuals in this sample averaged tions alone, we covaried the effects of scale elevation out of
32.68 years of age (SD = 11.94), and the majority were Euro- the discrepancy scores in the feigning sample. First, we re-
American (n = 135), with smaller numbers of Hispanic (18), gressed each scale score (DEP, ANX, and SCZ) onto each
Asian (7), and African American (5) participants; 27 partici- discrepancy score, and we saved the standardized residuals.
pants were either other ethnicities or did not report their eth- These residuals represented the discrepancy after controlling
nic identity. The average score on NIM for the feigning sam- for scale score. We ran a 3 × 3 ANOVA in the feigning sam-
ple was 78.9T (SD = 22.5). Data from this sample were used ple using these three standardized residuals as the within-
in the development of RDF (Rogers et al., 1996), and thus subjects scale factor and diagnosis as the between-subject
identification rates of feigning in this sample were quite high. factor. We constructed efficiency tables, and we computed
In this sample, the average raw score on the RDF was +1.50 hit rates and kappas as we described previously.
(SD = 1.22), and 83.5% were above the cutting score for
feigning on the RDF. We note that although this sample con-
stituted the development sample for the RDF, in this study, RESULTS
our hypotheses did not address the effectiveness of the RDF
for detecting feigning; rather, the study question concerns the The means and standard deviations for the observed NIM,
effectiveness of a strategy for detecting the specific nature of ANX, DEP, and SCZ scales; the NIM predicted ANX, DEP,
the feigning when the RDF (and other indicators) suggest and SCZ scales; and the resulting discrepancy scores are de-
46 HOPWOOD, MOREY, ROGERS, SEWELL
picted in Table 1 and Table 2 for the feigning sample and also present the effect size estimates (d) for the pertinent hy-
clinical samples, respectively. Consistent with predictions, pothesized differences between observed and NIM predicted
we observed a three-way interaction, F(4, 798) = 20.04, p < scores. Effect sizes can also be computed for between-
.001, which indicated that a consideration of scale, disorder, subject effects. For example, is the discrepancy score on
and feigning versus clinical group were each important in DEP larger for individuals feigning depression than for indi-
predicting the discrepancy score. The Disorder × Scale inter- viduals feigning either of the other two diagnoses? Large ef-
action was also significant, F(4, 798) = 99.87, p < .001, fect sizes indicated that this was the case for each scale (DEP
which indicated that the scale discrepancies were largely de- d = 1.33; ANX d = 1.69; SCZ d = 0.84). Effects can also be
termined by which disorder was relevant whether feigned or compared for the same scale between feigners and genuine
consistent with the respondent’s experience. Tables 1 and 2 clinical participants. The magnitude of the relevant (e.g.,

TABLE 1
Scale, NIM Predicted, and Discrepancy Scores for DEP, ANX, and SCZ Scales Among Individuals With
Feigned MDD, GAD, or Schizophrenia Diagnoses

Disorder

MDD GAD Schizophrenia

PAI Scale T Score M SD M SD M SD

NIM 79.88 21.45 68.77 21.31 86.97 21.18

DEP 93.17 15.56 72.04 12.17 71.80 14.01
NIM predicted DEP 77.49 14.64 69.90 14.55 82.33 14.46
DEP discrepancy 15.68 16.71 2.13 13.87 –10.53 12.44
d (observed vs. predicted DEP) 1.04 0.16 –0.74
ANX 69.05 15.13 82.46 14.21 68.14 14.48
NIM predicted ANX 73.73 13.29 66.84 13.21 78.12 13.13
ANX discrepancy –4.68 11.83 15.61 17.62 –9.98 11.47
d (observed vs. predicted ANX –0.33 1.14 –0.72
SCZ 72.75 15.44 68.07 16.86 89.26 20.22
NIM predicted SCZ 73.56 15.64 65.45 15.54 78.72 15.45
SCZ discrepancy –0.81 9.58 2.62 8.86 10.54 15.53
d (observed vs. predicted SCZ) –0.05 0.16 0.59

Note. Hypothesized relevant discrepancies are underlined. Discrepancies and d values predicted to be largest are underlined. NIM = Negative Impression
Management scale; DEP = Depression scale; ANX = Anxiety scale; SCZ = Schizophrenia scale; MDD = major depressive disorder; GAD = generalized anxiety
disorder; PAI = Personality Assessment Inventory; d = Cohen’s d.

TABLE 2
Scale, NIM Predicted, and Discrepancy Scores for DEP, ANX, and SCZ Scales Among Individuals WithTrue
MDD, GAD, or Schizophrenia Diagnoses

Disorder

MDD GAD Schizophrenia

PAI Scale T Score M SD M SD M SD

NIM 65.05 15.28 53.93 9.56 68.42 19.42

DEP 76.34 16.04 56.93 9.96 66.02 15.24
NIM predicted DEP 67.36 10.43 59.77 6.53 69.66 13.26
DEP discrepancy 8.98 12.56 –2.84 8.13 –3.64 10.87
d (observed vs. predicted DEP) 0.68 –0.34 –0.26
ANX 69.76 13.77 65.89 11.62 64.68 14.33
NIM predicted ANX 64.53 9.47 57.64 5.92 66.62 12.03
ANX discrepancy 5.22 10.86 8.25 10.46 –1.94 10.65
d (observed vs. predicted ANX) 0.45 0.94 –0.15
SCZ 64.25 15.45 53.39 11.40 67.50 17.75
NIM predicted SCZ 62.74 11.15 54.63 6.97 65.20 14.16
SCZ discrepancy 1.51 10.83 –1.24 7.07 2.30 13.89
d (observed vs. predicted SCZ) 0.11 –0.13 0.14

Note. Hypothesized relevant discrepancies are underlined. Discrepancies and d values predicted to be largest are underlined. NIM = Negative Impression
Management scale; DEP = Depression scale; ANX = Anxiety scale; SCZ = Schizophrenia scale; MDD = major depressive disorder; GAD = generalized anxiety
disorder; PAI = Personality Assessment Inventory; d = Cohen’s d.
 PAI FEIGNING 47
DEP for depressed group) discrepancy score was larger in in question. Table 4 depicts the efficiency of the NIM pre-
the group feigning a diagnosis than in the clinical group re- dicted discrepancy method after we covaried scale level for
ceiving that diagnosis, with moderate effect sizes obtained the feigning sample. The hit rate (72.0%) remained signifi-
(DEP d = 0.49; ANX d = 0.48; SCZ d = 0.56). cantly above chance, Pearson χ2(4, N = 182) = 131.98, p <
Within-subjects comparisons of discrepancies are particu- .001 (κ = .58), which indicated the viability of the proposed
larly important for the determination of which scale is likely method for identifying the disorder individuals were at-
to be feigned. Table 3 compares the scale hypothesized to be tempting to feign over and above a consideration of relative
feigned (i.e., the scale that demonstrated the largest discrep- scale elevations. Table 4 also includes efficiency statistics
ancy score for a particular participant) with the actual diag- for each scale discrepancy predicting the relevant disorder.
nosis that was feigned. The convergence between predicted Although hit rates decreased slightly, the results clearly indi-
and actual feigned diagnosis occurred at a level much greater cate that the NIM predicted discrepancy method provided
than chance (hit rate = 79.7%), Pearson χ2(4, N = 182) = moderate efficiency in identifying efforts at feigning specific
184.09, p < .001 (κ = .69). The most commonly observed disorders that was above and beyond the information pro-
misclassifications involved individuals who attempted to vided by the elevation of the scales themselves.
feign GAD or MDD and occasionally achieved the highest
discrepancy scores on SCZ. Efficiency statistics of the dis- DISCUSSION
crepancy method in detecting the correct (feigned) disorders
versus one of the incorrect disorders are also included in Ta- It has been argued that feigning research should move from
ble 3. It revealed a reasonable efficiency for the procedure in general questions regarding the detection of negative distor-
determining the diagnosis being feigned. tion toward questions regarding the nature of negative distor-
As we discussed previously, it could be argued that a more tion and the relation between feigning and particular disor-
parsimonious procedure would be to simply compare scale ders. A method for answering these questions using PAI data
elevations without considering them in the context of NIM has been proposed involving a comparison of observed Clini-
predicted scores and categorize the most elevated scale as the cal scale scores and those scores predicted by NIM, a PAI
one being feigned. To test this hypothesis, we covaried the scale indicative of negative distortion. Results suggest that
effect of elevation out of discrepancy scores for the feigning this discrepancy (observed – NIM predicted scores) provides
sample by regressing each scale score onto its respective information that is useful for identifying the specific nature
NIM predicted score and saving the standardized residuals. of feigning attempts for respondents with negatively dis-
We then examined these residuals, which served as indexes torted profiles. Data indicate that the difference between an
of elevation-controlled discrepancies, as dependent variables observed scale score and a NIM predicted scale score indeed
in a 3 (between subject: feigned disorders) × 3 (within sub- helps to identify which disorders feigners were attempting to
jects: PAI scales) ANOVA for the feigning sample. The sta- simulate. When individuals intentionally distort the clinical
tistically significant interaction between scale and disorder, picture, the discrepancy tended to be quite large for the scale
F(4, 358) = 59.04, p < .001, indicated that discrepancy scores that represents the disorder they were trying to feign, and the
were larger for the scale relevant to the disorder being discrepancy for other scales was low or negative (i.e., NIM
feigned even after controlling for the elevation of the scales predicted score greater than or equal to the observed score).

TABLE 4
TABLE 3 Efficiency of Discrepancy (With Trait Level
Efficiency of Discrepancy for Identifying Statistically Accounted for) in Identifying
Diagnosis in Feigning Sample Diagnosis in Feigning Sample

Largest PAI Discrepancy Largest PAI Discrepancy

PAI PAI PAI PAI PAI PAI

Feigned Diagnosis DEP ANX SCZ PPP NPP Feigned Diagnosis DEP ANX SCZ PPP NPP

Major depressive disorder 48 3 9 .87 .91 Major depressive disorder 43 8 9 88 87

Generalized anxiety disorder 1 41 15 .87 .88 Generalized anxiety disorder 1 41 15 66 88
Schizophrenia 6 3 56 .70 .91 Schizophrenia 5 13 47 66 84
Sensitivity .80 .72 .86 Sensitivity 72 72 72
Specificity .94 .95 .79 Specificity 95 84 85
Note. Pearson χ2(4, N = 182) = 184.09, p < .001, hit rate = 79.7%, κ = .69. Note. Pearson χ2(4, N = 182) = 131.98, p < .001, hit rate = 72.0%, κ = .58.
Efficiency statistics were based on comparison of discrepancy scores Efficiency statistics were based on comparison of discrepancy scores
between relevant group and combined data from other groups. PAI = between relevant group and combined data from other groups. PAI =
Personality Assessment Inventory; DEP = Depression full scale; ANX = Personality Assessment Inventory; DEP = Depression full scale; ANX =
Anxiety full scale; SCZ = Schizophrenia full scale; PPP = positive predictive Anxiety full scale; SCZ = Schizophrenia full scale; PPP = positive predictive
power; NPP = negative predictive power. power; NPP = negative predictive power.
48 HOPWOOD, MOREY, ROGERS, SEWELL
These findings suggest a three-step process to investigat- REFERENCES
ing negative distortion on the PAI. First, the evaluator should
examine various PAI validity indicators to determine the ex- Bagby, M. R., Nicholson, R. A., Bacchiochi, J. R., Ryder, A. G., & Bury, A.
istence and extent of negative distortion. Second, the config- S. (2002). The predictive capacity of the MMPI–2 and PAI validity scales
and indexes to detect coached and uncoached feigning. Journal of Person-
uration of NIM and the RDF scores and other indexes of ality Assessment, 78, 69–86.
negative dissimulation (e.g., PAI Malingering index, indexes Calhoun, P. S., Earnst, K. S., Tucker, D. D., Kirby, A. C., & Beckham, J. C.
from other instruments, information about the assessment (2000). Feigning combat-related posttraumatic stress disorder on the Person-
context) should be consulted to assess the degree to which ality Assessment Inventory. Journal of Personality Assessment, 75, 338–350.
Morey, L. C. (1991). The Personality Assessment Inventory professional
distortion is feigned, covert, or both (Morey, 2003). If feign-
manual. Odessa, FL: Psychological Assessment Resources.
ing is suspected, NIM predicted scores should be computed Morey, L. C. (1996). An interpretive guide to the Personality Assessment In-
for each scale, and discrepancies between these scores and ventory. Odessa, FL: Psychological Assessment Resources.
observed scores should be considered. This discrepancy can Morey, L. C. (1999). PAI interpretive explorer module manual. Odessa, FL:
then be used to identify the disorder the respondent was at- Psychological Assessment Resources.
Morey, L. C. (2003). Essentials of PAI assessment. Hoboken, NJ: Wiley.
tempting to feign, with particular attention given to discrep-
Morey, L. C., & Lanier, V. W. (1998). Operating characteristics of six re-
ancies larger than 10T, as this 1 SD threshold was achieved sponse distortion indicators for the Personality Assessment Inventory. As-
for all feigning groups in this study and represents a differ- sessment, 5, 203–214.
ence unlikely to have resulted from measurement error alone. Rogers, R., Ornduff, S. R., & Sewell, K. (1993). Feigning specific disorders:
Although the results of the study suggest that the NIM pre- A study of the Personality Assessment Inventory (PAI). Journal of Per-
sonality Assessment, 60, 554–560.
dicted discrepancy method holds some promise for assisting
Rogers, R., Sewell, K. W., Cruise, K. R., Wang, E. W., & Ustad, K. L.
in the interpretation of negatively distorted profiles, certain (1998). The PAI and feigning: A cautionary note on its use in forensic-
limits to the generalizability of these results must be ac- correctional settings. Assessment, 5, 399–405.
knowledged. First, data were only available to represent a Rogers, R., Sewell, K. W., Morey, L. C., & Ustad, K. L. (1996). Detection of
limited number of feigned and true disorders, and feigning feigned mental disorders on the Personality Assessment Inventory: A
discriminant analysis. Journal of Personality Assessment, 67, 629–640.
data came exclusively from undergraduate and graduate stu-
dents. In addition, the clinical sample included relatively few
participants in whom noteworthy covert distortion might be Leslie C. Morey
suspected; as such, it was not possible to evaluate hypotheses Texas A&M University
concerning the ability of the NIM predicted discrepancy Department of Psychology
method to identify core psychopathology in such College Station, TX 77843–4235
nonfeigning individuals. More research involving the NIM Email: lcm@psyc.tamu.edu
predicted discrepancy method across a variety of feigned and
covertly distorted diagnoses with large and diverse samples Received October 29, 2005
is to both replicate and extend these results. Revised February 27, 2006

View publication stats