Malingering response styles on the memory assessment scales and symptom validity tests

Pergamon

Archives of Clinical Neuropsychology, Vol. 10, No. 1, pp. 57-72, 1995 Copyright © 1994 National Academy of Neuropsychology Printed in the USA. All fights reserved 0887-6177/95 $9.50 + .00

0887-6177(94)E0005-A

Malingering Response Styles on the Memory AssessmentScales and Symptom Validity Tests John T. Beetar and I. Michael Williams Hahnemann University, Philadelphia, PA

This study identified malingering strategies of test performance and investigated their presence in the responses to computer-mediated versions o f Rey's DotCounting and 15-1terns tests, a forced-choice symptom validity procedure and the Memory Assessment Scales (MAS). Sixty volunteer subjects were randomly assigned to control (n = 30) or malingering (n = 30) groups. The control subjects were instructed to perform their best and the malingerers were instructed to fake a poor performance on the tests. As expected, malingering subjects scored significantly worse than control subjects on virtually all tests. Malingerers had slower response times on most tests. They also performed worse on recognition tasks in contrast to performance on recall tasks. Their response style was characterized by intentional wrong and random responding on recognition tasks. Malingerers did not show the expected worse-than-chance responding on the forced-choice symptom validity procedure. Current tests of symptom validity may not have sufficient sensitivity to detect milder forms of malingering.

The study of malingering on intellectual tests is part of the general investigation of extraneous influences on test performance. These influences include anxiety and psychological depression as well as malingering (Williams, Little, Scates, & Blockman, 1987). They compromise the validity of assessment because they introduce specific variance in test scores that are unrelated to true score variance (Nunnally, 1978). Although a test score may be designated as a measure of Address correspondence to: J. Michael Williams, Hahnemann University, Philadelphia, PA 19102. 57

58

J. T. Beetar and J. M. Williams

a specific construct such as memory, extraneous factors may systematically influence the test score and confound the interpretation of the score as a measure of this construct. These factors may be so influential that the memory test now becomes a test of depression or anxiety rather than a measure of the construct it was designed to measure. In the domain of intellectual ability measurement, malingering is a general term used to describe a test performance that has been influenced by the extraneous factor of factitious motivation. In this situation, the subject intentionally produces a poor performance that is not characteristic of "true" ability. When factitious motivation is present, a test of an intellectual ability becomes, to some degree, a measure of malingering response style. Neuropsychological assessment is vulnerable to malingering since the validity of testing depends upon the patient's cooperation and motivation to produce the best possible performance. If a poor performance will be used in litigation to justify personal injury and financial claims (Heaton, Smith, Lehman, & Vogt, 1978), or reduce one's responsibility in the case of criminal behavior (Rubinsky & Brandt, 1986), this cooperation may not always be present. Besides direct rewards, there are various indirect social and emotional benefits of presenting oneself as brain injured and these will likewise produce motivation to perform consistent with the brain-injured patient role (Heaton et al., 1978; Lezak, 1983). Some research in the area of malingering and neuropsychological assessment focused on markers of a malingering performance on conventional neuropsychological tests. Heaton et al. (1978) compared 16 subjects instructed to malinger to 16 head-trauma patients using the Wechsler Adult Intelligence Scale (WAIS) and the Halstead-Reitan neuropsychology battery. The malingering subjects did worse on the Speech-Sounds Perception Test, finger tapping, finger agnosia, sensory suppressions, hand dynamometer, and the Digit Span. Head trauma patients did worse on the Category Test, Trails-B, and the "Factual Performance Test. In addition, 10 neuropsychologists made blind judgments of the test protocols. Their ability to correctly classify the subjects as either malingering or impaired was only slightly better than chance. This finding suggested that clinicians cannot detect markers of malingering in the conventional summary of performance on these tests. Goebel (1983) expanded the investigation of Heaton et al. (1978) by using a control group, four malingering groups, and an increased sample size (n = 254). The malingering groups consisted of subjects instructed to produce a performance consistent with their understanding of injury to the left hemisphere, right hemisphere, both hemispheres, and general brain injury. Goebel subjectively examined the data and was able to correctly classify 98% of the unimpaired subjects and 80% of the impaired subjects. Although identifying information was removed from the testing protocol, Goebel had prior knowledge of all the impaired subjects and the base rate of malingering subjects in his sample. These aspects of the study probably contributed to the high classification rate. Faust, Hart, and Guilmette (1988) investigated the ability of 42 neuropsychologists to blindly interpret the malingered performance of three children on

Malingering and the Memory Assessment Scales

59

the Halstead-Reitan test battery. None of the clinicians detected faking although abnormality was diagnosed in 93% of the judgments. Using adolescents instead of children, Faust, Hart, Guilmette, and Arkes (1988) also found that neuropsychologists were unable to detect malingering. Bernard (1990) investigated malingering on the Wechsler Memory ScaleRevised, the Complex Figure Test, the Auditory Verbal Learning Test, and the Rey 15-Items Test. Subjects were randomly assigned to a control group, a malingering group with financial incentive, and a malingering group without financial incentive. Bernard did not find significant differences between the two malingering groups. These two groups, however, performed significantly worse than the control group on 16 of 22 tests. Bernard also performed a discriminant function analysis and was able to correctly classify approximately 75% of the malingerers and controls. This analysis found that malingerers performed worse on recognition versus recall tasks. Another general approach to the study of malingering involves the development of specific "symptom validity" tests. One technique to evaluate short-term recall involves a two-item, forced-choice procedure administered over many trials to determine response probabilities. Binder and Pankratz (1987) reported a case study in which a black pen or yellow pencil was placed before a patient and then removed. After counting aloud to 20, the patient was then asked to report which object was previously presented..Malingering was indicated by the patient's worse-than-chance performance over 100 trials. Hiscock and Hiscock (1989) also used a forced-choice technique in an attempt to detect malingering in a subject complaining of memory impairment. Their forced-choice test consisted of eight 5-digit numbers presented individually on stimulus cards. Response cards consisted of two 5-digit numbers, the target, and the distracter. Testing consisted of 72 trials divided into three blocks of 5 s, 10 s, and 15 s response delays. This subject also performed worse than chance and longer response delays were associated with the greatest number of errors. Although the single cases reported in these studies performed worse than chance, recent studies of these techniques among groups of patients in litigation and subjects instructed to malinger suggest that most malingering subjects do not perform worse than chance (Binder, 1993; Guilmette, Hart, & Giuliano, 1993; Prigatano & Amin, 1993). A second symptom validity procedure, the 15-Item Memory Test (Rey, 1964), is a simple memory test that is introduced to the subject as being difficult. The examiner presents a single figure for 10 s, containing the following 15 items: A, B, C, 1, 2, 3, a, b, c, a circle, a square, a triangle, I, II, and III. Lezak (1983) suggested that even moderately impaired subjects should be able to recall at least three rows or nine items following a 10-second presentation. Schretlen, Brandt, Krafft, and Van Gorp (1991) administered the 15-Item Memory Test to nine subject groups to study its effectiveness in detecting malingering. Although malingering subjects performed worse than controls, brain-injured subjects also did poorly on the task. Since brain-injured subjects were also impaired on the test, the investigators concluded that the 15-items

60

J. Z Beetar and J. M. Williams

test cannot serve as a diagnostic instrument for the detection of malingering. Lee, Loring, and Martin (1992) used the 15-Item Test to gather normative data on patients with genuine memory impairments. They suggested a cutoff score of seven items recalled based on a study comparing outpatients with neurological disorders in litigation to similar outpatients not in litigation. Another symptom validity procedure is the Dot-Counting Test developed by Rey (1941) and reported in Lezak (1983). This task consists of random displays of grouped and ungrouped dots. The subject is requested to count the dots and the examiner records the number counted as well as the counting time. Rey (1941) hypothesized that cooperative subjects will decrease the counting time over successive trials and report more accurate counts. Malingering subjects will produce the opposite pattern. These hypotheses were never systematically investigated but suggest another possible mechanism for the manifestation of malingering response style. These investigations of malingering responses on conventional tests and the study of symptom validity instruments suggest some consistent findings regarding malingering response style. In general, malingerers perform worse on timed tests (e.g., finger tapping, TPT, and specific WAIS Performance tests), tests that have a recognition response format (e.g., Speech Sounds and recognition memory tests including forced-choice techniques) and other tests sensitive to degree of effort (e.g., dynamometer). These findings indicate that malingering may be accomplished by cbaracteristic patterns of responding and some tests are more sensitive to these patterns than others. The following is a presentation of possible malingering strategies and the manner by which malingering may appear in the results of conventional clinical instruments and tests specifically designed to assess symptom validity. STRATEGIES OF M A L I N G E R I N G Random Responding

Tests employing a recognition format over many trials are very susceptible to random responses. Such tests include the Speech-Sounds Perception Test, that Heaton et al. (1978) found significantly discriminated malingerers from controls. Random responding will result in depressed scores. However, true random responding for all trials of a test will not result in performance that is worse than chance. Some intentional wrong responding must occur in order for the subject to respond worse than chance levels. Intentional Wrong Responses

Subjects who employ this strategy know the correct answer, but intentionally respond incorrectly. In addition to fabricated incorrect answers, subjects

Malingering and the Memory Assessment Scales

61

may also withhold answers in an attempt to intentionally produce a poor performance. Thirty percent of the experimental group in Goebel's (1983) study reported giving wrong responses as part of their faking strategy. Further, one can presume that the subjects in the case studies reported by Binder and Pankratz (1987) and Hiscock and Hiscock (1989) were giving intentional wrong responses. Intentional wrong responding combined with random responding may produce worse-than-chance performance over many trials on forced choice symptom validity tests and any other tests that have a recognition format (e.g., Wiggins & Brandt, 1988). Delayed Responding Constructing simple response delays will adversely affect performance on a variety of tasks. On memory tests, subjects who delay unrehearsed responses are more likely to forget what they were asked to remember (Squire, 1986). There are also scoring penalties for a slow performance on most timed tests (e.g., the Block Design and Object Assembly subtests of the Wechsler Adult Intelligence Scale). These time penalties are incurred if the subject goes over the time limit or performs so slowly that the subjects gets no credits for a quick response. Post-study interviews of malingering subjects indicated that 36% reported that they were slow to perform or acted as if they were slow to comprehend (Goebel, 1983). They also reported using poor motor coordination and slowness. Resnick (1984) also described response latency as a factor in detecting a malingered performance. Inattentiveness An individual who engages in this strategy does not attend to test stimuli. This behavior enables the malingerer to misperceive test directions so that incorrect responding may be less effortful. Specific tasks that require sustained attention and effort will be more sensitive to this strategy of malingering. Inattention may also be present with delayed responding. The purpose of this investigation was to examine the presence of malingering strategies in the responses to a conventional battery of memory tests, the Memory Assessment Scales (MAS), and three specific symptom validity tests. These were the Rey 15-items Test, Rey Dot-Counting Test, and a forced-choice recognition and recall test based upon the procedures used by H i s c o c k and H i s c o c k (1989). A m e m o r y battery was chosen because complaints of memory impairment are the most common among patients suspected of malingering and symptom validity tests most often employ a memory assessment strategy (Binder & Pankratz, 1987; Hiscock & Hiscock, 1989).

62

J. Z Beetar and J. M. Williams

METHOD

Subjects Student volunteers (n = 60) from H a h n e m a n n University, Philadelphia, P e n n s y l v a n i a w e r e r e c r u i t e d a n d p a i d $5 e a c h f o r t h e i r p a r t i c i p a t i o n . S t u d e n t s w h o w e r e w o r k i n g in the m e d i c a l l i b r a r y c o m p u t e r l a b o r a t o r y were a p - p r o a c h e d and asked to volunteer for the study. These subjects were r a n d o m l y a s s i g n e d to c o n t r o l a n d m a l i n g e r i n g c o n d i t i o n s . T h e c o n t r o l group c o m p r i s e d 17 males and 13 females. These subjects a v e r a g e d 16.60 years o f education (SD = 2.22) and had a mean age o f 23.97 (SD = 4.54). The e x p e r i m e n t a l group consisted o f 13 males and 17 females. T h e y averaged 17.17 years o f education (SD = 1.53) and had a mean age o f 25.63 (SD = 7.11).

General Procedure Subjects were tested individually in a library area that was isolated from other computer users and reasonably quiet. All auditory test items were presented using earphones. The examiner gave each subject a sealed envelope before testing began and left the room. The envelopes were used to randomly a s s i g n s u b j e c t s to e i t h e r c o n t r o l or m a l i n g e r i n g g r o u p . T h e y c o n t a i n e d informed consent releases and instructions to either try hard to do well or try to fake a poor performance. The examiner was therefore blind to group membership. The instructions for the control group were as follows:

The test you are about to take is a memory test. Your memory will be tested in a variety of ways and all that is asked is that you try to do your best. Please do not communicate these instructions to the examiner.

The instructions to the malingering group were adapted from Faust et al., (1988). They were as follows: When you do most things, you usually try your best. That's probably especially true when you take a test. Now you are being asked to do something very different. You are going to be asked to remember a variety of things, but you are being asked not to do your best. This is a study to see if you can fake a poor performance. This may sound easy, but the hard part will be to convince the person giving the test that you are not faking a bad performance. In other words, it can't be obvious that you're not performing at your best. This is going to take some skill on your part. Try to remember throughout testing that you're trying to fake a convincingly poor performance. Please do not communicate these instructions to your examiner.

Malingering and the Memory Assessment Scales

63

After the subject read the direction's and signed the consent form, the examiner was summoned to begin testing. The tests were administered in the order described below. The entire protocol required approximately 1 hour and 30 minutes.

Tests All tests used in this study were mediated on the Apple Macintosh personal computer. The subtests that required sound were presented using earphones. These stimuli were natural human voice digitally sampled at 22 kHz. Computer mediation of the tests allowed for the measurement of subtle error patterns and response time latencies.

Memory Assessment Scales (MAS; Williams, 1991). The MAS is an individually-administered battery of tests developed to assess memory functions in normal and clinical populations. It consists of 12 subtests that assess three general areas of cognitive functioning: (a) attention, concentration, and short-term memory; (b) learning and immediate memory; and (c) memory following a delay. The following is a summary of the MAS subtests in the order of administration. List learning. This task required the subject to learn a list of 12 common words, three from each of four categories: colors, birds, cities, and countries. The list was presented for a maximum of six trials or until the subject reported all 12 words. For each trial an array of the letters of the alphabet appeared on the computer screen. The screen was visible only to the examiner. As the subject recalled each word, the examiner selected the first letter of the word from among the full alphabet displayed on the screen. This general list-learning procedure was derived from a study of memory disorder by Rubin and Butters (1981). Total number of words recalled constituted the List Acquisition score. The computer recorded the list items reported by the subject and the response latency associated with the recall of each list element. Prose memory. The short story and cued recall questions of the MAS were taken from a study by Rawling and Lyle (1978). The subject was requested to listen to and recall a short story about a bank robbery. Memory for story details was tested using nine recall questions (e.g., What time did the robbery occur?). The number of correct responses to these questions constituted the Immediate Prose Recall Score. Delayed list recall. The subject was next asked to recall the 12-item list again after a short delay. The subject was then asked to recall the words within semantic categories, as prompted by the examiner. These trials rendered free recall and cued recall scores. Response latencies were also recorded for these trials of the list.

64

J. T. Beetar and J. M. Williams

Verbal Span. This is a short-term auditory memory task using forward and backward number spans (Miller, 1956; Wikelgren, 1978). The series range in length from two to nine numbers with two attempts permitted at each series length. For each trial the numerals 0 to 9 appeared on the computer screen visible only to the examiner. As the subject recalled each number, the examiner clicked the same number on the screen. The longest sequence of digits that the subject could recall forward and backward constituted the Verbal Span score. Response time was the latency associated with the recall of each digit. Visual Span. This nonverbal analog of the Verbal Span subtest is a variation of the block-tapping test designed by Corsi (described in Milner, 1971). An array of randomly distributed stars was presented to the subject on the computer screen. A series of stars flashed in a specified sequence and the subject then pointed to the same stars in the correct sequence. The examiner clicked the stars indicated by the subject. The number of stars in each sequence was increased over trials. The longest sequence successfully recalled constituted the Visual Span score. Response time was the latency associated with the recall of each star location. Visual Recognition. This is a task which measures recognition memory for geometric designs. Each design appeared on the screen for 5 s followed by a distraction task. The distraction task re/tuired the subject to search and count the occurrence of a target design among a large array of similar figures. The designs were displayed in a grid which was exposed to the subject for 15 s. The subject was then asked to recognize the design that was presented before the distraction task. Five trials required a same-or-different recognition response and five trials required recognition of the design from a selection of five similar designs. The basic presentation format is a modification of the Brown-Peterson distraction technique (Brown, 1958; Peterson & Peterson, 1959). Scores for all 10 trials were combined to provide the Immediate Visual Recognition score. Visual Reproduction. The subject viewed a geometric figure for 10 seconds, completed the same distraction task described above, and was then asked to draw the original figure. Two trials were administered in this manner. Drawings were scored for the presence or absence of specific details. Scores for the two drawings were totaled to produce the Visual Reproduction score. Names-Faces learning. This task required the subject to learn the names of 10 individuals and associate them to individuals portrayed in photographs. As each face was presented on the computer screen, the subject heard the person's name. Following this presentation, the subject was presented with individual faces and asked to recognize the correct name from a list of three, one of which was the correct name. Scores for two learning trials were combined to produce the Immediate Names-Faces score.

Malingering and the Memory Assessment Scales

65

Delayed List Recall. The subject was required to recall words presented in the List Learning subtest. The subject was then asked to recall words within categories as prompted by the examiner. The number of words correctly recalled constituted the Delayed List Recall and Delayed Cued List Recall score.

Delayed Prose memory. The subject was asked to produce a free recall of the story. Following this, the subject was asked nine questions concerning the details of the story. The number of correct responses to the nine questions constituted the Delayed Prose Recall score.

Delayed Visual Recognition. Twenty geometric designs, 10 of which were designs presented in the Visual Recognition subtest, were presented on the computer screen. The subject was requested to indicate the figures previously presented. The number of designs correctly recognized constituted the Delayed Visual Recognition score.

Delayed Names-Faces Recall. The subject was asked to recognize the correct names of individuals portrayed in photographs, as presented in the Names-Faces subtest. The total number of correctly identified names was the Delayed Names-Faces score. MAS standard scores were derived from data collected from 843 adult volunteers. These subjects ranged in age from 18 to 90 years. Standard scores for 12 individual subtests were constructed to have a mean of 10 and a standard deviation of three. Scoring of the MAS also includes four Summary Scales derived from factor-analytic studies: Short-Term Memory, Verbal Memory, Visual Memory, and Global Memory. These are designed to have a mean of 100 and standard deviation of 15. The Short-term Memory score was based on scores for the Verbal Span and Visual Span subtests. The Verbal Memory Summary Scale score was based on the List Recall and Immediate Prose Recall subtest scores. The Visual Memory Summary Scale score, which provides a measure of visual-spatial memory abilities, was derived from the Visual Reproduction and Immediate Visual Recognition subtest scores. The Global Memory Scale score is a measure of general memory ability. It was derived from the Verbal and Visual Memory Summary Scale scores.

Symptom Validity Tests Memorization of 15 items test (Rey, 1964). This task required the memorization of 15 different items arranged in five rows of three characters. They were displayed on the computer screen for l0 s. The subject was instructed to remember the items for immediate recall. The score was the total number of items correctly recalled.

Dot counting: Grouped and ungrouped. Based on a procedure reported by Rey (1941), this test consisted of a random presentation on the computer of l0 tri-

66

J. T. Beetar and J. M. Williams

als of ungrouped dots and 9 trials of grouped dots. The number of ungrouped dots ranged from 10 to 28 and the number of grouped dots ranged from 12 to 42. The subject was instructed to count the dots and then report the total number displayed for each trial. The score was the number of trials for which an incorrect total was counted. The computer recorded the time required to count each set of dots. Letters and numbers. This task was derived from a forced-choice method used

by Binder and Pankratz (1987) and Hiscock and Hiscock (1989) and requires either recall or recognition of a string of l0 letters and numbers. A randomly chosen letter began each string followed by letters and numbers generated in random order. There were 100 recall/recognition trials presented randomly for 3 s each. The assignment of recall or recognition trial was randomized within the total set of 100 trials. Individual trials had a delay of l0 s or 20 s before the subject's response was elicited. The delay incorporated a visible countdown on the computer screen. The recognition component required the subject to choose between the correct response and a distracter. The recall component required the subject to recall the string of letters and numbers. Errors were computed as a fraction of the total number of items for recall and recognition sets. Response time for the recognition task was the time interval between the presentation of the two recognition choices and the identification of either target or distracter. Response time for the recall task was the time required to provide items remembered of each string.

RESULTS A multivariate analysis of variance (MANOVA) indicated that control and malingering subjects significantly differed on measures of test performance, F(38, 76) = 574.800, p < .001, and response times, F(16, 98) = 84.701, p < .001. Univariate analyses using independent t-tests were used to determine which tests significantly discriminated the two groups (see Table 1). The only measure for which there was no large, significant difference was the recall component of Letters and Numbers. On this task the mean performances of both malingerers and control subjects were similar. Malingerers scored worse than control subjects on the recognition component of this test. All of the control subjects had MAS scores well within the range of their age and education normative standard. The malingerers scored in the low average range. Since the sample was drawn from the same pool of volunteers and subjects were randomly assigned to study conditions, this level of performance for the malingerers is probably much lower than their true memory ability. The pattern of performance of malingerers was remarkable for two major features that are inconsistent with organic memory disorder. First, measures of immediate recall, such as Digits Forward, were within the same range as the

Malingering and the Memory Assessment Scales

67

TABLE 1 Test Results of Subject Groups Control

SD

M

SD

t(58)

10.13 8.86 11.27 11.13 10.80 10.50 11.67 10.00 10.60 8.63 10.53 9.43

1.50 2.57 2.00 3.27 3.40 2.50 1.90 2.45 1.83 2.53 2.54 2.13

6.16 6.37 6.53 7.23 7.93 5.67 9.47 4.70 4.83 6.03 5.60 3.77

2.53 2.03 2.62 3.00 3.68 3.09 2.27 2.78 2.65 2.12 4.01 3.05

7.38"* * 4.19"** 7.86*** 4.82*** 3.14"* 6.67*** 4.07*** 7.83*** 9.80*** 4.33*** 5.69"* * 8.35 * **

110.00 103.67 109.23 107.57 14.70 1.97

14.69 10.18 10.92 10.72 .92 1.54

88.83 83.53 88.47 83.93 13.07 5.50

18.28 9.42 13.42 11.03 2.69 3.12

4.94*** 7.95*** 6.57*** 8.42*** 3.13"* -5.52***

.21 .00

.15 .01

.21 .17

.15 .16

-0.00 -5.63***

Response Time b,c List Acquisition Trial 1 Trial 2

2.56 2.00

1.31 1.07

4.82 4.19

2.30 2.18

-4.67*** -4.95***

Verbal Span Digits Forward Digits Backward

.83 1.07

.20 .36

1.33 1.51

.39 .54

-6.25*** -3.68***

1.47 5.93

.31 .98

1.80 6.92

.53 1.53

-2.92** -2.94*

11.31 4.20

2.77 .33

13.13 4.59

3.23 32.25

Test Performancea Memory Assessment Scales (MAS) Subtests List Acquisition Immediate Prose Recall List Recall Verbal Span Visual Span Immediate Visual Recognition Visual Reproduction Immediate Names-Faces Delayed List Recall Delayed Prose Recall Delayed Visual Recognition Delayed Names-Faces Summary Scores Short-Term Memory Verbal Memory Visual Memory Global Memory 15-Items Test Dot Counting Letters and Numbers Recall Errors Recognition Errors

Visual Span Dot Counting Letters and Numbers Recall Recognition

M

Malingerers

-2.31" -3.43***

Note. MAS subtest scores are expressed as standard scores with a mean of 10 and standard deviation of 3. MAS summary scores have a mean of 100 and standard deviation of 15. Scores on the 15-Item Test represented the total recalled. Dot Counting performance scores represent the number of trials for which there was an incorrect dot count. Dot Counting time represents seconds per trial. Recall and Recognition errors on the Letters and Numbers task are expressed as the proportion of errors for 50 trials. aperformance Multivariate F(38, 76) = 574.800, p < .001. bRespo n se Time Multivariate F(16, 98) = 84.701, p < .001. CAll times are expressed in seconds. *p < .05.; **p < .01 .; ***p < .001.

68

J. T. Beetar a n d J. q14. Williams

consolidation measures. Standard scores ranged from 83.53 to 88.83. They also performed worse on visual recogmtlon ' " " measures in contrast to visual recall (see Figure 1).

DISCUSSION These findings clearly indicated that volunteer subjects were able to effectively fake a bad performance on the Memory Assessment Scales without performing so poorly that their malingering was obvious. In general, the subjects instructed to malinger performed in the low average range on subtests of the MAS. They performed best on measures of free recall, such as the list acquisition and visual recall, and worse on recognition measures and delay components of all tasks. This pattern of poor recognition and better recall stands in contrast to the pattern associated with amnesic disorders and normal memory performance. Amnesic subjects and normals perform best on recognition tasks and worse on recall tasks (Hirst, Johnson, Kim, Phelps, & Volpe, 1986). This result may be explained by reference to one compelling feature of malingering: Volunteer subjects usually have no specific understanding of amnesic disorder or the cognitive psychology of memory. Therefore, they do not have sufficient knowledge to mimic true memory disorder. In particular, malingering subjects do not know what constitutes a poor recall performance and this may have resulted in a discrepancy between recall and recognition. For example,

12

N

FIGURE 1. Study group comparisons on the subscales of the Memory Assessment Scales (MAS).

Malingering and the Memory Assessment Scales

69

on the 12-item list-learning of the MAS, subjects probably do not know that a mean of eight items recalled for the six learning trials is still within the average range for unimpaired subjects. Likewise they do not know how poorly reproduced a figure must be to qualify as impaired. The results of this study suggest that malingering subjects still produce a recall performance within the average range even when they believe they are producing a poor performance. In contrast to recall performance, malingering subjects have a clear idea of poor performance on recognition tasks, such as the multiple choice trials of the visual memory section of the MAS, or the Letters and Numbers symptom validity test. Since all of the possible responses are available, the malingering subject who presumably knows the correct answer can easily choose another incorrect, distracter item from among the multiple choices. By thus choosing incorrectly or randomly, the malingering subject can easily produce a very poor performance on recognition tasks. This feature of malingering may influence performance on any test that utilizes a multiple-choice, recognition format. Such tests include the Speech Sounds Perception and Rhythm tests of the Halstead-Reitan Neuropsychological Test Battery, symptom validity tests, recognition components of memory tests, and a variety of other conventional clinical assessment procedures. The finding that malingerers delayed their responses and had longer response times supports previous reports describing this faking strategy (Goebel, 1983; Resnick, 1984). The usb of computer-mediated versions of the MAS and the s y m p t o m validity tests permitted accurate measurement. However, there are no normative data available for response times and it is important to note that a delayed response is not always evidence of malingering. Impaired subjects may also respond slowly and they were not examined in the present study. Other extraneous factors, such as test-related anxiety and depression, may also affect response time. An important finding of this investigation was that malingerers scored significantly worse than control subjects on all tests except the recall component of Letters and Numbers. On the recognition component of this test a response style characterized by intentional wrong answers was evident. This suggests a conscious effort to respond incorrectly or randomly when given the opportunity to choose between a target response and a distracter. Consistent with other studies (Guilmette et al., 1993; Prigatano & Amin, 1993; Trueblood & Schmidt, 1993), no subject in this investigation performed worse than chance on the forced-choice recognition task. Given the educational level of the current sample, this may be explained by their awareness that some tasks may be too easy for performance to be worse than chance. The finding that the experimental and control groups performed alike on the recall component of Letters and Numbers again suggests that it is easier to choose rather than construct an incorrect answer. Malingering subjects may have adopted intentional-wrong responding on the Dot Counting Test. They gave incorrect dot counts on approximately one-

70

J. T. Beetar and J. M. Williams

third of the counting trials. However, their counts were only incorrect by a small amount. They also counted more slowly than the control subjects. As with most other tasks, this represented the strong tendency of malingering subjects to perform slowly on every task presented to them. The results obtained on the 15-Item Memory Test indicate that malingerers withheld more responses than control subjects. The malingerers, however, performed above suggested cutoff scores for this particular test (Lezak, 1983; Schretlen et al., 1991). This finding supports the conclusion made by Schretlen et al. (1991) that the 15-Item Memory Test is so easy that it does not discriminate malingerers from controls or brain-injured subjects. These malingering subjects probably engaged in all four of the strategies described earlier. Delayed responding was clearly evident on all tasks for which response times were recorded. Malingerers were consistently slower on every task, including individual trials within tasks. This pattern even extended to response times for individual digits recalled on the Digit Span task. Random and intentional-wrong responding were likely the strategies that resulted in poor scores on most tests. However, it is currently impossible to determine whether the malingering subjects adopted an exclusive intentionalwrong or random response strategy. The unique demographic characteristics of these subjects cannot be overstated as possible extraneous influences on the generalizibility of findings. The subjects in this study were well-educated and were not motivated to malinger in the same way as malingering patients in a clinical setting. However, the major purpose of the study was to investigate strategies of malingering, and bright, highly educated subjects may utilize the full range of malingering strategies. They may even invent a number of new approaches to malingering that subjects with lower education or intelligence might not use. Although these subjects can accomplish the malingering task, they have a demographic background that is probably different than the background of clients or patients referred for clinical treatment. These differences might be reflected in different motivations and approaches to malingering. The drop in scores associated with malingering may also vary with overall ability levels. Malingering subjects whose true ability is below average may not be able to lower their scores as far as subjects whose ability is above average or superior. In addition, the contextual features of this simulation study are different than the situation of "true" malingering. However, it is difficult to perform group studies of genuine malingerers. Genuine malingerers would not be malingering if they admitted their dishonest performance and volunteered for a research study on malingering. Although these simulation studies do not absolutely replicate the everyday clinical situation, the conditions required of the subject are very similar to those required of a genuine malingerer. A malingerer is presumably an unimpaired person who attempts to fake a poor performance on tests. Presumably subjects instructed to malinger who take the task seriously will invent and apply similar strategies attempted by the "true" malingerer. The

Malingering and the Memory Assessment Scales

71

evidence from numerous studies (e.g., Heaton et al., 1978) suggests that these subjects actually do perform similar to clinical malingerers and to brain-injured subjects. The simulation model allows investigators to identify numerous strategies that an unimpaired subject might employ to fake a bad performance on particular tests. However, the findings of the present study rest on the inference that the simulation was valid. Of course, the simulation may not be valid and these results may not be generalized to the true malingering situation. It would appear that subjects employed a variety of strategies to accomplish a malingered performance. Strategies probably varied across tasks and even across items within tasks. It is apparent that malingering is a skill, and some subjects were more skillful and innovative than others in the invention and application of malingering strategies. Some malingerers may be subtle and intelligent in the application of these techniques and others are more obvious and concrete in their approach. Many subjects will produce a worse performance than expected without performing so poorly that their scores fall below the average range, or to worse-than-chance levels on symptom validity tests. Some of these subjects may be sufficiently intelligent to notice that the task is simple and worse-than-chance levels will be viewed as too extreme by examiners. As a consequence, symptom validity tests that rely on worse-thanchance performance may only detect naive, obvious malingering subjects. This analysis suggests that there may be no specific symptom validity or malingering tests. Instead, there are tests of memory recognition or other constructs that are especially vulnerable to malingering strategies. Many conventional clinical instruments may serve the same role as symptom validity tests if the malingering strategies can be discerned in responses to test items. The present study found only one sign of malingering and some indications that response delay may eventually serve in this capacity. The end result of these investigations may be that there are few pathonomonic signs characteristic of all malingering. Most malingerers may perform so similar to impaired subjects that they cannot be discriminated by any sign or unique profile.

REFERENCES Bernard, L. C. (1990). Prospects for faking believable memory deficits on neuropsychological tests and the use of incentives in simulation research. Journal of Clinical and Experimental Neuropsychology, 5, 715-728. Binder, L. M. (1993). An abbreviated form of the Portland Digit Recognition Test. Clinical Neuropsychologist, 7, 104-107. Binder, L. M., & Pankratz, L. (1987). Neuropsychologicalevidence of a factitious memory complaint. Journal of Clinical and Experimental Neuropsychology, 9, 167-171. Brown, J. (1958). Some tests of the decay theory of immediate memory. Quarterly Journal of Experimental Psychology, 10, 12-21. Faust, D., Hart, K., & Guilmette, T. J. (1988). Pediatric malingering: The capacity of children to fake believable deficits on neuropsychological testing. Journal of Consulting and Clinical Psychology, 56, 578-582.

72

J. T. Beetar and J. M. Williams

Faust, D., Hart, K., Guilmette, T. J., & Arkes, H. (1988). Neuropsychologists' capacity to detect adolescent malingerers. Professional Psychology: Research and Practice, 19, 508-515, Goebel, R. A. (1983). Detection of faking on the Halstead-Reitan neuropsychological test battery. Journal of Clinical Psychology, 39, 731-742. Guilmette, T. J., Hart, K., & Giuliano, A. J. (1993). Malingering detection: The use of a forcedchoice method in identifying organic versus simulated memory impairment. Clinical Neuropsychologist, 7, 59-69. Heaton, R. K., Smith, H. H., Lehman, R. A. W., & Vogt, A. T. (1978). Prospects for faking believable deficits on neuropsychological testing. Journal of Consulting and Clinical Psychology, 46, 892-900. Hirst, W., Johnson, M. K., Kim, J. K., Phelps, E. A., & Volpe, B. T. (1986). Recognition and recall in amnesics. Journal of Experimental Psychology: Learning, Memory, and Cognition, 12, 445-451. Hiscock, M., & Hiscock, C. K. (1989). Refining the forced-choice method for the detection of malingering. Journal of Clinical and Experimental Neuropsychology, 11,967-974. Lee, G. P., Loring, D. W., & Martin, R. C. (1992). Rey's 15-item visual memory test for the detection of malingering: Normative observations on patients with neurological disorders. Psychological Assessment, 4, 43-46. Lezak, M. D. (1983). Neuropsychological assessment (p. 615). New York: Oxford University Press. Miller, G. (1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63, 81-97. Milner, B (1971). Interhemispheric differences in the localization of psychological processes in man. British Medical Bulletin, 27, 272-277. Nunnally, J. (1978). Psychometric theory (pp. 627-677). New York: McGraw-Hill. Peterson, L. R., & Peterson, M. J. (1959). Short-term retention of individual verbal items. Journal of Experimental Psychology, 58, 193-198. Prigatano, G., & Amin, K. (1993). Digit Memory Test: Unequivocal cerebral dysfunction and suspected malingering. Journal of Clinical Experimental Neuropsychology, 15, 537-546. Rawling, P., & Lyle, J. G. (1978). Cued recall and discrimination of memory deficit. Journal of Consulting and Clinical Psychology, 46, 1227-1229. Resnick, P. J. (1984). The detection of malingered mental illness. Behavioral Sciences and the Law, 2, 21-38. Rey, A. (1941). L'examen psycho logique duns le cas d'encephalopathie traumatique. Archives de Psychologie, 37, 126-139. Rey, A. (1964). L'examen clinique en psychologie. Paris: Presses Universitaires de France. Rubin, D. C., & Butters, N. (1981). Clustering by alcoholic Korsakoff patients. Neuropsychologia, 19, 137-140. Rubinsky, E. W., & Brandt, J. (1986). Amnesia and criminal law: A clinical overview. Behavioral Sciences and the Law, 4, 27-46. Schretlen, D., Brandt, J., Krafft, L., & Van Gorp, W. (1991). Some caveats in using the Rey 15item memory test to detect malingered amnesia. Psychological Assessment, 3, 667-672. Squire, U (1986). Mechanisms of memory. Science, 232, 1612-1619. Trueblood, W., & Schmidt, M. (1993). Malingering and other validity considerations in the neuropsychological evaluation of mild head injury. Journal of Clinical and Experimental Neuropsychology, 15, 578-590. Wiggins, E. C., & Brandt, J. (1988). The detection of simulated amnesia. Law and Human Behavior, 12, 57-78. Wikelgren, W. A. (1973). The long and short of memory. Psychological Bulletin, 80, 425-438. Williams, J. M. (1991). Memory Assessment Scales. Odessa, FL: Psychological Assessment Resources. Williams, J. M., Little, M. M., Scates, S., & Blockman, N. (1987). Memory complaints and abilities among depressed older adults. Journal of Consulting and Clinical Psychology, 55, 595-598.