Assessment of dissimulation in a forensic population

MISSING IMAGE

Material Information

Title:
Assessment of dissimulation in a forensic population
Physical Description:
viii, 117 leaves : ill. ; 29 cm.
Language:
English
Creator:
Kapel, Lawrence
Publication Date:

Subjects

Subjects / Keywords:
Malingering   ( mesh )
MMPI   ( mesh )
Forensic Psychiatry   ( mesh )
Residential Facilities -- Florida   ( mesh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1991.
Bibliography:
Includes bibliographical references (leaves 112-116).
Statement of Responsibility:
by Lawrence Kapel.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 25668322
ocm25668322
System ID:
AA00011174:00001

Full Text












ASSESSMENT OF DISSIMULATION IN A FORENSIC POPULATION


By

LAWRENCE KAPEL
















A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1991
















ACKNOWLEDGEMENTS


The author would like to gratefully acknowledge the

following persons whose contributions were essential to

this project. Tonia Kapel's clerical support was surpassed

only by her emotional support. Illana Interrante spent

countless hours with me over the computer puzzling over

statistics. Dr. Hugh Davis provided guidance throughout

this project and my entire graduate school experience. Dr.

Monte Bein provided guidance throughout this project and

coordinated the administrative chores necessary to conduct

research at the NFETC. Finally, I would like to

acknowledge the staff at the NFETC who contributed at an

astounding response rate on this project.
















TABLE OF CONTENTS


page

ACKNOWLEDGMENTS........... ............................ .. ii

LIST OF TABLES............................ .. .............

LIST OF FIGURES......................................... vi

ABSTRACT................................................ vii

CHAPTERS

1 INTRODUCTION........................................ 1

Population....................................... 5
The Problem and Related Research................. 10
Purpose of Current Study and Specific
Predictions ................................... 35

2 METHODS............................................. 39

Setting and Subjects............................. 39
Materials and Measures........................... 44
Procedure......................................... 47
Analysis........................................ 51

3 RESULTS.......................................... 56

Inter-Professional Agreement......................... 56
Group Differences...... ...................... ... 60
Division into Groups............................. 61
MMPI Analysis .................. .................. 65
Regression Analysis................ ...... ..... .. 88

4 DISCUSSION.......................................... 92

Professional Agreement............................ 92
Baserates of Dissimulation...................... 93
Group Differences................................ 95
MMPI Differences.................................. 96
Prediction Equations............................. 105
MMPI-2: Applications of This Research............ 106
Conclusions........................................ 107


















APPENDIX .............................................. 110

REFERENCES............................................. 112

BIOGRAPHICAL SKETCH.................................... 117















LIST OF TABLES


Table page

2-1: Legal Status of Residents at NFETC.............. 40

2-2: Severity of Crimes per Resident at NFETC......... 42

2-3: Response Rates per Professional Group........... 49

2-4: Classification of Residents..................... 54

3-1: Dissimulation Rating per Professional Group..... 57

3-2: Consciousness Rating per Professional Group..... 58

3-3: NDISS Groups.................................... 62

3-4: VAR by Groups................................... 63

3-5: VAR2 by Groups................................... 64

3-6: MMPI Variables Significantly Different on
the NDISS Variable.............................. 79

3-7: MMPI Variables Significantly Different on
the VAR Variable............................... 81

3-8: MMPI Variables Significantly Different on
the VAR2 Variable............................. 83

3-9: MMPI Variables Significantly Different
by Status..................................... 85

3-10: MMPI Variables Significantly Different
by Status Variable............ .................. 87

3-11: MMPI Variables Significantly Different in
ITP Group Comparing Exaggerators and the
Rest of the Sample............ ................. 89
















LIST OF FIGURES


Figure age

2.1: Dissimulation Rating Form....................... 46

3.1: Standard MMPI Profile of Total Sample........... 66

3.2: Supplemental Scales of MMPI for Total Sample.... 67

3.3: Standard MMPI of NDISS Groups................... 68

3.4: Supplemental Scales on MMPI for NDISS Groups... 69

3.5: Standard MMPI Profile of VAR Groups............. 70

3.6: Supplemental Scales of MMPI for VAR Groups....... 71

3.7: Standard MMPI Profile of VAR2 Groups............ 72

3.8: Supplemental Scales of MMPI for VAR2 Groups..... 73

3.9: Standard MMPI Profile by Status.................. 74

3.10: Supplemental Scales of MMPI for Status........... 75

3.11: Standard MMPI Profile by Felony Code............. 76

3.12: Supplemental Scales of MMPI for Felony
Code Groups.................................... 77
















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

ASSESSMENT OF DISSIMULATION IN A FORENSIC POPULATION

By

Lawrence Kapel

August 1991

Chairman: Hugh Davis, PhD
Major Department: Department of Clinical and Health
Psychology

The incidence of exaggerating and minimizing of

psychopathology in a forensic setting was investigated.

Primary therapists, psychologists and psychiatrists

evaluated 144 consecutive admissions at the North Florida

Evaluation and Treatment Center (NFETC) using two scales

designed to assess the residents' portrayal of

psychopathology and motivation for the portrayal of their

behavior. Inter-rater agreement was established to

differentiate those seen as exaggerating, minimizing and

accurately reflecting their pathology.

Baserate analysis suggested that the rate of

malingering in this forensic facility is lower than would

have been predicted from the literature. Defensive

responding was more common than malingering, but most of

the residents were seen as accurately portraying their


vii











pathology. Severity of crime was not related to

dissimulation. Only 10% of the sample was seen as

exaggerating pathology, and 28.6% of the sample was seen as

minimizing pathology. Most of the population was seen as

accurately reflecting their pathology. Those seen as

exaggerating their pathology were also considered to be

malingering. All of the malingerers were in the

Incompetent to Proceed (ITP) category, although this was

not statistically significant.

Minnesota Multiphasic Personality Inventory (MMPI)

analysis revealed several statistically and clinically

significant differences when comparing the accurate,

exaggerating and minimizing groups. MMPI differences were

found comparing the ITP and Not Guilty by reason of

Insanity (NGI) group. The bulk of the differences were

between the exaggerators and the rest of the sample.

Significant differences were found on MMPI scales

indicative of general pathology and scales to assess

dissimulation. Prediction equations did not, however, aid

in the prediction of malingering.

Malingering in this forensic sample appears to be

manifested by a global endorsement of psychopathology

indicative of a variety of psychiatric disorders. The

general over-reporting of psychopathology might be the best

indicator of malingering. Suggestions for future research

in the area of malingering is discussed.


viii














CHAPTER 1
INTRODUCTION


The revised third edition of the Diagnostic and

Statistical Manual of Mental Disorders (DSM-3-R) defines

malingering to be the intentional production or

exaggeration of physical or psychological symptoms

motivated by an external situation (e.g., incentives) and

not the result of a mental disorder (APA, 1987). As such,

malingering is a conscious attempt at portraying oneself in

a fraudulent manner in order to gain an external advantage.

Malingering is not seen as a mental disorder although

several psychoanalytic theorists have considered

malingering to be a sign representative of a mental disease

more severe than the neurotic disorders (Eissler, 1951).

This formulation is based on clinical observations which

suggest that those who malinger tend to develop mental

difficulties subsequent to the malingering.

Pollack (1982) suggests that there are four dimensions

to the construct of malingering. The first dimension is

the ability to identify a true illness with which to

contrast the malingerer. This is somewhat difficult in the

mental health profession given the debate concerning the

veracity of the existence of mental health disorders

(Szasz, 1987) and in the utility of classification (Rogers,











1954). There is a sizable literature on this debate.

Current practice in the mental health fields recognizes the

necessity of diagnosis and the third edition of the DSM has

helped to alleviate many of the concerns regarding

classification (Blashfield, 1984). The second dimension

concerns the suspected malingerer's desire to assume a

sufficiently instrumental sick role. This refers to

society's reactions to the attempted dissimulation in terms

of benefits that it might accrue for someone who

legitimately suffers from such a disorder. The third

dimension concerns the purposefulness of the dissimulation.

Accordingly, the portrayal must be under control of the

suspected malingerer. Psychologically, the act must be a

conscious portrayal. The fourth dimension concerns the

rationality of the act. In other words, the benefits

accrued to the malingerer should be considered desirable by

persons not mentally impaired. In general, Pollack

suggests that mental health professionals are less likely

to suspect and identify malingering because they tend to

recognize many variables that might be construed as related

to the above dimensions (e.g., need for love) that other

professionals would not. As such, the legal model of

malingering is more inclusive than the model employed by

mental health professionals and stresses the purposive

intent to deceive. Concomitantly, attorneys tend to be the











professionals who most often will raise the specter of

malingering.

Resnick (1988) delineates five reasons why people

might desire to malinger psychosis. These reasons can be

extended to the construct of malingering in general.

Criminals may seek to avoid punishment in some form. This

is often accomplished by attempting to appear Incompetent

to Stand Trial (IST, also known as Incompetent to Proceed

(ITP)) or Not Guilty by Reason of Insanity (NGI). Persons

may seek to avoid military service. Persons may seek to

gain financial benefit through agencies (e.g., social

security). Prisoners may malingerer to obtain better

quarters or drugs. Persons may also malinger in order to

gain admittance into the hospital and get free room and

board. This study is concerned with the first group

mentioned above which involves a forensic population.

The diagnosis of a mental disorder does not preclude

the presence of malingering. For example, a person

diagnosed with schizophrenia might malinger by reporting

the experience of hallucinations in order to receive

financial benefits in the absence of the symptoms. The

DSM-3-R does not provide information concerning the

prevalence of malingering in the general population or in

special populations. Melton (1987) and Grisso (1986) both

suggest that forensic populations are likely to be

over-represented by malingerers due to the potential











benefits of a successful portrayal of insanity or

diminished capacity. Rogers (1986a) found that

approximately 21% of persons being assessed for criminal

insanity were suspected of malingering. There is little

corroborating demographic data concerning this population.

Nevertheless, this data would support the above notion

concerning the over-representation of forensic populations

in relation to malingering.

The construct of defensiveness is not included in the

DSM-3-R. Defensiveness is the polar opposite of

malingering and is the conscious denial or minimization of

physical or psychological symptoms (Rogers, 1988). As with

malingering, this minimization must be seen as rational and

portrayed in order to gain an external advantage. Much of

the above discussion related to the motivation to malinger

applies to defensiveness. In forensic settings,

defensiveness is most likely to be seen in persons who have

already been convicted or found NGI and are looking to be

released or gain privileges (Grisso, 1986). It is

important to distinguish between someone who is protecting

their ego resources through defense mechanisms and one who

is defensive in terms of the malingering-defensiveness

continuum. Defensiveness, as used in this context, is not

a psychological defense mechanism, rather it is an

instrumental behavior designed to gain external benefits.











Population

The constructs of malingering and defensiveness are of

significant importance in forensic settings which provide

strong external incentives for behavioral dissimulation.

Forensic psychiatric facilities (herein referred to as

forensic facilities) vary from state to state in terms of

criteria for admittance and relationship with the state's

corrections department. In the state of Florida the

forensic facilities are under the auspices of the

Department of Health and Rehabilitative Services (HRS).

There are three such facilities in Florida. This study

will involve the North Florida Evaluation and Treatment

Center (NFETC) located in Gainesville, Florida. There are

three general psychiatric units at the NFETC which are

almost entirely comprised of persons who have been

adjudicated either NGI or IST.

The majority of persons at the NFETC are those who

have been found to be Incompetent to Stand Trial. These

persons have yet to stand trial. As such, Grisso (1986)

posits several reasons why this population might be

motivated to malinger a mental disorder. Overtly crazy

behavior during incarceration prior to trial might help in

establishing an insanity defense. The delay while the

defendant is being considered incompetent allows the

defense council greater time to prepare the defense. This

delay may also result in greater difficulty for the











prosecutor obtaining witnesses willing to testify.

Finally, the prosecution might be willing to plea bargain

rather than have to interact with the protracted process of

establishing someone's competency.

The national precedent for the current use of the

incompetency criteria was set in the case of Dusky vs.

United States (1960). This case stressed the current

status of the defendant and his ability to have a rational

and factual understanding of proceedings for which he is

undergoing, as well as his ability to assist in these

proceedings. Competency issues relate to the defendant's

current presentation and has no relation to his mental

status at the time of the offense.

Prior to 1989, the Florida rules of criminal procedure

in 1987 listed the following criteria for the finding of

IST. The defendant should be able to demonstrate the

following: an appreciation of the charges, an

understanding of the range and nature of possible penalties

associated with these charges, an understanding of the

adversarial nature of the legal process, the capacity to

disclose to his attorney pertinent facts surrounding the

alleged offense, the ability to relate to his attorney, the

ability to assist his attorney in planning a defense, the

capacity to realistically challenge prosecution testimony,

the capacity to testify relevantly, the motivation to help

oneself in the legal process and the capacity to cope with











the stress of incarceration while awaiting trial. In 1989,

the Florida rules of criminal procedure concerning

competency were abbreviated to six criteria. The current

criteria are as follows: the defendant demonstrates an

appreciation of the charges and range and nature of

concomitant penalties, has an understanding of the nature

of the legal process, has the capacity to disclose to his

attorney facts pertinent to the proceedings at issue, and

has the ability to manifest appropriate courtroom behavior

and has the capacity to testify relevantly.

The competency criteria do not directly relate to

mental illness. As such, one may be IST and not be

diagnosed with a mental disorder. Equivalently one may

have a major psychiatric disorder and not be considered to

be incompetent to stand trial. Grisso (1988) reported that

between 10% and 25% of those who are found to be legally

competent have major psychotic disorders diagnosed.

Lawrence (1985) cautions that mental health professionals

often report on the defendant in terms of psychiatric

diagnosis and psychological constructs in lieu of the

legally oriented competency questions. My personal

experience suggests that this is less of an issue at the

present time.

Grisso (1988) suggests that there are four major

reasons why defendants appear to have the functional

deficits required to be considered IST. They are mental











disorder, mental retardation, ignorance and malingering.

Grisso suggests that only deficits associated with a mental

disorder will be sufficient for the legal determination of

IST. The final determination of competency is a legal one

and is not under the purview of mental health

professionals. As such, mental health professionals should

provide a descriptive service from which a judge might base

his legal decision. Scheidemandel and Kanno (1969)

conducted a survey of forensic facilities and found that

persons classified as IST outnumbered the NGI population by

a ratio of approximately 13:1. Lawrence (1985) cited

numerous studies supporting the preponderance of IST

persons in forensic settings.

Despite the dominance of IST persons in forensic

settings, the NGI population receives significantly more

attention in the popular media. The attempted

assassination of Ronald Reagan focused national attention

on the insanity defense (Melton, 1987). As with the

competency criteria, the criteria required in order to be

considered legally insane at the time of the offense vary

from state to state. Quen (1974) traced the history of the

insanity defense to biblical days. The current standards

that most states in the United States adhere to are based

on English Common Law. Specifically, the M'Naughten

decision in 1843 provides the guidelines most commonly

applied in the United States (including the state of











Florida). There are two facets to the M'Naughten rule.

The first is that the defendant suffer from a mental

disease or defect. The second is that the defendant had a

defect in reason which resulted in the defendant not

understanding either the nature and quality of the act or

that the act was wrong.

Florida's rules of criminal procedure (1987) state

that when an NGI plea is contested the court must order the

defendant to be evaluated by two or three examiners. The

examiners must submit a report to the court including

techniques employed during the evaluation, a description of

the defendant's mental status at the time of the offense

and the supporting facts or opinions. The ultimate

decision regarding sanity is an issue of fact resolved by

the judge or jury. Unlike the competency decision,

insanity does not relate to the defendant's current

condition and requires that the examiner reconstruct the

defendant's condition at the time of the offense.

Melton (1987) points out that the diagnosis of a

mental disorder or impairment is necessary for the finding

of NGI; however, this is not sufficient. The mental

disorder must in some way be causally linked to the act.

Further, most NGI defendants have committed serious crimes

(often murder) which strongly argues the instrumental value

of having been found NGI for those in forensic populations.

However, 60%-90% of those found NGI are still diagnosed as











psychotic after acquittal. Recidivism of NGI defendants

are roughly the same as matched felons. Howard and Clark

(1985) suggest that the irrationality of the crime might be

the biggest contributor to the success of the NGI defense.

They further stated that approximately 8% of a Michigan

sample evaluated for the NGI defense actually received it.

Melton (1987) suggested that approximately 25% of those

evaluated are successful. However, Melton points out that

the NGI defense is only considered in less than .5% of the

criminal proceedings.

The NGI and IST populations are inter-related in

several aspects. Howard and Clark (1985) point out that a

period of internment as incompetent contributes positively

to a later finding of insanity. As mentioned above, this

provides motivation for incompetent defendants to malinger

during the trial. Grisso (1986) relates the constructs of

legal insanity and legal incompetence in terms of

competency. The insanity criteria can be looked at in

terms of competency at the time of the offense.


The Problem and Related Research

The identification of malingering and defensiveness in

general and in forensic populations in specific have not

been popular topics in the psychological literature

(Rogers, 1988). A clinical psychologist's resources for

the identification of the above constructs are largely










interview information, psychological testing, and

psychophysiological methods. Rogers (1988) suggests that

psychophysiological methods are not practical to a

clinician outside of a laboratory setting or for select

cases for which significant amounts of money are available.

Resnick (1988) offers information which would help to

identify persons attempting to malinger psychosis during a

clinical interview. Brandt (1988) offers similar

information for those attempting to malinger amnesia.

Unfortunately, these are not standardized measures and they

rely heavily on a skilled interviewer with ample time to

spend in a one on one situation. Rogers (1986) pointed out

the haphazard manner in which clinical interviews are used

in identifying persons suspected of malingering or being

defensive. In response, he developed the Structured

Interview of Reported Symptoms sirsS). The SIRS is a

structured interview designed to identify malingerers based

on strategies observed by experts. The initial validity

and reliability data appear promising (Rogers, 1988).

Unfortunately, further psychometric data are needed before

it can become an accepted tool in clinical practice. A

major disadvantage of this type of instrument involves the

widespread use of such an interview as a screening

instrument. Manpower in large facilities is likely to be

insufficient and given the SIRS limited purpose it does not

provide breadth of information which would justify standard










implementation. It is probably best employed when

malingering is suspected.

Psychological testing provides a potentially fruitful

area for investigation concerning malingering and

defensiveness. Major areas of psychological testing

include objective and projective personality tests,

neuropsychological tests and intelligence tests.

Intelligence tests have limited usefulness in relation to

dissimulation in a forensic population who will likely be

more invested in attempting to appear psychotic. As such,

there are few studies involving forensic populations and

intellectual measures. Schretlen (1986) reviewed 11

studies involving malingering and intellectual tests. None

of the studies involved forensic populations. Schretlen

reported that response scatter was the most effective

detection strategy based on the rationale that fakers are

likely to fail items that non-fakers pass and fakers pass

items that non-fakers fail. This strategy is likely to

have limited effectiveness as most current intellectual

measures are obviously hierarchical. In addition to the

above reasons, intelligence tests are not likely to be

efficient as a screening measure in order to determine if

one is malingering or being defensive because they tend to

be time consuming and require individual administration by

a skilled examiner.











Neuropsychological tests are gaining acceptance in

court for civil suits regarding injury claims. However,

similar to intellectual tests they have a limited role in

relation to incarcerated forensic populations consisting of

persons found to be NGI and IST. Pankratz (1988) discusses

the role of neuropsychological measures in the

identification of malingering. He suggests using symptom

validity testing which requires a forced choice of the

testee. There is a percentage correct that one should get

simply from chance, and if the testee does significantly

worse than this dissimulation should strongly be suspected.

This is effective for somatic complaints (e.g., vision

problems) and memory problems. Lezak (1983) describes a

technique in which the testee is asked to memorize 15

different items. The number of items is stressed. In

actuality the testee needs to remember three or four ideas

to recall most of the items. If few items are recalled,

malingering is suspected. Schretlen (1986) reviewed five

studies involving the Bender-Gestalt test and suggested

when figures are intentionally distorted they might be

identified; however, further studies are needed before

detection strategies are implemented. In general,

neuropsychological tests appear to hold promise for the

identification of malingering in relation to specific

disorders and complaints; however, they do not appear to be

useful as a global screening instrument.











Both intellectual and neuropsychological tests have

limited usefulness in the detection of defensiveness. They

require skilled administration and have little connection

with the types of dissimulation seen in an incarcerated

forensic facility.

Projective personality techniques include the Rorshach

and the Thematic Apperception Test (TAT). Projective

techniques are grounded in psychodynamic theory which

states that one's inner perception shapes one's outer world

(Rabin, 1986). Projective techniques tap into this process

by presenting the testee with an ambiguous stimuli or task

which presents an unlimited number of responses. The

manner in which the response is made is considered to be a

reflection of the testee's inner world (Anastasi, 1988).

Inferences are made into the unconscious functioning of the

testee. These unconscious aspects are often not considered

by those who adhere to objective tests (McClelland, 1981).

As such, proponents of projective techniques point to the

relatively richer pool of data which they receive from

their techniques. In relation to forensic evaluations,

projective techniques are difficult to use in a legal

situation due to their theoretical grounding. Ziskin

(1981) points to the difficulty in using projective

techniques in the courtroom.

There has been little research in the areas of

malingering and defensiveness in relation to performance on











projective techniques. Stermac (1988) attributes this

paucity of information to the erroneous belief that these

methods are immune to dissimulation as a result of the

ambiguous nature of the projective task. The Rorshach is

the most widely used projective personality technique

(Anastasi, 1988). However, Carp and Shavzin (1950)

demonstrated that the Rorshach is not immune to

dissimulation. Both Schretlen (1986) and Stermac (1988)

have reviewed the related studies and have found this to be

a robust finding. Schretlen (1986) further points out that

while the Rorshach can be susceptible to malingering and

defensiveness, there does not appear to be a stable and

identifying pattern allowing for detection. Albert, Fox

and Kahn (1980) had experts from the Society for

Personality Assessment rate Rorshach protocols of six

psychiatric patients with a psychotic diagnosis and 18

undergraduate students. Six of the undergraduates were

informed of the nature of paranoid schizophrenia via an

audio tape. They and six other students were instructed to

take the Rorshach as if they were suffering from paranoid

schizophrenia. Results showed that the informed fakers

were diagnosed paranoid schizophrenic 72% of the time;

while the psychiatric patients were diagnosed as such 48%

of the time and the uninformed students were diagnosed as

paranoid schizophrenic 46% of the time. These results











clearly suggest that the Rorshach is susceptible to

dissimulation and that this is difficult to identify.

Exner (1986) reported that when using his

comprehensive scoring system malingering is identifiable.

Inparticular, malingerers tended to give more good form

responses with bizarre wording. Seamons et al. (1981)

suggested that malingerers tended to be more dramatic in

the content of their responses. Unfortunately, these

studies are sparse and are based on simulation designs.

Additionally, specific cut-off scores for the

identification of malingering have not been established

(Stermac, 1988). Schretlen (1986) suggests that

malingering should be suspected in the following

situations: reduced number of responses, slow reaction

times, frequent inanimate and animal movement responses,

vague or poor form responses, dramatic responses, an

attitude of pained compliance, frequent card rejections,

inconsistency, and failure to understand the task.

Exner (1986) reports several characteristics of a

Rorshach protocol that will help to identify a defensive

responder. These include a high frequency of popular

responses and frequent responses based on form; however,

Exner and Sherman (1977) demonstrated that the Rorshach is

difficult to reply to in a defensive manner when the testee

has been diagnosed as a schizophrenic.











Stermac (1988) reported that there were few studies

concerning dissimulation with other projective techniques.

In general, projective techniques hold promise for the

identification of malingering and defensiveness; however,

controlled studies on the topic are not currently available

to validate the current finding. Projective techniques

require a skilled assessor and are not likely to be

economical as screening instruments in large forensic

facilities.

Greene (1988) reported that significantly more studies

concerning the topic of dissimulation are available

concerning objective personality tests than any other type

of psychological test. This is largely due to the

Minnesota Multiphasic Personality Inventory (MMPI), which

is the most widely researched and used personality

measurement available to psychologists (Anastasi, 1988).

Objective personality tests, in general, and the MMPI, in

particular, offer several advantages as a screening measure

for the identification of malingering and defensiveness in

large forensic settings. The MMPI is easy to administer

and does not require professional training. Although it

takes the testee approximately one to two hours to complete

the MMPI, little supervision is needed and, therefore,

several persons may take it simultaneously at a minimal

expense in man-power terms. As a result the MMPI is easy

to administer in prison settings. The MMPI provides











information that is useful in addition to screening for

malingering and defensiveness. Scoring is objective and

computer scoring packages are readily available. Further,

the MMPI has scales that are designed to uncover attempts

at dissimulation. The MMPI appears to be the best

available screening measure for psychologists to administer

in forensic settings.

The MMPI was developed by Hathaway and McKinley in

1941 at the University of Minnesota (Dahlstrom, Welsh, &

Dahlstrom, 1972). The inventory consists of 566 items

which require the testee to respond either true or false.

The items used in the test were selected from an original

pool of more than 1000 items. The selection of these items

was based on empirical trials with criterion groups of

patients who were contrasted with normals.

There are currently ten clinical scales and three

validity scales which make up the standard MMPI profile.

Greene (1980) and Graham (1988) both point to the plethora

of specialized scales that have since been developed. The

validity scales were designed to detect the test taking

attitude of the testee and are related to the constructs of

malingering and defensiveness. The ten clinical scales are

related to psychopathology and consist of the following:

hypochondriasis, depression, hysteria, psychopathic

deviate, masculinity-femininity, paranoia, psychasthenia,

schizophrenia, hypomania and social introversion. There is











currently a debate concerning the use of empirically

derived scales. Briefly stated, empirical tests do not

offer a rationale for inclusion of items and, therefore,

those items that are selected might be a result of random

noise or variables related to the specific population

assessed (Jackson, 1971). This is one of the major

criticisms concerning the MMPI since the normative groups

were white residents of the state of Minnesota. For the

purposes of this study, the empirical debate suggests that

a priori hypothesis regarding the empirically derived

scales should be made conservatively and related only to

the constructs that the empirical scale directly measures.

As such, the empirically derived validity scales are

related to malingering and defensiveness; however, the

clinical scales are not and therefore, will not be included

in terms of a priori hypothesis in identifying

dissimulation. Many of the special scales since developed

were done so on rational grounds or were empirically

related to dissimulation and might be included in a priori

hypothesis.

The first of the validity scales is the Lie (L) scale.

The L scale consists of 15 items which are considered to be

virtuous by society but are unlikely to be honestly

endorsed. As such the L scale is a crude measure of

defensiveness. Unfortunately, Graham (1988) and Greene

(1988) both suggest that due to the obvious pull of these











items, a responder with minimal sophistication will score

low on this scale. Therefore, the L scale is likely to be

a good measure of psychological sophistication, but not a

good screening scale for identification of defensive

responders (Greene, 1980). However, the L scale might

prove useful in the differentiation of defensiveness in low

functioning populations.

The F scale consists of 64 items designed to detect

unique responses. Items for the scale were determined

based on infrequency of endorsement in the normative

sample. The F scale was considered to be the traditional

index of malingering; however, Greene (1988) points to

three reasons for elevation on this scale: (a)

inconsistent patterns of response (e.g., random

responding), (b) presence of actual pathology, and (c)

malingering. The F scale does not tap a unitary construct

and has components which are suggestive of self and social

alienation, as well as bizarre and psychotic experiences

(Dahlstrom, Welsh, & Dahlstrom, 1972). Elevations of the F

scale are often seen in pathological populations that are

not considered to be malingering. Greene (1988) suggested

that only extreme elevations on this scale are sufficient

to suspect malingering in lieu of serious pathology. Gough

(1950) suggested that the F scale is best interpreted in

relation to another validity scale (K scale). Greene

(1980) suggested that an extremely low score on the F scale











might be an indicator of defensiveness. This might be an

excellent scale to use in identifying defensive responders

in a forensic facility given that the high base-rate of

pathology and unusual circumstances experienced by the

residents prior to their internment would suggest a

relatively high F-score given accurate responding.

The K scale consists of thirty items that

distinguished persons who displayed significant pathology

yet had normal profiles from those with normal profiles and

no evidence of pathology. The K scale is considered to be

a measure of defensiveness (Greene, 1988). Mild to

moderate elevations in the K scale have been associated

with education level and are considered to be a reflection

of one's ego strength (Graham, 1988; Greene, 1980). Five

of the clinical scales have a proportion of K added to the

score in order to better differentiate the clinical

populations (McKinley, Hathaway, & Meehl, 1948).

Unfortunately, little research has been conducted to verify

this adjustment (Greene, 1980) or with special populations

such as in a forensic settings. As such, MMPI research

should be conducted with and without the K correction

(Butcher & Telligan, 1978). Research involving the K scale

has not differentiated populations well (Greene, 1980);

however, given the level of pathology in forensic settings,

the K scale might prove to be a valuable discriminator.











Currently the F and K scales are interpreted in

relation to each other when dissimulation is suspected.

The F-K index was posited by Gough (1950) as being able to

identify both defensive and malingering responders.

Specific cut-off scores have been suggested (e.g., Gough,

1950); however, Greene (1980) suggests that the population

assessed be considered in determination of establishing

cut-off scores. One would expect a relatively higher F in

populations with high pathology and a relatively higher K

in educated populations. Greene (1988) also suggests that

adolescents tend to have relatively higher F-scores. As

mentioned, assessing defensiveness using this scale might

be confounded by education level and healthy ego defenses;

however, this scale is likely to be more effective with

forensic populations. The F-K index offers a single

measure which addresses both malingering and defensiveness.

There are several other scales that might help to

identify persons attempting to dissimulate their MMPI

profile. The Gough Dissimulation Scale-Revised (DS-R)

consists of 40 items which differentiated a group of

neurotic patients from students and professionals asked to

simulate pathology (Gough, 1957). Anthony (1971) suggests

that the DS-R is more effective at identifying

dissimulators than any other scale. Cutting scores which

optimize identification of malingering have not been

established.











Wiener and Harmon (1948) developed a subtle-obvious

dichotomy for five of the ten clinical scales. This

division was based on a rational inspection of the items.

There are other subtle-obvious type scales reported in the

literature (Wales & Seeman, 1968; Burkhart, Christian, &

Gynther, 1978); however, there is less literature available

on these scales and normative data are not available for

them. The Weiner-Harmon scales can be used to help

identify both malingering and defensive responders. Graham

(1988) stated that the subtle-obvious scales are of little

interpretive value because most of the subtle items would

not have stood up to cross validation. However, Greene

(1988) suggests that one should be suspicious of

malingering if the T-score for the obvious items exceeds

the subtle items by 20 on each of the five scales. The

differentiation of defensiveness using the subtle-obvious

scales are more complicated. Given that the items on the

MMPI were empirically selected, there does not necessarily

need to be a theoretical rationale for an item's inclusion

on a scale. As such, the subtle items might not have any

connection with the criterion with which they are

correlated (Jackson, 1971). Further, Burkart et al. (1978)

found that dissimulators would over-endorse subtle items

when attempting to look favorable and would under-endorse

subtle items when attempting to look bad. Posey and Hess

(1985) have refuted the assertion that when attempting to











dissimulate favorably subtle scores tend to be

over-endorsed. This debate has not been settled.

Currently, the obvious items are interpretable and

contribute the most to the identification of malingering

and defensiveness. In a forensic setting with a high

base-rates of pathology, the under-endorsement of obvious

items might be a strong indicator of defensiveness.

The Carelessness Scale was developed by Greene (1978)

as a measure of consistency in responses. The scale

consists of 12 empirically selected pairs of items that are

considered to be psychological opposites. The scale might

be useful in the detection of a malingering responder as

the malingerer might be over-endorsing pathology and

therefore respond to items that are considered to be

psychological opposites.

Goldberg (1965) developed an index based on a linear

model designed to differentiate psychotic versus neurotic

responders. His index is (Scale L + Scale 6 + Scale 8) -

(Scale 3 + Scale 7). Forensic malingerers are likely to

attempt to portray themselves as psychotic and this index

might contribute to the identification of malingering. The

last two scales discussed have relatively little research

on dissimulation connected with them.

Wiggens (1966) developed 13 content scales based on

rational inclusion validated by psychometric techniques.

Lachar and Alexander (1978) claim that the content scales











are highly face valid and, therefore, susceptible to

dissimulations. There has been little research on this

topic. Given the above assertion concerning psychotic

portrayal, it is possible that the Wiggens content scale

for Psychoticism will be elevated in a forensic resident

attempting to malinger.

Megargee and Mendelsohn (1962) developed the

Overcontrolled-Hostility Scale based on their findings that

extremely assaultive persons tended to have lower hostility

ratings. Over-controlled responders are seen as rigidly

responding to situations involving aggressive impulses. A

forensic patient portraying himself in a defensive manner

is likely to be engaging in this type of response style.

Consequently, this scale might be beneficial in identifying

defensive responders.

Cuadra (1953) developed the Control Scale by comparing

profiles of outpatients and inpatients being treated for

the same disorder. The rationale for the scale was based

on the notion that staying out of the hospital required

self-control. This scale might aid in identifying

defensive responders who are attempting to present

themselves in a controlled fashion.

Research using the MMPI to identify dissimulation

usually involve intentional dissimulation which potentially

contributes a large amount of variance. For example, how

similar are profiles of college students asked to respond











as if they were trying to look psychopathological and a

person who is trying to avoid a long prison term by

appearing psychopathological? Other studies make an

assumption of malingering or defensiveness based on group

membership. Consequently, there is a dearth of studies

relating dissimulation to forensic populations who are

taking the test under realistic conditions.

Anthony (1971) had forty members of the United States

Air Force take the MMPI, first under standard conditions

and then with an instructional set to exaggerate pathology.

thirty-two of these exaggerated profiles were matched with

valid profiles of similar clinical configuration. Results

suggested that persons are able to exaggerate pathology

with the MMPI. Endorsement of the subtle items

differentiated the standard and exaggerated profiles with

the greatest proficiency; however, the obvious scale, DS-R

scale, F scale raw score and F-K index all differentiated

the two profiles with over 80% proficiency. The subtle

items were underendorsed in those groups attempting to

simulate pathology. The F raw score, F-K index and DS-R

scales all differentiated the matched MMPI profiles from

the simulated. The hit rates were much lower than the

first comparison.

This study is important because it points out several

flaws in this style of research. The intentionally

dissimulated profiles were relatively easy to identify











(compared to matched profiles), suggesting that this

situation might not mimic clinical administration. The

obvious and subtle items were significant differentiaters

in the first comparison but not in the second. This study

involved multiple administrations of the MMPI which enters

an added degree of variance. Finally, for forensic

purposes, this study involved using male military officers.

This might not be a good group from which to generalize

results.

Grow, McVaugh and Eno (1980) evaluated several

different MMPI detection techniques related to

dissimulation. One hundred and fifty undergraduate

students were asked to take the MMPI in conditions of

faking bad, faking good, and accurately. In general,

results suggested that the detection of the fake-bad group

was easier than the detection of the fake-good group.

Faking bad was best identified by an F > 14 or F-K > 6.

These counted for approximately 80% of the variance. The

DS-R and Obvious items also were excellent discriminators;

however, the subtle items did not contribute significantly

to detection. Faking good was best identified by F-K <

-11. This accounted for only 36% of the variance. The

cutting scores in this study are clearly not applicable to

a forensic setting which has a much higher base-rate of

pathology than undergraduate students.









28

Grow et al. (1980) cross-validated these results with

clinical populations. This was accomplished via a record

review of a state mental hospital and adjacent outpatient

clinic. The fake-bad group was identified by the following

criteria: (a) an MMPI profile which suggested more

pathology than presentation, (b) situational variables

which suggested motivation for faking bad, and (c) a

statement in the final report suggesting the possibility of

faking bad. The fake-good group was identified by the

following criteria: (a) an MMPI profile which suggested

less pathology than did the rest of the evaluation, (b)

situational variables which suggested motivation for faking

good, and (c) a statement in the final report suggesting

the possibility of faking good. The legitimate group was

based on MMPI profiles judged accurate by the testee's

therapist while undergoing therapy. Results of this

cross-validation were similar to the first experiment. The

cut-off scores for the fake-bad group were higher and less

of the variance was explained. The fake-good discriminator

was the same yet explained a greater percentage of the

variance. This suggests that with clinical groups, faking

good on the MMPI might be more easily detectable than in

normal groups. This study is obviously flawed by

circularity in that the MMPI profile is part of the

inclusion criteria to help differentiate MMPI profiles.

Further, one who is attempting to fake-bad or fake-good









29

would do so over the entire evaluation and not only on the

MMPI. As such, the criterion groups for this

cross-validation might be representative of variance

un-attributable to dissimulation.

Gallucci (1984) conducted a study designed in a manner

which did not directly instruct the participants to

dissimulate. Gallucci divided patients seeking

compensation or pension based on disability from the

Veterans Administration according to where in the process

the testee was in terms of receiving compensation.

Although not asked to dissimulate, the testees were taking

the test as part of an experiment. As such the

administration of the MMPI was still not "in vivo." Only

the F and F-K scales were looked at and results suggest

that the F-K is somewhat more effective between groups

discriminator than the F scale. This study is flawed by

the assumption of malingering based solely on group

membership. Severity of pathology in relation to the

various stages of seeking disability is a major confound.

Studies using the MMPI in forensic populations tend to

have many of the flaws mentioned above. Audubon and Kirwin

(1982) conducted a study designed to differentiate

defensive responders using the MMPI and 16 PF (another

objective personality inventory). Forty-five straight

admissions to a forensic unit of a state hospital were

administered the above instruments and 20 of these testees








30

were divided into high defensiveness and low defensiveness

groups based on their performance on these tests. Both the

low and high defensiveness groups did not appear to be

responding in a manner indicative of response bias. High

defensive responders tended to admit to less pathology, and

would admit to past pathology when making these admissions.

High defensive responders tended to have committed the more

serious crimes (e.g., murder). High defensive responders

tended to be least defensive regarding admissions of a

sociopathic orientation. This study is flawed by its

selection of groups based on the tests. Consequently,

results might be tautological in that persons are labeled

"defensive" because they don't admit to pathology and then

conclusions are made stating that "defensive" persons do

not admit to pathology. Further, the authors do not

address the notion that one is not admitting to current

pathology because it might no longer exist. Defensive

responding suggests that pathology exists that is not

admitted to. In this study the existence of pathology is

assumed but is not determined. Finally, given the

population studied, the "low defensiveness" group might be

a malingering group and they might be responsible for the

significant results.

Lanyon and Lutz (1984) also looked at the construct of

defensiveness in terms of MMPI profiles. The authors used

a population of sex offenders. The subjects were males who








31

were undergoing competency or insanity evaluations. Over

80% of the subjects were accused of child molestation and

convictions were either obtained or "anticipated by the

police." Subjects were divided into no denial, full denial

and partial denial based on admission of sexually deviant

behavior. The MMPI was administered as part of the

evaluation and not as part of an experimental protocol.

Results suggested that the full and partial denial groups

responded similarly on the MMPI and that they both differed

appreciably from the no denial group. A combination of the

validity scales seemed to discriminate denial from no

denial the best. Criterion selection for groups in this

study is appreciably better than the previous studies;

however, the assumption of group placement based admission

is tautological.

Walters (1988) compared the MMPI profiles of three

different populations within a federal penitentiary and

inferred motivation in terms of dissimulation based on

group membership. The three groups in this study were:

(a) those considered for parole (presumed defensive), (b)

those considered for solitary confinement (presumed

malingering), and (c) those considered for therapy

(presumed honest). The MMPI's were administered as part of

the standard operating procedure and were not part of an

experiment. Results suggest that it is easier to detect

exaggeration than defensiveness on the MMPI. The F, F-K,











Weiner Obvious Scales, Mania and Hysteria Subtle Scales,

and Ds-R scales all showed differences between the three

groups. Only the DS-R, Depression-Obvious and

Hysteria-Obvious differentiated the honest and defensive

groups. Hit-rates greatly improved on chance when

employed. The assumption in regard to dissimulation based

on group membership is a major flaw of this study. As

mentioned by the authors, the honest group might well have

been defensive. Further single-cell requesters are likely

to have a higher base-rate of pathology compared to the

other groups. However, the realistic use of the MMPI and

the populations used both combine to make this a useful

study.

A study involving the detection of malingering using

the MMPI and a forensic population was conducted by Wasyliw

et al. (1988); however, this study also made the assumption

of malingering based on MMPI profiles without external

validation. This study compared a group of persons

undergoing a competency/insanity evaluation (inferred to be

malingering) and a group already found NGI (inferred to be

a control group). Subjects were further divided according

to the severity of the crime. The MMPI profiles were

divided into not exaggerated, equivocal and clearly

exaggerated based on the validity scales. Comparison

between the two groups suggested that the evaluation group

was significantly higher on the F, F-K, Ds-R, and











Obvious-Subtle scales. There were not significant

differences between the two groups on measures of

defensiveness (e.g., L and K scales) or when the subtle

items were compared without the obvious ones. The range of

profiles showing evidence of malingering were between 26%

to 57% for the evaluation group and between 13% and 33% in

the control group. The potential punishment correlated

positively with the incidence of suspected malingering in

the evaluation group. This study has several flaws. There

were no external criteria for malingering except for the

MMPI. The control group has undergone treatment and the

evaluation group has not. Further, the control group has

incentive to be defensive and differences might well be due

to defensive responding and not malingering. This study is

important because it does suggest that assumptions

regarding group membership and dissimulation might be

accurate. Further, baseline information (based solely on

MMPI profiles) concerning the incidence of malingering is

offered. Finally, this study considers the relative

differences in external motivation (i.e., crime severity)

in terms of the incidence of malingering.

Walters, White and Greene (1988) conducted a study

involving malingering in forensic facilities using the MMPI

which did not infer malingering status based on group

membership or MMPI results. Inmates in a maximum security

federal prison were administered the Psychiatric Diagnostic









34

Interview (PDI) and those with significant psychopathology

based on the interview were included in the study. The

MMPI was also administered. Two psychologists rated these

inmates on a four point scale in which Level One was no

behavioral verification of illness suggested in interview

(malingering) to Level Four which was inmates displaying a

pattern of behavior consistent with the results of the

interview. Unfortunately, this study is flawed in that the

raters were aware of the standard scales on the MMPI and

the ratings were made approximately 1-14 months after the

interview. This study was important because it employed a

realistic situation and made an external effort at

verifying malingering behavior.

Walters et al. (1988) suggest that approximately 60%

of the inmates are accurately portraying their behavior.

The traditional MMPI scales did not significantly

differentiate the groups; however, five of the specialty

scales were significant. They were DS-R, Obvious total,

Obvious-Subtle ratio, Depression subtle and Mania obvious.

Only the DS-R and Depression subtle scales improved

accurate classification compared with the base-rates. This

study demonstrates the importance of establishing an

accurate baserate of disorder when attempting to predict

it's occurrence.











Purpose of Current Study and Specific Predictions

This study is intended to explore the incidence of

malingering and defensiveness in a forensic setting, as

well as, using the MMPI to differentiate these groups and

allow for identification based on MMPI results.

Relationships between the constructs of dissimulation,

legal status and severity of crime will be explored.

Further, inter-professional opinions concerning the

incidence of these constructs will be explored. As

indicated, the artificial nature of previous research into

the use of the MMPI in the identification of dissimulation

are the major flaws in this body of research. Realistic

studies involving patient populations and independent

criteria concerning the identification of malingering and

defensiveness are needed. This study addresses this issue

by having therapists, psychologists and psychiatrists

evaluate each of the residents in terms of a

malingering-defensiveness continuum. Inter-rater agreement

will result in three broad classifications of exaggerating

pathology, minimizing pathology or accurate portrayal of

pathology. This rating will not infer conscious intent.

Specific hypothesis concerning these broad constructs are

not offered a priori; however, their MMPI profiles will be

contrasted with the other broad groups and the relationship

between membership in these broad groups and severity of

crime and legal status will be explored.











Those seen as minimizing or exaggerating pathology

will be further classified into two more circumscribed

groups if such portrayal is considered to be the result of

conscious intent. These two groups will be classified as

malingering and defensive. These ratings will be based on

the professional's interaction with the resident and will

form the criteria groups. No predictions are made

concerning differences between the professionals and the

suspected incidence of dissimulation. However, it is

believed that; (a) more of the residents who are NGI will

be classified into the defensiveness group, (b) more of the

residents who are IST will be classified into the

malingering group, and (c) severity of crime will be less

significant a factor than legal status in terms of being

classified as a malingerer or defensive responder. These

hypothesis are important as their validation would suggest

that some of the assumptions made in previous research

described above were likely to be accurate.

The MMPI will be administered as part of the standard

operating procedure at the NFETC and therefore no variance

due to an artificial experimental situation will be

entered. The circumscribed groups (malingering and

defensive) will be contrasted with their broader groups

(exaggerating and minimizing); however, no specific

hypothesis are offered. It is hypothesized that the three

groups (accurate, malingering and defensive) will be









37

differentiatable based on the MMPI profile. In particular,

it is hypothesized that the malingering group will be

differentiated from the accurate group by the following

seven scales.

1. The DS-R will be greater in the malingering group

than in the accurate group.

2. The F-K index will be greater in the malingering

group than in the accurate group.

3. The K scale will be less in the malingering group

than in the accurate group.

4. The ratio of total obvious to total subtle items

on the Weiner-Harmon obvious-subtle scales will be greater

in the malingering group than in the accurate group. This

difference should hold for each of the individual scales

and will mainly be due to the increase in endorsement in

the obvious items and not due to preferential responding to

the subtle items.

5. The Wiggens Psychoticism scale will be higher for

the malingering group than in the accurate group.

6. The Goldberg index will be greater for the

malingering group than for the accurate group.

7. The Carelessness index will be greater for the

malingering group than the accurate group.

It is predicted that the defensive group will

differentiate themselves from the accurate group by the

following six scales.











1. The F scale will be lower in the defensive group

than in the accurate group.

2. The F-K index will be lower in the defensive group

than in the accurate group.

3. The K scale will be higher in the defensive group

than in the accurate group.

4. The overall ration of obvious to subtle items on

the Weiner-Harmon obvious-subtle scales will be lower in

the defensive group than in the accurate group. This will

likely be due to the relative underendorsement of obvious

items rather than the subtle items.

5. The content scale for Over-controlled Hostility

will be higher in the defensive group than in the accurate

group.

6. The content scale for Control will be higher in

the defensive group than in the accurate group.

In the above comparisons, individual analysis of each

of the separate subtle-obvious scales will also be

conducted. For those scales which do significantly

differentiate the above groups, further analysis will be

run in order maximize correct classification.














CHAPTER 2
METHODS


Setting and Subjects

The incidence of malingering and defensiveness, as

well as, a more broadly defined group of persons seen as

exaggerating or minimizing pathology in a forensic setting

was investigated. The forensic facility used in this study

was the North Florida Evaluation and Treatment Center

(NFETC), located in Gainesville, Florida. The NFETC is one

of three maximum security forensic treatment centers in the

state of Florida and is run by the Health and

Rehabilitative Services (HRS) branch of the state

government. The NFETC is a three unit, 10 building

facility which included a Mentally Disordered Sex Offender

Unit (MDSO) for part of the study; however, this population

was excluded from this research. The other two units are

herein referred to as general units and are composed mostly

of persons adjudicated Incompetent to Proceed (ITP), as

well as a smaller percentage of those found Not Guilty by

Reason of Insanity (NGI) (see Table 2-1).

This population has been considered as requiring a

restrictive setting. This is usually a result of being

seen as dangerous to either themselves or others. Those

requiring less restrictive settings are not sent to NFETC.





























Table 2-1: Legal Status of Residents at NFETC



Status Number Percent


ITP 98 68.1


NGI 45 31.2


Other 01 0.7


Total 144 100.0








41

A third general unit (also composed of persons adjudicated

ITP or NGI) was opened midway through this study and

persons admitted to this unit were included in this study.

All residents at the NFETC were male. The age range for

the residents included in this study was from age 18 to 59.

The subjects in this study approximate the racial

distribution of the residents at the NFETC and are

approximately 40% black, 40% white and 20% Hispanic. The

severity of the crimes of the residents is presented in

Table 2-2. One hundred and forty four residents at the

NFETC were included in this study. This represents all of

the admissions from April 1, 1989 to December 1, 1989

excluding those admitted for the MDSO program.

Upon arrival at the NFETC each resident participates

in an admissions interview, as well as, weekly evaluations

during building rounds which are conducted throughout his

first month following admission. Members of his

interdisciplinary treatment team are present throughout

this process. Treatment team members include a therapist,

psychiatrist and psychologist. The admissions interview is

conducted primarily by the psychiatrist; however, the other

professionals are actively involved in the interview.

Initial screening for both medical and psychological

therapies are conducted during the admissions interview. A

partial focus of this interview is diagnostic. The

resident has an opportunity to ask questions during this





























Table 2-2: Severity of Crimes per Resident at NFETC



Felony Status Number Percent


1 i 33 22.9


2 23 16.0


3 88 61.1


Total 144 100.0











interview. The resident is also informed of the general

expectations regarding his behavior during his treatment at

the NFETC. The weekly rounds are also conducted primarily

by the psychiatrists and focus on the maintenance and

reformulation of the treatment plan. This includes both

the medical and psychological treatments. In addition to

these planned interactions between the resident and his

treatment team, individual members of the team are

consulted as needed if problems arise.

The therapists have variable training; however, most

have some graduate work in the field of human services.

All of the therapists have undergraduate college degrees.

Twenty-five therapists participated in this study. They

have regular contact with the resident and are primarily

responsible for the implementation of non-medical therapies

and treatment plans. The therapists have the most frequent

contact with the resident among the treatment team members.

Therapists had case-loads varying from six to ten

residents. The psychiatrists are licensed physicians with

a specialty in the area of psychiatry. Six psychiatrists

participated in this study. They are primarily responsible

for the medical therapies. The psychiatrists' case-loads

were variable and ranged from twenty to seventy residents.

The psychologists are licensed in the field of psychology

(three at the doctoral level). There was one master's

level clinician. They are involved in the supervision of












non-medical therapies. The psychologists were each

assigned to a unit which includes approximately sixty

residents.

Evaluations of the residents were routinely completed

by Psychology Services within the first thirty days of

admission. The evaluations were tailored to the needs of

the individual resident; however, most evaluations included

a clinical interview and an attempt to administer the MMPI.

A copy of the psychology services evaluation is placed in

the residents NFETC record and is available to the

treatment team. The resident is informed of this, as well

as, the general limits of confidentiality concerning the

evaluation. At times; however, the resident is not

considered able to participate in formal testing. This is

usually because he is seen as grossly psychotic.

Psychology Services consists of the three licensed

psychologists that are mentioned above and two

masters-level assistants who are graduate students at the

University of Florida. The psychology assistants

participated in the thirty-day assessments but are not part

of the treatment teams.


Materials and Measures

All of the general admissions (those not admitted to

the sex offender unit) to the NFETC over an eight-month

period extending from April 1, 1989 to December 1, 1989









45

were included in this study. In addition, other residents

who were seen by Psychology Services and were administered

the MMPI as part of the assessment were included; however,

the same resident was not included twice. This situation

only occurred once during this period. Approximately

thirty days after the resident's admission the three

members of his treatment team mentioned above (therapist,

psychologist and psychiatrist) were sent a two scale

instrument (Figure 2.1) which was accompanied by

instructions (see Appendix). The first scale is a

continuum designed to differentiate the resident in terms

of whether he is seen as exaggerating or minimizing his

psychopathology. This scale is known as the dissimulation

(DISS) scale. The scale had a range of 0 to 15.5 c.m.

This did not allow for the classification of malingering or

defensiveness as defined in the introduction; however, it

allowed for broad classification of the resident as

accurate, exaggerating or minimizing without inferring

intent. The second scale is also a continuum and allowed

for the differentiation of that subset of residents who

were seen as exaggerating or minimizing that could be

further classified as malingering or defensive. This scale

is known as the conscious intent (CON) scale. This scale

also had a range of 0 to 15.5 c.m. Ratings were completed

before the psychological evaluation by Psychology Services.











Figure 2.1: Dissimulation Rating Form


Please rate the following resident prior to his thirty-day
review by Psychology Services based on your opinions formed
during your contact with him and historical information
available to you.


Name:


Date:


Place an X mark on the position in the line to which you
feel the above resident's portrayal belongs on the
following two continuums.


Resident behavior
and self-report
appear to be a
minimization of
his level of
psychopathology.


Resident behavior
and self-report
appear to be an
exaggeration of
his level of
psychopathology.


Above portrayal
appears to be
the result of
unconscious
intent or without
external motivation.


Above portrayal
appears to be
the result of
conscious intent
due to external
motivation.











In addition to the above ratings, information

concerning the residents legal status (ITP or NGI), and

severity of crime based on the Florida felony statutes

(Felony 1, Felony 2 and Felony 3) were gathered. Persons

classified as ITP were coded as 1 and persons classified as

NGI were coded as 2. One resident was admitted on a Baker

Act (a civil commitment) and he was excluded from the rest

of the study. Felony one offenses are the most severe and

life sentences were classified as Felony one. Felony three

offenses represent the least severe of the felonies and

misdemeanors were coded as Felony three. If there were

more than one offense, the most severe offense was

recorded. Felony one offenses were coded as 1, felony two

offenses were coded as 2 and felony three offenses were

coded as 3.


Procedure

There were 144 admissions during the period of this

study. Ratings forms were sent out via inter-center mail

for all of these. Responses were requested within three

days of receipt of the rating scales. Responses were

returned via inter-center mail. If no response came a

second mailing was initiated; however, if raters were no

longer ignorant of test results they were not sent ratings

forms. Therapists responded to 142 of the forms,

psychologists responded to 142 of the forms and








48

psychiatrists responded to 137 of the forms. The response

rates for each of the professional groups as well as

overall response rate are listed in percentage form in

Table 2-3. There were 134 residents for whom complete

rating forms were returned by all three professionals.

Therefore, 93% of the initial pool of admissions were

included in this study.

MMPI profiles that were administered by Psychology

Services were gathered and information regarding the

validity scales, clinical scales and additional scales

discussed in the introduction were collected. The raw and

T-scores for the clinical scales that were K-corrected were

gathered. The Wiener-Harmon obvious and subtle scores were

collected. Additionally, the following special scales were

also collected: Goldberg Index, Psychoticism, Control,

Over-controlled hostility, Gough dissimulation

index-revised, Carelessness index, Cannot Say scale.

MMPI's were not administered as part of this study. The

residents took the MMPI in the normal clinical setting and

the data was gathered via record review. The rationale for

the inclusion of the scales listed above are offered in the

introduction. The raw scores for those scales that are not

K-corrected (scales 2,3,5,6,0, the validity scales and

special scales) were not separately analyzed as this would

duplicate the analysis of the T-scores. The total obvious

and total subtle scores were derived by summing the





























Table 2-3:


Response Rates per Professional Group


Profession # Sent # Returned % Returned


Therapist 144 142 98.6


Psychologist 144 142 98.6


Psychiatrist 144 137 95.1


Total 432 421 97.5
i -









50

T-scores of the individual obvious and subtle scales. For

those scales which are K-corrected both the raw scores and

T-scores were analyzed. As such, the utility of the

K-correction with this population was assessed. The F-K

index was derived by subtracting the raw score on the F

scale by the raw score on the K scale.

MMPI profiles were included only if they were

considered "valid." The following criteria were considered

in determining whether to exclude MMPI profiles:

1. Cannot Say Scale > 30 were excluded. Greene

(1980) noted that when this scale is less than thirty the

profile is not appreciably altered.

2. Exclude all T profiles.

3. Exclude all F profiles.

4. Exclude all random profiles according to Greene

(1980).

This was done by both visual inspection of both the profile

and answer sheet. No completed MMPI profiles were excluded

from the study although several administrations were

discontinued by the examiners as a result of the above

processes.

Only complete MMPI profiles were included. No short

forms were included. The psychology clinic utilizes both

Form G and Form R of the MMPI. Although the 566 items are

presented in different order on the two forms, they include

the same questions. The MMPI could be administered in the











standard fashion or via an audio tape. The audio tape

version is available in both English and Spanish. All

residents with questionable reading skills are screened

using the Wide Range Achievement Test-Reading Test. Those

with below a sixth grade reading score were administered

the audio tape version. Further, those with vision

problems were also administered the audio tape version.

Residents with limited intellectual capacities would often

not be administered the MMPI. This determination is made

by the person who is conducting the assessment. Those

residents who completed the MMPI were coded 1. Those

residents for whom the MMPI was not attempted were coded 2.

Those residents for whom the MMPI was attempted but not

completed were coded 3. The failure to complete the MMPI

was due to factors related to the resident's behavior and

not external time considerations. There were 68 completed

MMPI profiles, this is 47% of the original sample.


Analysis

All statistical analysis were run using SAS

statistical programs or were done by hand. The differences

between the ratings of CON and DISS were compared between

the three professional groups. This was done using a

repeated measures ANOVA. As such, the normality of each of

the distributions on the CON and DISS variable by

professional was assessed. Appropriate transformations











were run for those that were not normally distributed.

Repeated measures was used since the same resident was

being rated by three separate professionals. Where there

were no significant differences between the professional

groups the scores were averaged together. Outlying groups

were analyzed separately on the dimension for which they

did not agree with the other professionals. Differences

between the categorical variables of legal status (status),

f-code and MMPI status (MMPI) were compared with the DISS

and CON ratings using analysis of variance. Differences

between the categorical variables were determined using Chi

square analysis. Chi square analyses were run comparing

fcode x status, MMPI x status and fcode x MMPI.

The DISS scores were divided into those seen as

exaggerating, accurate and minimizing based on the

distribution of these ratings. The range of responses for

the DISS variable was divided into thirds. Those in the

left most third were classified as minimizing. Those in

the middle third were classified as accurate. Those in the

right most third were classified as exaggerating. These

groups were compared on the collected MMPI variables using

a multivariate analysis of variance (MANOVA). The MANOVA

controlled for family-wise error common when making many

comparisons. Prior to running the MANOVA, the normality of

each of the collected MMPI variables was assessed. Those











that did not have a normal distribution were transformed

appropriately.

MANOVA analysis was also run to compare the added

dimension of CON and the MMPI variables. Conscious intent

was inferred when the resident received a rating greater

than or equal to 7.0 on the CON scale. Those persons who

were classified as exaggerating on the DISS scale and

received a CON rating of greater than 7.0 were classified

as malingering. Those persons classified as exaggerating

on the DISS scale with a CON score of less than 7.0 were

considered as exaggerating. Residents who were classified

as minimizing on the DISS scale and received a CON score

greater than 7.0 were classified as defensive. Residents

who were classified as minimizing on the DISS scale and had

a CON score of less than 7.0 were classified as minimizers.

The accurate group was not affected by the CON ratings.

Table 2-4 schematically represents the groupings mentioned

above.

Additionally, MANOVA's were run to compare MMPI

variables and the categorical variables of status and

fcode. The mean MMPI profiles for all the groups and the

overall MMPI profile were plotted. The MMPI variables

Cannot Say and Careless were analyzed using the Chi Square

technique because they were not distributed in a fashion

which allowed for transformation to normality. The

Careless variable was translated into a categorical




























Variable








Variable


Table 2-4: Classification of Residents



Group IRule


DISS Minimizing DISS Lower 1/3


Accurate DISS Middle 1/3


Exaggerating DISS Upper 1/3


CON Minimizing DISS Lower 1/3, CON < 7.0


Defensive DISS Lower 1/3, CON 7.0


Accurate DISS Middle 1/3


Exaggerating DISS Upper 1/3, CON < 7.0


Malingering DISS Upper 1/3, CON > 7.0
I









55

variable in which scores greater than four were considered

"high" and scores less than or equal to four were

considered "low." This was based on Greene's (1980)

recommendation of significance concerning this variable.

The Cannot Say scale was translated into a categorical

variable in which scores greater than five were considered

as "high" and scores less than this were considered as

"low."

Discriminant analysis, principle component analysis

and multiple regression analysis were run to determine if a

prediction equation into DISS categories could be obtained

using the data gathered. Further these analyses determine

the multicollinearity of the data and allow for the

determination of independent factors contributing to the

overall variance of the model.















CHAPTER 3
RESULTS


Inter-Professional Agreement

The mean dissimulation (DISS) ratings given by the

therapists, psychologists and psychiatrists regarding the

residents are presented in Table 3-1 along with the

standard deviations. The mean consciousness (CON) ratings

given by the therapists, psychologists and psychiatrists

regarding the residents are presented in Table 3-2 along

with the standard deviations. There are fewer

consciousness rating scores than dissimulation scores as

would be expected given the instructions which state that a

consciousness score is not needed for one who is seen as

accurate.

The DISS and CON ratings by each professional were

assessed for normality. The skewness, kurtosis, box plots,

normal probability plots and tree diagrams were evaluated.

The raters' responses were normal for both DISS and CON and

therefore repeated measures ANOVA's were appropriate to

analyze these data.

A repeated measures ANOVA was run to determine if

there were differences between the three professional

groups on the dissimulation dimension. The overall DISS

model yielded an F score of 2.14 which correlates to a





























Table 3-1: Dissimulation Rating per Professional Group



Standard
Profession Number Mean Deviation


Therapist 142 I 6.14 2.98


Psychologist 142 6.15 2.43


Psychiatrist 137 5.47 3.26




























Table 3-2: Consciousness Rating per Professional Group



Standard
Profession Number Mean Deviation


Therapist 115 7.11 3.70


Psychologist 126 7.50 2.93


Psychiatrist 130 6.08 3.83











P-value of .12. As such, there were no significant

differences between professional groups on the DISS scale.

This allowed for the creation of a new variable, NDISS =

(therapist DISS score + psychologist DISS score -

psychiatrist DISS score)/3, which is the average of the

three professional groups. The NDISS variable will be used

in the rest of the analysis involving dissimulation.

A repeated measures ANOVA was run to determine if

there were differences between the professional groups on

the CON dimension. The overall CON model yielded an F

score of 2.99 which correlates to a P-value of .05. As

such, there were significant differences between the three

professional groups. A post-hoc test, the studentized

range statistic (Weiner, 1969), was run to determine which

groups differed from each other. The psychologist and

psychiatrist comparison yielded a q-value of 6.76 which is

significant at the .01 level. The therapist and

psychiatrist comparison yielded a q-value of 4.90 which is

also significant at the .01 level. The therapist and

psychologist comparison yielded a q-value of 1.86 which was

not significant. Consequently, the psychiatrists differed

significantly from the therapists and psychologists while

the therapists and psychologists did not differ from each

other. As a result, a new variable VAR = (psychologist CON

+ therapist CON)/2 was created. This is an average of the

therapist and psychologist rating on the CON dimension. A











second variable VAR2 = (psychiatrist CON) is simply the

psychiatrist rating of CON. Subsequent analysis using the

CON dimension were run using both VAR and VAR2.


Group Differences

Differences between the variable NDISS and the

variables of felony code (fcode), legal status and ability

to complete the MMPI (MMPI) were assessed using analysis of

variance (ANOVA's). There were no significant differences

in the DISS ratings and the resident's legal status, felony

code or ability to complete the MMPI. ANOVA's were also

run comparing the VAR and VAR2 variables and the variables

of fcode, legal status and MMPI. As with DISS there were

no significant differences in the CON ratings and the

resident's legal status, felony code or ability to complete

the MMPI.

Additionally Chi-square analyses were run to compare

the categorical variables of status, MMPI and fcode. There

were no significant differences on these. Severity of

felony is not related to legal status or ability to

complete the MMPI. Neither is the resident's legal status

related to his ability to complete the MMPI. Overall, the

ratings of CON and DISS were not correlated with the

categorical variables of status, fcode or MMPI.











Division into Groups

The NDISS rating was divided into three groups. These

groups were known as accurate, minimizers and

exaggerators. This was done by dividing the range of

scores (0-12.9) on the NDISS variable into thirds. The

minimizers (min) were those with NDISS ratings of 0-4.3.

The accurate (acc) were those with NDISS ratings of

4.3-8.7. The exaggerators (egg) were those with NDISS

ratings of 8.7-12.9. The breakdown of residents in terms

of numbers and percentage of sample in each of the three

groups is presented in Table 3-3. Those with DISS scores

but not CON scores were excluded. This only affected the

size of the accurate group which was the majority of

responses. This was done so that the CON and DISS

populations would be the same sample size.

The groups were further divided based on the CON

ratings as delineated in the methods section. As mentioned

there were significant differences between the CON ratings

of the psychiatrists and the other two professional groups.

Therefore this additional division will include VAR and

VAR2. The breakdown of residents in terms of numbers and

percentage of sample for the five groups (egg, mal, acc,

min, and def) for VAR is presented in Table 3-4. The

breakdown of residents in terms of numbers and percentage

of total sample for the five groups for VAR2 is presented

in Table 3-5. The difference between VAR and VAR2 is that





























Table 3-3: NDISS Groups



Group Number %


Accurates 56 61.5


Minimizers 26 28.6


Exaggerators 09 9.9


Total 91 100.0


























Table 3-4: VAR by Groups


Group Number %


Accurates i56 61.5


Minimizers 15 16.5


Defensive 11 12.1


Malingerers 09 9.9


Exaggerators 00 0.0


Total 91 100.0
























Table 3-5: VAR2 by Groups


Group Number %


Accurates 56 61.5


Minimizers 19 20.9


Defensive 07 7.7


Malingerers 06 6.6


Exaggerators 03 3.3


Total 91 100.0
i _








65

the psychiatrists tended to rate lower on the CON variable

and, therefore, implied less conscious intent to

dissimulated behavior.

For all the groups there tended to be a large number

of persons considered to be accurate and a very small

number considered exaggerating or malingering. Further

these numbers are even smaller when considering those with

completed MMPI profiles. As such statistical significance

when comparing group differences on the NDISS, VAR and VAR2

variables will require large differences in clinical

significance.


MMPI Analysis

The overall mean MMPI profile was gathered and is

plotted in Figure 3-1. The collected supplemental scales

are plotted in Figure 3-2. The mean MMPI profile and

supplemental scale profiles for each of the categorical

group variables are presented in Figures 3-3 through 3-12.

It should be noted that since all of the MMPI profiles

garnered in the exaggerator category were seen by the

psychologists and therapists as malingering there is not a

separate figure for the VAR = malingering as this is

exactly the same profile as NDISS = egg. Further there is

no figure for VAR = egg as all of these were also

considered to be malingerers and are the same as the NDISS

= egg category.















0

O
0






E--4











oC

,,













.I
aa










































LO o mo 0 un C m o L o m 0 L
C m a am oD r- r- W LO U io g
-4










,,


C,/




c'





















i I I p I I i p pp



In O I 0 I'n 0 u'n 0 In 0 In 0 In

























aquoo


33AO


PTOD


tpAsa


T6noo


S6


q06


s9


q09


s7


qoV


SE




st

SZ
qo c


in 0 cn C o 0 o 0 o C) o IV
w3o co r r- %Do D n m -r -W

aeoOS 3I


















0

4J N

3 0


- 4-)
-.4E


In

0)
'qi
V


U 0 -" 0 nLn 0 00 0 In 0 U"
r r
aIOOS JI








69




o

4. N 1


O -74 a
III


a







SA
< PTo9





o,.. O 9 9 ,
S. no9
o s 6 \

0 m r
m q06





- q09


(, .._.._.X> ,,

= S q


,-- .-- qo






fc- -
SqoE


I i | a p a | *

On 0 O 0 Un 0 in 0 En 0 En 0 n
m co o r on n m

aJoOS Ji











70






a) 0 a)
-4 N -


I C '
aU -D c0

E I


0 m,














n. I 0 ." -. C




-C w




I \


-o s -, '
-In








1-1 y

a c-' / I





-3-


















-4~~~~~~
ero(


















S4) 4
-, N 4)

co -4


I I


Ln O L 0 Lo n L 0 Lo 0 Ln o Ln
O D 'o 0 DLn n m

alooS I


Juoo


I AO



PTOO


qtoAsd


qbnoo


S6


q06


S9 -4
s9






sp


qo


sC-





qor
sZ

s0t




















Oi
0 w,

Q) Mi 4J @)


Q) -, I -,"
- C -4
S E m
43


un) 0 i) 0 in 0 In o0 n O iO
ON OC OD co r- r- %Do %D lO I n


in


rn
-4
C.)

ci,


In O
O o
1- i-


alooS I

















0 wL
4) ID 41 4.
>J 4H ( 01
C 6 1O C
f)0'
O M "4

S- -x














> PTOD
C

o I









0
4-4





44 >
0 -





U .- \ s-
^ \ _q06


_ --9 -







c o V
~J -. 09







I SE 1
0000 q- 0 0
















tn 0 0 4 L 0n C t 0 L 0n Ln 0 V
CN a% co CD r- r- ko w LO O lw m










































































































In O0 L
0 0 0i
o- o I
1-1 1-


%D




LO


0)
*-4
.q. a
-
CO


o In 0 un 0 uo o Ln 0 n
CA\ oD O r- r- W w LO U) 'cO


aloos Ir













SH























q
0


















o
= I l 1 I = I = I = '


l/ l q 3uoo


4-) A -O





0 PTO
aoo






U q06




E < q 09











~ qoZ
0n LO a / ) n 9oL









OO -D < n LOa qO

azoos a;


















II II II

0 0 0
u 0) 0

>1 >1 >1
C: C; c
0 0 0
,-4 ,4 ,,,
cl r^ r'



I i 0


n 0o O
r-l ,-


o ur o 0 t
SO co


0 )n 0 D 0 0 U)
r- tD %D In Ln .-W


arooS I


\0









U



e














II II II
a ( aD
o 0o 0o
U L) u
>4 >1 >r



I
1cp^'-


o oi 0 o> 0 tc 0 L n
co r- C.- %D CD n LO -

83o0S I


I7uoo






PTo9


tjoAs3


qino9


S6


q06


S9



q09


Sv


qo0


SC


qoE


s


qOZ


o nO
.z. n








78

The individual MMPI scales were checked for normality.

The skewness, kurtosis, box plots, normal probability plots

and tree diagrams were evaluated. The following variables

were not normally distributed: raw 1, raw 7, raw 8, total

obvious and six-obvious. These were transformed to

normality using the log transformation. Upon

transformation these variables were normally distributed.

The careless and cannot say scales were not transformable

to normality. As mentioned, these variables were

transformed into categorical variables and analyzed

separately.

The variable NDISS (those seen as exaggerating,

accurate or minimizing based on the DISS variable) was

analyzed with the MMPI variables using a multivariate

analysis of variance (MANOVA). The overall MANOVA

statistics were not significant. Wilk's Lambda had an

F-value of 1.17 and a P-value of .29. Pillai's Trace had

an F-value of 1.20 and an P-value of .26. Univariate

analysis suggest that 12 of the MMPI variables were

significantly different as a result of group membership.

These are listed in Table 3-6. Analysis of the Tukey

statistics suggests that the differences tended to be a

separation of the exaggerators from the other two groups

with the exaggerators scoring higher on these scales

indicating greater pathology. The Chi-square analysis

comparing NDISS x cannot was significant (p = .023) with















Table 3-6: MMPI Variables Significantly Different
on the NDISS Variable


Variable


F-Value


P-Value


T2 6.59 .0027


T3 3.87 .0269


Raw 4 5.48 .0068


T4 4.57 .0146


T6 3.63 .0331


T7 4.79 .0122


TO 3.23 .0471


Total OB 4.41 .0168


20B 9.17 .0004


30B 4.99 .0102


40B 4.65 .0137


60B 3.86 .0267


i










the exaggerators tending to be high. The Chi-square

analysis comparing careless and NDISS was not significant.

The variable VAR (the added variable of CON considered

for the psychologists and therapists) was analyzed with the

MMPI variables using a MANOVA. The overall MANOVA

statistics were not significant. Wilke's Lambda had a

P-value of 1.36 and an F-value of .10. Pillai's Trace had

an F-value of 1.41 and a P-value of .07. Univariate

analysis suggest that 10 of the MMPI variables were

significantly different as a result of group membership.

These are listed in Table 3-7. Only the nine subtle

variable is different from those observed in the univariate

NDISS analysis. The further subgroupings created an even

smaller number of members per group. Analysis of the Tukey

statistics suggests that the differences tended to be a

separation of the malingerers from the other groups (note

that malingerers and exaggerators are synonymous for this

variable) with the malingerers endorsing items indicative

of greater pathology. The Chi-square analysis comparing

var with careless and cannot were both insignificant.

The variable VAR2 (the added variable of CON

considered for the psychiatrists) was analyzed with the

MMPI variables using a MANOVA. The overall MANOVA

statistics were not significant. Wilke's Lambda had an

F-value of 1.26 and P-value of .1393. Pillai's Trace had

an F-value of 1.32 and a P-value of .0843. Univariate


















Table 3-7: MMPI Variables Significantly Different
on the VAR Variable



Variable F-Value P-Value


T2 4.36 i .0081


T3 2.92 .0426


Raw 4 3.58 .0196


T4 3.04 .0369


T7 3.14 | .0330


Total OB 2.89 .0439


20B 6.14 .0012


30B 3.30 .0273


40B 3.06 .0360


9S 3.38 .0248











analysis suggest that five of the MMPI variables were

significantly different as a result of group membership.

These are listed in Table 3-8. These five variables were

also found to be different in the analysis of NDISS and

VAR. As with the VAR analysis the further subgroupings

resulted in even smaller numbers per group. Analysis of

the Tukey statistics suggested that the differences found

were between the malingering group and the accurate with

the malingerers endorsing items indicative of greater

pathology. The Chi-square analysis comparing VAR2 with

careless and cannot were both insignificant.

The variable VAR differentiated five more MMPI

variables based on group membership than did VAR2.

Individual T-tests were run comparing VAR and VAR2 on the

MMPI variables. There were no significant differences.

This is in part due to the small sample sizes. However,

this does suggest that the VAR2 measure does not add any

additional variance and does not need to be considered in

lieu of the average of the therapists and psychologists

since fewer differences were found on the MMPI variables

and group membership.

Overall, adding the CON dimension did not produce

greater differentiation between the groups. More

differences were found on the analysis of NDISS than the

analysis of VAR and VAR2. This is in large part due to the

small sample sizes. As such the utility of further


























Table 3-8: MMPI Variables Significantly Different
on the VAR2 Variable



Variable F-Value P-Value


T2 3.01 .0263


Raw 4 2.87 .0319


T7 2.52 .0510


20B 4.56 .0031


30B 2.50 .0537










analysis of the CON variable is dubious and was not

undertaken. It is important to note that with the

exception of one scale (nine subtle) all the observed

differences on the MMPI variables when analyzing the CON

variable were found when only considering the resident's

DISS score. The psychiatrists' ratings of CON accounted

for the least amount of variability on the MMPI. Further

the NDISS groupings are the most powerful as they include

the ratings of all three professional groups.

The variable of status was analyzed with the MMPI

variables using a MANOVA. The overall MANOVA statistics

were not significant. Wilke's Lambda had a F-value of .80

and a P-value of .73. Pillai's Trace had an F-value of .80

and a P-value of .73. Univariate analysis suggested that

25 of the MMPI variables were significantly different as a

result of the resident's legal status. These are listed in

Table 3-9. It should be noted that there are only two

groups in the status variable (NGI and ITP) and therefore

there was a relatively large n for each group. Analysis of

the Tukey statistics suggested that the residents who were

ITP tended to score higher on the MMPI scales indicative of

pathology or exaggeration of symptoms. The ITP persons

scored lower on measures of defensiveness (the Overcontrol

Scale and the K scales). The Chi-square analyses comparing

status with careless and cannot were both insignificant.











Table 3-9:


MMPI Variables Significantly Different
by Status


Variable F-Value P-Value

TF 10.22 .0023

TK 11.01 .0016

Raw 1 9.85 .0027

Raw 4 4.39 .0408

T6 6.17 .0160

Raw 7 12.07 .0010

T7 8.17 .0060

Raw 8 11.12 .0015

T8 9.30 .0035

Raw 9 9.09 .0039

T9 7.34 .0090

TO 6.17 .0161

Total OB 11.60 .0012

20B 9.24 .0036

2S 4.02 .0500

30B 10.50 .0020

40B 5.07 .0284

60B 11.26 i .0014

90B 8.01 .0065

Gough 12.49 .0008

Psych 11.67 .0012

Goldberg 5.23 .0269

Overcon 16.49 .0002

Control 9.60 .0031








86

Since all of the persons seen as exaggerating were ITP

and the greatest number of univariate MMPI differences were

found comparing status levels, a second MANOVA was run on

the status variables excluding those persons seen as

exaggerating. The overall MANOVA statistics were not

significant. Wilke's Lambda had a F-value of 1.15 and a

P-value of .3938. Pillai's Trace had an F-value of 1.15

and a P-value of .3938. Univariate analysis suggested that

22 of the MMPI variables were significantly different as a

result of the resident's legal status despite eliminating

those seen as exaggerating. These are listed in Table

3-10. As with the previous status analysis, the Tukey

statistics suggested that those who were ITP tended to

endorse items indicative of pathology and when compared to

the NGI group. The ITP group also tended to endorse fewer

items than the NGI group on scales measuring defensiveness.

The Chi-square analyses comparing status with careless and

cannot were insignificant.

Since there were significant status effects on the

MMPI, an additional MANOVA was run comparing only the ITP

group subdivided into two groups; those found exaggerating

and the rest of the sample. The overall MANOVA statistics

were not significant. Wilke's Lambda had a F-value of .05

and a P-value of .1053. Pillai's Trace had an F-value of

.99 and a P-value of .1053. Univariate analysis suggested

that 10 of the MMPI variables were significantly different













Table 3-10: MMPI Variables Significantly Different
by Status Excluding Exaggerators

Variable F-Value P-Value

TF 8.36 .0057

TK 8.50 .0053

Raw 1 7.51 .0085

T6 3.94 .0527

Raw 7 8.67 .0049

T7 5.44 .0237

Raw 8 8.19 .0061

T8 6.83 .0118

Raw 9 7.37 .0091

T9 6.06 .0173

TO 4.75 .0340

Total OB 8.51 .0053

20B 6.93 .0112

30B 8.42 .0055

3Sub 7.66 .0079

60B 7.86 .0072

90B 6.71 .0125

Gough 9.93 .0028

Psycho 9.24 .0038

Goldberg 4.93 .0310

Overcon 15.10 .0003

Control 8.51 .0053










as a result of exaggeration within the ITP population.

These are listed in Table 3-11. The group who was

considered to be exaggerating pathology had higher MMPI

scores on all 10 of the variables found to be significantly

different.

The variable of fcode was analyzed with the MMPI

variables using a MANOVA. The overall MANOVA statistics

were insignificant and none of the MMPI variables were

significantly different as a result of fcode. The

Chi-square analysis comparing fcode and careless and cannot

were both insignificant.


Regression Analysis

Multiple regression analyses were run to determine a

prediction equation for the dependent variable DISS and the

independent variables of the MMPI and demographic data.

Models were run using both forward/backward and step-wise

selection methods. The models which only included those

variables found significantly different on the DISS

variable produced the best results; however, the best model

found explained only 26% of the total variance, significant

at the .036 level. The variables used for this model were

T2, T3, Raw 4, T4, T6, TO and Three OB. The single

variable that explained the most variance Two OB, was

excluded from the complete model. The collinearity of the

12 MMPI variables that were found significantly different



















Table 3-11: MMPI Variables Significantly Different in ITP
Group Comparing Exaggerators and the Rest of the Sample



Variable F-Value P-Value


T2 7.38 .0102


T3 3.95 .0546


Raw 4 7.16 .0113


T4 5.88 .0206


T7 5.40 .0260


Total OB 5.12 .0300


20B 10.67 .0024


30B 5.21 .0287


40B 6.30 .0169


N 60B 4.15 .0493








90

in regard to DISS was assessed. Eight of the 12 variables

had tolerance factors of less than .10. As such

multi-collinearity is likely to be a problem and would help

account for the relatively small amount of total variance

explained by the model.

Given that all the Tukey differences found on the MMPI

variables when compared with DISS were between the

exaggerators and the other two groups, a multiple

regression was run using the dependent variable of DISS in

which only exaggerators and others accurates and

minimizers averaged together) were considered. This did

not improve the prediction equation as still only 26% of

the variance was explained. The prediction equation was

used to predict DISS and only one of the exaggerators was

accurately classified. The equation tended to classify

persons into the accurate group which is overweighted in

the sample and a high percentage of correct responses would

be gathered simply by predicting everyone as accurate. As

such, the prediction equation did not aid in providing

additional information for classification.

A principal components analysis was run to help

elucidate the small amount of variance explained by the

regression model. The first principal component explained

64% of the variance. The first five principal components

explained 95% of the variance. The first principal

component does not have any loadings greater than .45 but








91

has six variables with loadings greater than .30 (T2, T6,

T7, Two OB, Three OB and Four OB). The second component is

primarily a positive loading on T2 and a negative loading

on T6. The third component is primarily related to T4.

The fourth component is primarily related to T3. The fifth

component was primarily related to T7 and negatively

related to Three OB.














CHAPTER 4
DISCUSSION


Professional Agreement

All three groups of professionals who took part in

this study (psychologists, therapists and psychiatrists)

had exceptionally high response rates. As such, it is fair

to state that the responses received were fully

representative of the NFETC admissions during the course of

the study and not reflective of a response bias.

There was agreement among the professionals in terms

of the dissimulation scale. This would suggest that the

professionals are similar in their attributions of behavior

in terms of whether they view the resident as exaggerating

or minimizing his pathology. Further, given this agreement

among the raters the overall rating of dissimulation is a

powerful measure as it is the mean of the three different

raters.

However, the raters disagreed as to the conscious

intent of the dissimulation. The psychiatrists tended to

see behavior as less consciously motivated than did the

therapists and psychologists. This is likely to be the

result of the different training that psychiatrists receive

as compared to other human services professionals.

Psychiatrists are physicians and are indoctrinated into the