PARTICIPANTS

1
False Positive Error Rates of WCST Malingering
Formula in Head Injury Litigants and a Mixed
Clinical Sample Rael T. Lange, Ph.D.,
Riverview Hospital, Canada, Graeme J. Senior,
Ph.D., University of Southern Queensland,
Australia, Gordon J. Chelune, Ph.D., Cleveland
Clinic, Ohio., Lucille A. Douglas, Ph.D.,
Assessment Services in Psychology, Australia,
Sharron Dawes, B.Sc (Hons)., University of
Southern Queensland, Australia

• PARTICIPANTS
• Group 1 Mixed Clinical Sample
• 614 patients from a large Hospital in
  Cleveland, Ohio
• Referred as part of standard medical treatment
  and/or
• rehabilitation
• Includes 13 Clinical and 1 Suspected
  Exaggerator Group
• The 13 Clinical Groups are outlined in Table 1
• Individuals in the Suspected Exaggerator group
  were selected
• based on poor performance on the VSVT
• All individuals were not in litigation at the
  time of the
• Assessment

ABSTRACT It has been proposed that the Wisconsin
Card Sort Test (WCST) may be useful for detecting
cognitive exaggeration. Bernard et al. (1996) and
Suhr Boyer (1999) developed two malingering
formulas based on standard WCST scores, while
Greve et al. (2002) has proposed the use of
'unique' items on this test. The purpose of this
investigation was to examine the false positive
error rate of the two WCST malingering formulas
and the "unique-item" method proposed by Greve et
al. Participants were 614 hospital patients (13
diagnoses 1 suspected exaggerator group) and
105 head injury litigants. In the hospital
sample, false positive error rates ranged from 0
to 16.3 (Suhr/Boyer) and 0 to 25.6 (Bernard)
depending on clinical diagnoses. However, similar
base rates were also found in the suspected
exaggerator group (Suhr/Boyer 16.7 Bernard
8.3). When both formulas are combined, the false
positive error rate decreased in the hospital
sample (0 to 11.6), while only 5.6 of the
suspected exaggerators were identified as true
positives. In the head injury litigants,
unacceptably high false positive error rates were
found using all three formulae/methods
(Suhr/Boyer 9.5 to 18.1 Bernard 15.2
Greve 10.5 to 21.9), however, acceptable
false positive error rates were demonstrated when
any two of the formula's/methods were combined
(i.e., lt 6.7). Combining the Greve method with
either of the two malingering formula's produced
the lowest false positive error rate. These
findings provide some support for the use of
Greve et al's unique items, in combination with
existing malingering formula's, as a useful
innovation for detecting malingering on the WCST.
However, further base rate and clinical outcome
studies using suspected exaggerators are
essential before these methods can be established
as a viable means for detecting cognitive
exaggeration. INTRODUCTION Within the context
of personal injury litigation, one of the
greatest challenges for a neuropsychologist is
the evaluation and detection of biased
responding.   One common approach for evaluating
cognitive effort is the use of tests designed
specifically to assess biased responding through
the representation of a relatively easy task as
being more complex (e.g., TOMM).   However, the
analysis of unusual performance patterns on
standard tests of cognitive ability has become an
increasingly popular strategy from both a
clinical and research perspective.   Such
pattern analyses have most frequently focused on
tests of memory and learning (e.g., RAVLT, RCFT),
however, the emergence of such patterns on
non-memory tasks is also increasing (e.g.,
Seashore Rhythm Test Booklet Category Test).
  A number of researchers have proposed that the
Wisconsin Card Sort Test (WCST) may be useful in
this role as a means for detecting cognitive
exaggeration. Bernard et al. (1996) and Suhr
Boyer (1999) developed malingering formulae,
using standard measures from the WCST, that were
considered useful for differentiating genuine
responding from poor cognitive effort (see Figure
1). More recently, Greve et al. (2002) proposed
an innovative method based on the identification
of 'unique' items on this test (see Figure 1).
The purpose of this investigation was to
examine the false positive error rate of the
three WCST malingering formulae/methods in a
mixed clinical and head injury litigant sample.
RESULTS (cont) Objective 2 False Positive
Error Rate in a Head Injury Litigant Sample The
false positive error rate of the three
malingering formulae/methods in the head injury
litigant sample are presented in Table 2.
Using a criterion base rate of 10, only the
Suhr/Boyer formula demonstrated an acceptable
false positive error rate (i.e., 9.5).
However, acceptable false positive error rates
were demonstrated when any two formulae/methods
were combined, with the lowest rates demonstrated
when combining the Greve et al. method with
either two malingering formulae.
Table 2. False Positive Error Rates of WCST
Malingering Formula in a Medicolegal
Population n
Suhr Boyer (1999) 19 9.5 Bernard et al.
(1996) 16 15.2 Greve et al.
(2002) 23 10.5 Suhr OR Bernard 28 20.0
Suhr OR Greve 36 20.0 Bernard OR
Greve 32 21.9 Bernard OR Suhr OR Greve
41 4.8 Suhr AND Bernard 7 4.8 Suhr
AND Greve 6 0 Bernard AND Greve
7 3.8 Bernard AND Suhr AND Greve 3 0
DISCUSSION CONCLUSION In the mixed clinical
sample, the results of this investigation are
consistent with past base rates studies that have
reported high false positive error rates in
clinical populations using the Suhr/Boyer and
Bernard et al. malingering formulae (e.g., Greve
Bianchini, 2002) In order for the WCST
malingering formulae to be recognized as useful
tools for detecting exaggeration, (a) low false
positive error rates must be established in
Genuine Responders, and (b) high true positive
base rates established in Suspected
Exaggerators. In contrast to this notion, (a)
the false positive error rate of the Suhr/Boyer
and Bernard et al. malingering formulae in this
investigation were unacceptably high in the vast
majority of clinical groups, and (b) the rate of
true positive scores in suspected exaggerator
group was unacceptably low. When both formulae
are combined, false positive error rates did fall
within acceptable limits for almost all clinical
groups. However, the percentage of true positives
in the suspected exaggerator group also declined
accordingly. In the head injury litigant sample,
overall, the false positive error rate of the
three malingering formulae/methods was
high. Contrary to expectations, Greve et al's
(2002) proposed method of identifying "unique"
items on the WCST did not produce lower false
positive error rates than either the Suhr/Boyer
or Bernard formulae. While a combination of any
two formulae/methods was again successful in
decreasing the false positive error rate to
acceptable limits, the contribution of Greve et
al's method in combination with either the
Suhr/Boyer and Bernard et al formula separately
produced the lowest false positive error rates.
These findings indicate that Greve et al's
method of identifying "unique" items on the WCST,
in combination with existing malingering
formulae, was useful for decreasing false
positive error rates in genuine responders.
However, in order to establish the potential
contribution of Greve et al's proposed method as
a means for detecting cognitive exaggeration,
further base rate data and clinical outcomes
analyses are required using suspected
exaggerators in a clinical setting.
Table 1. False Positive Error rate of the
Suhr/Boyer Bernard et al (1996) WCST
Malingering Formulae in a Clinical Sample
Suhr Bernard Suhr/Boyer Boyer et
al AND (1999) (1996) Bernard Cog
Dis. NOS 11.8 8.8 5.9 CVDs
11.0 15.9 6.1 Alazhiemers Dem
7.4 22.2 3.7 Dementia NOS
16.3 25.6 11.6 Drug Abuse -- --
-- Encephalitis 12.5 -- -- ETOH
Abuse 6.3 -- -- MI Dementia
13.0 13.2 3.8 Multiple Sclerosis --
-- -- Parkinsons 15.0 20.0 5.0
Seizure Disorder 9.5 10.9 6.1 TBI
mixed 5.0 20.0 3.3 Tumour 9.4
6.3 3.1 Suspected Exagg 16.7 8.3
5.6 TOTAL 10.3 12.9 5.6
Figure 1 WCST Malingering Formula/Methods
Bernard, McGrath, Houstan (1996) Formula
Derived by Discriminant Functions Analysis
Scores Included Perseverative Errors
Categories Suhr Boyer (1999) Formula Derived
by Logistic Regression Scores Included
Categories Failure to Maintain Set
Greve, Bianchini, Mathias, Houston, Crouch
(2002) Concept of Method Identification of
Unique Responses and Perfect Matches- Missed
(PM-M) Unique Responses Failure to
correctly sort a response card that matches
perfectly with one of the key cards. Perfect
Matches-Missed Eight perfect match cards are
possible in the WCST 128 card version (i.e.,
16, 29, 41, 62 in each deck). Perfect Matches-
Missed is the number of cards incorrectly sorted
on these items.