The impact of DSM-III-R and transitive diagnoses on perceptions of a man with a psychological problem


Material Information

The impact of DSM-III-R and transitive diagnoses on perceptions of a man with a psychological problem
Physical Description:
133 leaves : ill. ; 29 cm.
Raskin, Jonathan D
Publication Date:


bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1995.
Includes bibliographical references (leaves 123-132).
Statement of Responsibility:
by Jonathan D. Raskin.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002057555
notis - AKP5581
oclc - 33832554
System ID:

Full Text








For Shay


I am very grateful to the members of my doctoral

committee: Franz Epting, Harry Grater, David Suchman, Robert

Ziller, and Bernard Paris. Each has encouraged me in a

unique way, and all have helped me to develop a sense of

myself as an emerging professional psychologist. In

addition, I want to thank Dr. Martin Heesacker, whose

knowledge of research design and statistical analysis has

proved invaluable to me during my graduate school education.

I also want to acknowledge all the members of Dr. Epting's

research group, whose many suggestions positively influenced

the direction of my dissertation; I am especially thankful to

Stacy Gorman for helping me to test subjects.

I wish to pay special tribute to my advisor, Dr. Franz

Epting. He has been more than just an advisor during the

past five years. He has also been a good friend. His soft

spoken wisdom has given me much insight into both my

profession and myself. His influence on my development has

been immeasurable, and I thank him for it.

Finally, I want to acknowledge my wife, Shay Humphrey.

Her caring, love, and support are constant, and my

accomplishments would mean very little to me without her

there to share them with me.


ACKNOWLEDGMENTS ........................................... iii

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

LIST OF FIGURES ........................................... vii

ABSTRACT ................................................. viii

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

The Success of DSM-III and DSM-III-R..................... 2
Personal Construct Theory and Transitive Diagnosis........3
Comparing the Way People React to Psychological
Diagnoses .............................................. 6

REVIEW OF THE LITERATURE ....................................7

DSM Diagnosis.............................................. 8
Components of a DSM-III-R Diagnosis ...................13
Reliability and Validity ............................. 18
Transitive Diagnosis .................................. .. 31
Kelly on Transitive Diagnosis .........................32
Recent Developments ................... ................37
Common Criticisms of Psychiatric Diagnosis...............44
Difficulty Identifying the Mentally Ill ...............45
Mental illness As Myth ................................ 47
Labeling Theory and Mental Illness .................... 50
Other Labeling-Related Research .......................56
Sexism and Diagnosis .................................. 63
Racism and Diagnosis .................................. 68
Criticisms of Diagnosis in Relation to the Present
Study .............................................. 72
Hypotheses and Overview of the Study.....................73

METHOD ...................................................... 77

Materials................................................ 77

RESULTS .................................................. ... 85

DISCUSSION .................................................100

Evaluating the Jim Johnson Vignette..................... 102

Critical Evaluation of Results..............
Conclusion ..................................

............ 105
............ 110


APPENDIX B SOCIAL DISTANCE SCALE .........................119


APPENDIX D FUNNEL QUESTIONNAIRE .........................122

REFERENCES .................... .................... ........123

BIOGRAPHICAL SKETCH ...... ....................... .......... 133


Table 1. Descriptive Statistics for Dependent Measures
Yielding Significant Results For Type of Diagnosis...... 86

Table 2. Oblique Factor Loadings for Principal
Components Factor Analysis of Social Rejection
Measures ................................................ 86

Table 3. Pairwise Comparisons of Type of Diagnosis for
Social Distance Ratings Using Tukey-Kramer Procedure... 89

Table 4. Oblique Factor Loadings for Principal
Components Factor Analysis of Eleven Bipolar
Dimensions .............................................. 91

Table 5. Pairwise Comparisons of Type of Diagnosis for
Negative Outward Impression Ratings Using Tukey-Kramer
Procedure .............................................. 95

Table 6. Pairwise Comparisons of Type of Diagnosis for
Unacceptable Behavior Ratings Using Tukey-Kramer
Procedure .............................................. 95

Table 7. Descriptive Statistics for Dependent Measures
Yielding Significant Results For Type of Diagnostic
Vignette................................................. 98

Table 8. Pairwise Comparisons of Type of Diagnostic
Vignette for Social Distance Ratings Using Tukey-
Kramer Procedure........................................ 98

Table 9. Pairwise Comparisons of Type of Diagnostic
Vignette for Negative Outward Impression Ratings Using
Tukey-Kramer Procedure ..................................99

Table 10. Pairwise Comparisons of Type of Diagnostic
Vignette for Unacceptable Behavior Ratings Using
Tukey-Kramer Procedure.................................. 99


Figure page

Figure 1. The Effects of Type of Diagnosis on Social
Distance Ratings......................................... 90

Figure 2. The Effects of Type of Diagnosis on Negative
Outward Impression Ratings.............................. 94

Figure 3. The Effects of Type of Diagnosis on
Unacceptable Behavior Ratings........................... 95

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



Jonathan D. Raskin

August, 1995

Chair: Dr. Franz R. Epting
Major Department: Psychology

This study investigated the ways in which mental illness

diagnostic labels are perceived. It was hypothesized that

George Kelly's transitive diagnosis approach to interpersonal

problems--wherein the unique situation, background, and

personal constructions of clients are emphasized--would

result in less negative reactions to clients than a

traditional, DSM-III-R diagnosis approach--wherein clients

are understood by summarizing their psychological problems in

a mental illness diagnostic category. Subjects were 364

undergraduate psychology students. Each subject was

presented with a vignette describing a fictional man named

Jim Johnson. One group of subjects received a vignette

indicating that Jim Johnson had a psychological problem, and

this problem was described and assigned one of three


DSM-III-R labels. Another group of subjects received a

vignette indicating that Jim Johnson had a psychological

problem, but for these subjects the problem was described

according to one of three transitive diagnoses, each

comparable to one of the DSM-III-R diagnoses. A third group

functioned as a control, receiving no information about Jim

Johnson having ever had a psychological problem.

Results indicated that subjects responded negatively to

Jim Johnson when he was assigned either a DSM-III-R or

transitive diagnosis. Subjects desired greater social

distance from Jim Johnson when he was assigned any form of

diagnosis than when he received no diagnosis. Further,

subjects evaluated Jim Johnson's behavior as less socially

acceptable and his outward impression as less positive when

he was given a psychological diagnosis. Negative reactions

to Jim Johnson occurred regardless of type of diagnosis. The

results can be interpreted as indicating that merely

identifying someone as having a psychological problem is

stigmatizing for the person being labeled.


The ways in which psychiatric diagnoses affect

perceptions of persons has long been debated. Scheff (1966)

proposed the controversial idea that labeling an individual

in a particular manner influences both the labeled person's

self-perceptions and others peoples' perceptions of the

labeled person. In other words, Scheff suggested that labels

create a self-fulfilling prophecy, in which labeled persons'

behaviors come to conform with the labels assigned to them.

While research has not fully backed up all of Scheff's (1966)

labeling theory formulations, his ideas have been fruitful

for mental health professionals, who have been led to

investigate the impact of psychiatric diagnostic labels on

perceptions of persons. The current study hopes to add to

the expansive research literature in this area. In

particular, the current study examines differences in the way

people respond to a client labeled with a psychological

problem when the client is diagnosed using the Diagnostic and

Statistical Manual of Mental Disorders (Third Edition.

Revised) (DSM-III-R) versus when the client is diagnosed

using George Kelly's transitive diagnosis. In addition, the

study compares impressions of a person not diagnosed with any

psychological problem to impressions of a person diagnosed

with either DSM-III-R or transitive diagnoses. However,

before progressing any further it is necessary to briefly

introduce DSM diagnosis and transitive diagnosis, and to

differentiate them from one another.

The Success of DSM-III and DSM-III-R

Before DSM-III was published in 1980, psychiatric

diagnosis was generally regarded as both unreliable and

lacking in validity (Kirk & Kutchins, 1992). Due to many

improvements in the way in which diagnoses are made, DSM-III

and DSM-III-R have been widely embraced by the mental health

community, both in the United States and around the world

(APA, 1987). DSM-III has been translated into a variety of

languages, including French, German, Italian, Japanese,

Norwegian, Portuguese, Spanish, and Greek (APA, 1987). There

are numerous reasons for the success of DSM-III and DSM-III-

R. Both manuals adopted new and innovative ideas for

diagnosing mental disorders, and both have had reputations as

highly reliable scientific instruments (APA, 1980, 1987).

The elements that have made DSM-III, and later DSM-III-R, so

successful are discussed in detail in Chapter 2.

DSM-III and DSM-III-R propose distinct categories of

disorder. The American Psychiatric Association states:

The text of DSM-III-R systematically describes each
disorder in terms of current knowledge in the following
areas: essential features, associated features, age at
onset, course, impairment, complications, predisposing
factors, prevalence, sex ratio, familial pattern, and
differential diagnosis. (APA, 1987, p. xxv)

By providing such detailed information about each disorder,

clinicians are supposed to be able to reliably and

effectively diagnose each client with one or more specific

disorders. While some people have criticized DSM's

categorical approach (Adamson, 1989; Dumont, 1984; Vaillant,

1984), it has been useful in a number of ways. For example,

it has made it easy to include DSM categories in successive

versions of The International Classification of Diseases

(ICD), the medical community's disease manual. It has also

allowed clinicians a simple, quick way to convey information

about clients. Partly because of its ability to provide

diagnostic information so efficiently, categorical diagnoses

are useful for insurance purposes. While some objections to

the diagnostic approach found in DSM-III and DSM-III-R

remain, including challenges to claims that DSM-III and DSM-

III-R have enhanced reliability, both manuals have been

strikingly successful. In the introduction to DSM-III-R, the

authors observe that "the impact of DSM-III has been

remarkable. Soon after its publication, it became widely

accepted in the United States as the common language of

mental health clinicians" (APA, p. xvii).

Personal Construct Theory and Transitive Diagnosis

George Kelly's personal construct theory (PCT)

"emphasizes the creative capacity of the living thing to

represent the environment" (Kelly, 1991a, p. 6). PCT holds

that each person represents the environment through a system

of personal constructs. A personal construct is "a

representation of the universe, a representation erected by a

living creature and then tested against the reality of that

universe" (Kelly, 1991a, p. 9). Because personal constructs

are erected by individuals, they can also be torn down or

altered by individuals. People can try out new personal

constructions of the world if they find current constructions

limiting or stagnant, or even simply because they want to see

the world from an alternative perspective for a while. The

main idea is that there are an infinite number of ways in

which the world can be construed, and that people need not

cling solely to one construction of the world. In other

words, in PCT "the universe is real, but it is not

inexorable" unless a person "chooses to construe it that way"

(Kelly, 1991a, p. 6). The idea that the world can be viewed

in an infinite number of ways is known in PCT as constructive

alternativism (Kelly, 1970, 1991a, 1991b).

while constructive alternativism holds that DSM-III-R is

one potentially useful construct system for understanding

abnormal behavior, Kelly (1991b) wished to encourage

alternative ways to think about abnormal behavior. According

to constructive alternativism, the DSM-III-R construction of

abnormal behavior is one of an infinite number of ways to

think about abnormal behavior. Kelly felt that "a

psychological theory should be considered ultimately

expendable. The psychologist should therefore maintain

personal independence of his theory" (Kelly, 1991a, p. 31).

DSM-III-R's approach to abnormal behavior, while often useful

in conceptualizing clients, is only one method of construing

such behavior. Since people are "always faced with

constructive alternatives" they "need not continue

indefinitely to be the absolute victim either of past

history or present circumstances" (Kelly, 1991a, p.

30). In other words, we need not be enslaved by our own

constructions (Kelly, 1991a, 1991b).

Kelly (1969, 1991b) did not care for categorical

approaches to psychological diagnosis because he felt that

they often failed to help clinicians comprehend the unique,

individualized construct systems of their clients. Rather

than employing categorical diagnostic labels, Kelly preferred

an idiographic approach to understanding psychological

problems. By employing various construct elicitation

techniques, Kelly felt that clinicians could effectively gain

access to each client's construct dimensions, and then could

evaluate these dimensions as part of understanding the

client's psychological problem. Little research has been

done on Kelly's (1991b) original formulation of transitive

diagnosis. However, recently Faidley and Leitner (1993) made

an attempt to actively employ transitive diagnosis with real

life clients. A more detailed discussion of transitive

diagnosis, including work by Landfield and Epting (1987) and

Faidley and Leitner (1993), will be presented in Chapter 2.

Comparing the Way People React to Psychological Diagnoses

Because transitive diagnosis and DSM-III-R diagnosis are

substantially different from one another, it seems logical to

hypothesize that each might impact differently on perceptions

of persons. Data on the effects of traditional, DSM-style

diagnoses on impressions of persons has provided mixed

results, at best. However, it does seem that, at least under

certain circumstances, traditional diagnostic labels produce

negative reactions towards the individual labeled. In the

present study, an initial attempt was made to investigate the

degree to which the effects of psychiatric labeling can be

lessened as a function of the method of diagnosis used.

Transitive diagnosis and DSM-III-R diagnosis were compared to

each other and to a no-diagnosis control group in order to

examine the extent to which each kind of diagnosis impacts on

perceptions of persons, and the extent to which transitive

diagnosis might produce less negative perceptions of persons.


The first section of this chapter focuses on DSM-style

diagnosis. It presents literature pertaining to the origins

and development of DSM-style diagnosis, focusing primarily on

the two most recent versions of the manual, DSM-III and DSM-

III-R. Attention is paid to the history of the DSM, using

DSM-III-R to make diagnoses, and to issues of diagnostic

reliability and validity. The second section of this chapter

discusses Kelly's (1991a, 1991b) transitive diagnosis. It

details Kelly's original outline of transitive diagnosis, as

well as the more recent approaches of Landfield and Epting

(1987) and Faidley and Leitner (1993). Further, it

summarizes PCP literature that explicates the differences

between DSM-III-R diagnosis and transitive diagnosis. The

third section of the chapter addresses common criticisms of

psychiatric diagnosis, with an attempt to relate these

criticisms, where applicable, to both DSM-III-R and

transitive diagnosis. Finally, the fourth section of this

chapter explains the major hypotheses of the present study.

The study compares the impact that DSM-III-R and transitive

diagnoses have on laypersons' impressions of a man described

as experiencing a past psychological problem.

DSM Diagnosis

The history of psychiatric diagnosis can be traced all

the way back to 2600 BC, when disorders such as melancholia

and hysteria were discussed by the Egyptians and Sumerians

(Webb, DiClemente, Johnstone, Sanders, & Perley, 1981).

There have been many well known historical figures who have

played important roles in the development of psychiatric

diagnosis, including Hippocrates, Sydenham, Pinel, and

Kraepelin (Webb et al., 1981) Emil Kraepelin is perhaps most

often referred to when discussing important influences on

present day DSM diagnosis. His early attempt at psychiatric

classification during the latter part of the nineteenth

century in Heidelberg, Germany, greatly influenced later

researchers interested in cataloguing psychological

disturbances (Kirk & Kutchins, 1992; Webb et al., 1981). In

Kraepelin's system, groups of symptoms were monitored over

time in order to identify mental diseases; his system

provided thorough descriptions of disorders, with the hope

that each disorder's etiology would eventually be discovered

(Zigler & Phillips, 1961). Kraepelin's idea of organizing

symptoms into descriptive, categorical diagnoses can be

clearly seen in the various editions of DSM (Kirk & Kutchins,


Kraepelin's system was not the only early attempt at

psychiatric diagnosis. Numerous efforts have been made in

the United States during the past century and a half. Most

of the early American psychiatric nosologies were developed

in part due to social demands, such as the requirements of

the U.S. census (Kirk & Kutchins, 1992). In fact, the first

categorization of mental disorders in the U.S. was performed

as part of the 1840 census. Idiocy was the only category, and

it included those deemed insane (Kirk & Kutchins, 1992; Webb

et al., 1981). By the 1880 census, seven mental disorder

categories existed: mania, melancholia, paresis, monomania,

dementia, epilepsy, and dipsomania (Kirk & Kutchins, 1992;

Webb et al., 1981).

Census officials became interested in a more

standardized diagnostic system, and eventually asked the

American Medico-Psychological Association to develop such a

system (Kirk & Kutchins, 1992). In 1918, the American

Medico-Psychological Association published the Statistical

Manual for the Use of Institutions for the Insane, a manual

which contained 22 distinct diagnostic categories (Kirk &

Kutchins, 1992). Ten editions of this manual were used

between 1918 and 1942, and in 1935 it was incorporated into

the American Medical Association's Standard Classified

Nomenclature of Disease (Kirk & Kutchins, 1992; Spitzer &

Williams, 1983).

The American Psychiatric Association published the

original edition of the Diagnostic and Statistical Manual of

Mental Disorders (DSM-I) in 1952 (APA, 1987; Eysenck,

Wakefield, & Friedman, 1983; Kirk & Kutchins, 1992). It

contained 106 categories (Kirk & Kutchins, 1992), and greatly

reflected the psychobiologic view of Adolf Meyer (APA, 1987).

Meyer's influence was readily seen in the manual's use of the

term "reaction," which reminded readers of Meyer's

conceptualization of mental disorders as "reactions of the

personality to psychological, social, and biological factors"

(APA, 1987, p. xviii).

Sixteen years passed before a revision of DSM-I

appeared. DSM-II was significant in a number of respects.

First, the term "reaction" was eliminated (APA, 1987).

Second, the number of diagnostic categories increased from

106 to 182 (Kirk & Kutchins, 1992). Third, DSM-II marked the

first time that DSM was planned in conjunction with the

International Classification of Diseases manual, which at

that time was in its eighth edition (ICD-8). Both DSM-II and

ICD-8 were published in 1968 (APA, 1987). Some clinicians

found DSM-II biased towards a psychoanalytic perspective

(Kirk & Kutchins, 1992). However, APA (1980, 1987)

maintained that--other than inclusion of the term "neurosis"-

-DSM-II was largely theoretical.

Work on DSM-III began in 1974, with the appointment of

the Task Force on Nomenclature and Statistics (APA, 1987;

Eysenck et al., 1983; Kirk & Kutchins, 1992). The Task Force

was headed by Robert L. Spitzer, and its goal was to revise

DSM-II so that the new version would be consistent with ICD-

9, but would also provide more detail about each mental

disorder than ICD-9 (APA, 1980, 1987; Kirk & Kutchins, 1992;

Webb et al., 1981). DSM-III was different from DSM-II in

numerous ways. First, the term "neurosis" was eliminated.

Second, the categories and accompanying five-digit codes for

each disorder that were used in ICD-9 were also used in DSM-

III; this resulted in almost all DSM-III categories being

included in ICD-9 (APA, 1980, 1987). Third, DSM-III marked

the first attempt to answer critics who claimed that DSM

diagnostic categories lacked reliability; extensive field

trials were performed as part of the process of developing

DSM-III, in an effort to enhance the diagnostic reliability

of its categories (APA, 1980, 1987; Eysenck et al., 1983;

Kirk & Kutchins, 1992; Spitzer, Endicott, & Robins, 1975;

Spitzer, 1984a, 1984b). Fourth, DSM-III introduced the

multiaxial approach to diagnosis, wherein diagnoses are made

along five specific axes. Perhaps most importantly, the

introduction of DSM-III marked a crucial turning point in the

manual's history. Unlike its predecessors, DSM-III was

widely embraced by the mental health community (APA, 1987;

Kirk & Kutchins, 1992), and was praised as a major step in

improving the diagnosis of mental disorders.

DSM-III-R was published in 1987 (APA, 1987; Kirk &

Kutchins, 1992). The APA's (1987) introduction to DSM-III-R

cites the emergence of new reliability data and the need to

refine diagnostic criteria as the primary reasons for

revising DSM-III. The introductory section of DSM-III-R

emphasized that each revision of DSM is only one "still

frame" in the scientific process of understanding mental

disorders, and that DSM-III-R was simply the latest still

frame. Major changes in DSM-III-R involved revising

diagnostic criteria, adding new disorders, and eliminating a

few old disorders. Childhood and adolescent disorders, in

particular, were substantially revised for DSM-III-R

(McConville & Steichen-Asch, 1990). Despite its many

changes, DSM-III-R maintained most of the innovations first

used in DSM-III. As with DSM-III, diagnostic criteria,

multiaxial diagnosis, and use of five-digit codes consistent

with ICD-9 were all part of DSM-III-R. Despite DSM-III-R's

conceptual similarities to DSM-III, Zimmerman (1990) felt

that the revisions made in DSM-III-R were so substantial that

it should have been called DSM-IV.

At the time of this writing, DSM-IV (APA, 1994) has been

recently published. This fifth version of the DSM was

originally planned to come into use in 1993 (Kendell, 1991;

Widiger, Frances, Pincus, Davis, & First, 1991), but was

eventually set for publication in May, 1994 (Laurence, 1994;

Roan, 1994). DSM-IV and ICD-10 were expected to be more

similar to each other than were DSM-III and ICD-9 (Kendell,

1991). While some researchers have criticized DSM-IV,

questioning the necessity for a revision of DSM-III-R so soon

after its publication (Zimmerman, 1988, 1990), others see

DSM-IV as an important step in the ever evolving science of

psychiatric diagnosis (Kendell, 1991; Widiger et al., 1991).

The present study was conducted prior to DSM-IV's

publication. However, the DSM-III-R descriptions used in the

present study remain similar to those found in DSM-IV.

Components of a DSM-III-R Diagnosis

DSM-III marked a major overhaul of DSM-style diagnosis.

It emphasized a descriptive approach to cataloging mental

disorders, introduced diagnostic criteria and multiaxial

diagnosis, and took great strides in addressing the issue of

diagnostic reliability. As mentioned above, most of the

changes introduced in DSM-III were also found in its 1987

revision, DSM-III-R. Below, the primary elements important

in making a DSM-III-R diagnosis are briefly discussed.

Definition of mental disorder. DSM-III-R cautions that

while it "provides a classification of mental disorders, no

definition adequately specifies precise boundaries for the

concept 'mental disorder'" (APA, 1987, p. xxii).

Nevertheless, DSM-III-R defines a mental disorder as being "a

clinically significant behavioral or psychological syndrome

or pattern that occurs in a person" (APA, 1987, p. xxii).

This syndrome or pattern must be considered "a manifestation

of a behavioral, psychological, or biological dysfunction in

the person" (APA, 1987, p. xxii). While proposing a

categorical system for cataloging mental disorders, DSM-III-R

does not claim that mental disorders have distinct

boundaries; the degree to which various disorders

qualitatively differ from each other continues to be

researched and debated (APA, 1980, 1987). At the same time,

DSM-III-R emphasizes that the diagnoses it contains do not

classify people, but "disorders that people have" (APA, 1987,


Descriptive approach. DSM-III-R's stated goal is to

present theoretical descriptions of mental disorders; this

is because the etiology behind most mental disorders is

unknown (APA, 1980, 1987). Further, the authors of DSM-III-R

felt that the inclusion of theoretical explanations for each

mental disorder would result in conflict between clinicians

of different theoretical orientations. As a result, DSM-III-

R only offers descriptions of disorders; it does not try to

explain the causes of the disorders.

Diagnostic criteria. Prior to DSM-III, diagnostic

criteria were not employed in diagnosing mental disorders

(Adamson, 1989; APA, 1980, 1987). However, in an effort to

enhance reliability, diagnostic criteria were introduced in

DSM-III. The diagnostic criteria for a disorder consists of

a list of various behaviors that together constitute that

disorder; as in DSM-III before it, the description of each

DSM-III-R disorder is accompanied by diagnostic criteria. In

making diagnoses, clinicians determine whether their clients'

behaviors coincide with the required number of behaviors

listed in the diagnostic criteria. Clients whose behaviors

meet the criteria are diagnosed with the disorder at hand.

Interest in developing diagnostic criteria for the

purpose of enhanced reliability can be traced back to the

early 1970s. One of the first attempts to develop such

criteria reached fruition in 1972, with the "Feighner

criteria" (Adamson, 1989; Kirk & Kutchins, 1992); the

"Feighner criteria" greatly influenced the criteria

eventually used in DSM-III and DSM-III-R (Adamson, 1989; Kirk

& Kutchins, 1992). Diagnostic criteria allowed for more

general agreement about the behaviors constituting each

disorder. However, while helpful for reliability purposes,

the authors of DSM-III-R warn that "it should be understood

that for most of the categories the diagnostic criteria

are based on clinical judgment, and have not been fully

validated" (APA, 1987, p. xxiv). Thus, while diagnostic

criteria have enhanced reliability, validity issues remain.

More on reliability and validity shortly.

Multiaxial evaluation. Williams (1985a) presents a

fascinating and thoroughly detailed history of multiaxial

approaches to psychiatric diagnosis. In 1952, the Danish

Psychiatric Association adopted a biaxial diagnostic system,

a forerunner of the DSM-III multiaxial system (Williams,

1985a). Proposals for a multiaxial approach in DSM began to

be discussed in the 1960s, and by the time a first draft of

DSM-III was circulated in 1976, a multiaxial system was being

developed for the manual (Williams, 1985a).

DSM-III-R describes its multiaxial system as one in

which "every case is assessed on several 'axes,' each of

which refers to a different class of information" (APA, 1987,

p. 15). Axes I and II are the axes used to classify all

major mental disorders. The disorders categorized along Axis

II constitute personality and developmental disorders, which

are believed to begin in childhood or adolescence, and which

are often overlooked "when attention is directed to the

usually more florid Axis I disorder" (APA, 1987, p. 16). All

disorders not categorized as personality or developmental

disorders are listed on Axis I. The distinction between Axis

I and Axis II allows clinicians to make simultaneous

diagnoses along both axes, or along one or the other axis as

is necessary. However, DSM-III-R also points out that it is

often appropriate to diagnose clients with more than one

disorder along either Axis I or Axis II. For example, both

DSM-III and DSM-III-R (APA, 1980, 1987) emphasize that a

client may have multiple Axis I disorders (e.g. a

psychoactive substance abuse disorder and a mood disorder),

as well as multiple Axis II disorders (e.g. a specific

developmental disorder and a personality disorder), and both

manuals instruct clinicians to diagnose all disorders for

which a client meets the necessary criteria. Further, when

no Axis II disorder is diagnosed, that axis can be used to

list specific personality traits. When multiple diagnoses

are made, clinicians are supposed to indicate which is the

"principal diagnosis"--that is, the disorder which was

principally responsible for the client's entering treatment.

Axis III is not actually used for diagnosing mental

disorders. Rather, it is used by clinicians to indicate

other physical illnesses afflicting a client that are not

mental disorders. Axis III is used to supplement diagnoses

made along Axes I and II, and to provide information about

other physical ailments that may be relevant to a client's

case (APA, 1987).

Axis IV of DSM-III was heavily criticized (Skodol, 1991;

Williams, 1985a). As a result, it was revised for DSM-III-R.

In DSM-III-R, Axis IV constitutes the Severity of

Psychosocial Stress Scale. Evaluations of psychosocial

stress are made along a six point scale, with "1" being "no

stress" and "6" being "catastrophic stress." Next to each

point on the scale are listed several examples of what

constitutes that particular level of stress. Two versions of

the scale are provided, one for adults and one for children

and adolescents. For example, a Severity of Psychosocial

Stress score of "3" is considered to constitute "moderate"

stress. For an adult, the following events are considered

instances that cause moderate stress: "marriage, marital

separation, loss of job, retirement, miscarriage" (APA, 1987,

p. 11). For a child, these events are considered instances

that cause moderate stress: "expelled from school, birth of a

sibling" (APA, 1987, p. 11). Clinicians are instructed to

make evaluations of psychosocial stress "based on the

clinican's assessment of the stress an 'average' person in

similar circumstances and with similar sociocultural values

would experience from the particular psychosocial

stressor(s)" (APA, 1987, p. 19).

Axis V also changed substantially between DSM-III and

DSM-III-R. In DSM-III-R, Axis V is the Global Assessment of

Functioning Scale. This scale ranges from 0 to 90. Lower

numbers indicate less ability to function effectively in

one's environment. For example, a score of "30" indicates

that "behavior is considerably influenced by delusions or

hallucinations" (APA, 1987, p. 12). By comparison, a score

of "60" indicates only "moderate symptoms" (p.12), while a

score of "90" indicates "absent or minimal symptoms" (p.12).

A score of "0" indicates not enough information to make a

judgment. Two assessments along Axis V are made. The first

describes current level of functioning at the time of a

client's diagnostic evaluation; the second indicates the

clinician's judgment regarding a client's highest level of

functioning within the past year.

Reliability and Validity

Research and debate addressing reliability and validity

in psychiatric diagnosis are not new. Zigler and Phillips

(1961), in their well known article critiquing psychiatric

diagnosis, discussed reliability and validity at length and

cited studies which found both high and low diagnostic

reliability levels. The publication of DSM-III in 1980

marked a turning point in the debate over diagnostic

reliability. Part of the reason that DSM-III and DSM-III-R

have been more successful than previous editions of the DSM

is because of the purported enhanced reliability of the

diagnostic categories they contain. DSM-III and DSM-III-R

have generally been considered highly reliable diagnostic

instruments (APA, 1980, 1987). Below, research on the

reliability and validity of DSM-III and its successor, DSM-

III-R, is reviewed.

Reliability and validity defined. In the most general

sense, validity is defined as the best approximation of the

truth or falsity of a proposition (Cook & Campbell, 1979).

While perfect validity is impossible--we can never

distinguish with complete certainty true from false--we can

know what has not yet been ruled out as false (Cook &

Campbell, 1979). A specific kind of validity relevant to

diagnosis is construct validity. Construct validity involves

the degree to which a theoretical construct measures what it

purports to measure (Cook & Campbell, 1979; Smith & Glass,

1987). It is crucial that diagnostic constructs--the many

categories employed when classifying abnormal behaviors--

measure what they claim to if they are to be useful for

clinicians; a clinician employing a diagnostic system unable

to differentiate between normality and disorder benefits very

little from using that system.

Important in achieving a high level of validity is

reliability. Reliability constitutes the degree to which a

measure is stable and consistent (Cook & Campbell, 1979). A

reliable measure provides similar results each time it is

taken. In categorical diagnosis, this is important. If

one's measures change each time they are taken, then knowing

what category of disorder a client belongs in will be

difficult. In a highly reliable system, clinicians reach the

same diagnosis each time they assess a client. Further, a

reliable system allows clinicians to reach the same diagnosis

of a client independently of one another. Most importantly,

a valid system cannot be achieved without a high degree of

reliability. Though a reliable system does not guarantee a

valid system, a valid system must be reliable (Cook &

Campbell, 1979; Smith & Glass, 1987). Because more attention

has been paid in the literature to diagnostic reliability

than diagnostic validity, most of the research presented

below addresses the former.

DSM-III field trials. Appendix F of DSM-III contains

the results of field studies performed in order to evaluate

the reliability of diagnostic categories found in DSM-III

(APA, 1980). Two phases of reliability testing were

undertaken as part of these field trials. Phase One used

criteria from a January, 1978 draft of DSM-III; Phase Two

used revised criteria (APA, 1980). DSM-III provides kappa

statistics indicating the degree of reliability for the major

classifications of mental disorders. The kappa statistic

measures improvements in clinician diagnostic agreement

beyond mere random chance (APA, 1980; Kirk & Kutchins, 1992;

Spitzer & Fliess, 1974), and is cited in DSM-III as

indicating improved reliability over past diagnostic

classification systems. Herein, the field trials of DSM-III

are discussed because there is no section of DSM-III-R that

reports the results of reliability field studies; research on

DSM-III-R's reliability was not mentioned at the time of its

publication or included in the manual (Kirk & Kutchins,


DSM-III indicates that a kappa statistic of 0.7 or

higher indicates a high degree of reliability (APA, 1980).

The authors of DSM-III claim that, based on kappa statistics,

DSM-III has improved reliability over its predecessors. In

fact, a good number of diagnostic classes produced kappa

statistics that were at or above the 0.7 level. However,

DSM-III does make clear that "a high kappa. indicates

good agreement as to whether or not the patient has a

disorder within that diagnostic class, even if there is

disagreement about the specific disorder within the class"

(APA, 1980, p. 468). In other words, clinicians need only

agree about the general class of disorder a client has, not

the specific disorder itself. As explained in DSM-III:

diagnoses of Schizophrenia, Paranoid Type and
Schizophrenia, Catatonic Type by two clinicians would be
considered agreement on Schizophrenia. Similarly, if
one clinician diagnoses Borderline Personality Disorder
and the other Schizotypal Personality Disorder, this is
considered agreement on whether or not there is a
Personality Disorder. (APA, 1980, p. 468)

Thus, the DSM-III field trials assessed diagnostic

reliability according to class of disorder, not according to

specific disorder type. In order to be considered in

diagnostic agreement with each other, clinicians merely had

to concur about the class of disorder a client had, not about

the specific disorder within that class.

The authors of DSM-III conclude that the field trials

point towards enhanced diagnostic reliability for DSM-III

categories. As indicative of the progress made in enhancing

reliability during the course of the field trials, the

authors of DSM-III point to the fact that Phase Two

reliability was, in general, higher than Phase One

reliability. Nevertheless, the authors acknowledge several

areas in which diagnostic reliability could be enhanced. For

example, they express disappointment that the reliability

kappas for childhood and adolescent disorders were lower in

Phase Two than in Phase One--particularly so for the

adjustment disorder category (APA, 1980). Further, they

plainly acknowledge that diagnostic reliability for specific

personality disorders is not very good, with kappas ranging

from 0.26 to 0.75 (APA, 1980). In spite of these

disappointments, they conclude that

several innovative features of DSM-III have undoubtedly
contributed to the generally improved diagnostic
reliability: changes in the classification itself (e.g.
grouping all the Affective Disorders together), the
separation of Axis I and Axis II, the systematic
description of the various disorders and, finally, the
inclusion of diagnostic criteria. (APA, 1980, p. 469)

Some aspects of DSM-III and DSM-III-R reliability have not

been closely examined. Williams (1985b) was unable to locate

any reliability studies evaluating DSM-III's Axis III, and no

Axis III reliability studies were found during the writing of

this literature review. Nevertheless, a significant number

of studies on DSM-III and DSM-III-R reliability have been

conducted, many pertaining to specific disorders. For

example, numerous researchers have concerned themselves with

the reliability of diagnostic criteria for various affective

disorders (Carroll, 1984; Fogel, 1990; Mazure, Nelson, &

Price, 1986). While space prevents an exhaustive review of

reliability data for every disorder within DSM-III and DSM-

III-R, several important issues regarding the reliability of

DSM-III and DSM-III-R are briefly discussed below.

Axis IV. Skodol (1991) reviewed the literature

pertaining to Axis IV's reliability. Axis IV, which measures

the degree of psychosocial stress a patient is experiencing,

was severely criticized when DSM-III was first published. As

a result, DSM-III-R altered Axis IV by providing separate

examples to help clinicians differentially evaluate clients

experiencing acute events and clients experiencing enduring

events (Skodol, 1991). Little research into Axis IV's

reliability has been done; of these few studies, the DSM-III

field trials produced the best reliability results; these

results indicated a moderate degree of reliability. Even so,

Skodol (1991) notes that nearly half of the clinicians

participating in the DSM-III field trial recommended making

major changes in Axis IV. As a result of problems with Axis

IV, a number of changes were considered for DSM-IV, including

possibly eliminating Axis IV altogether in favor of a list of

psychological problems (Skodol, 1991).

Axis V. Axis V changed substantially between DSM-III

and DSM-III-R. Perhaps this is because research on DSM-III's

Axis V yielded mixed to poor reliability levels. For

example, Fernando, Mellsop, Nelson, Peace, & Wilson (1986)

conducted a study that investigated Axis V's reliability in

DSM-III. In DSM-III, Axis V measured the highest level of

adaptive functioning during the past year, and ratings were

made along a six point scale. The results of Fernando et al.

(1986) produced an intraclass correlation coefficient of only

.49, which was substantially lower than the correlation found

during the DSM-III field trials. By the time DSM-III-R was

published Axis V had been changed entirely, with the new Axis

V constituting the Global Assessment of Functioning Scale

(discussed earlier). No reliability information for the

revised version of Axis V is provided in DSM-III-R.

Reliability and categorical diagnosis. Adamson (1989)

contends that while diagnostic criteria served as an

improvement, DSM-III and DSM-III-R continue to exhibit poor

reliability and validity. He believes that limitations in

current knowledge prevent the development of a fully adequate

diagnostic system, but also implies that the categorical

approach used in the DSMs may be interfering with the ability

of researchers to improve reliability and validity. Adamson

(1989) argues that a more dimensional approach towards

measuring mental disorders might succeed better than the

present, categorical approach, but he doubts that DSM-IV will

move towards greater use of dimensional measurement.

However, Adamson (1989) feels that, despite its problems,

each revision of DSM marks an important step forward.

Dumont (1984) also criticizes the categorical approach,

but is far less confident than Adamson (1989) about the

ability to improve DSM diagnosis. Dumont (1984) believes

that DSM-III marked only a minimal improvement in

reliability, much of which was "inflated by the use of

'characteristic' case histories which do not manifest the

much more characteristic ambiguities of real-life situations"

(p. 327). He argues that categorical diagnosis is bound to

be unreliable because it is assumes that "reality comes

packaged in well-bounded categories waiting to be discovered"

(Dumont, 1984, p. 333). While Adamson (1989) and Dumont

(1984) provide very different outlooks for the future

improvement of DSM diagnosis, they both present important

concerns regarding the difficulties involved in producing a

categorical diagnostic system with high levels of


Selling of DSM. In a recent book, Kirk and Kutchins

(1992) argue that the sense of greatly enhanced diagnostic

reliability that accompanied DSM-III, and later DSM-III-R,

did not occur because either of these manuals was

demonstrably more reliable than previous editions of DSM.

Rather, they contend that the sense of enhanced reliability

resulted from brilliant political maneuverings on behalf of a

small group of neo-Kraepelinian researchers intent on seeing

the medical model replace the psychodynamic influence present

in DSM-II. These researchers, led by DSM-III and DSM-III-R

Task Force chair Robert Spitzer, were instrumental in the

development of both DSM-III and DSM-III-R. According to Kirk

& Kutchins (1992), one way that Spitzer and his colleagues

advanced their argument that DSM-III was very reliable was by

introducing the kappa statistic into diagnostic reliability


Kirk and Kutchins (1992) describe how use of the kappa

statistic contributed to the perception of DSM-III as

reliable, even though they feel that kappa measures

themselves did not support such a claim. First, while

controversial, the kappa statistic was not understood by most

practitioners. This left reliability issues in the hands of

researchers and statisticians. As a result, most

practitioners were left out of the reliability debate (Kirk &

Kutchins, 1992). Second, the kappa statistic transformed

reliability from a conceptual problem, threatening the very

core of psychiatry and psychiatric diagnosis, "into a

technical problem to be solved by research specialists" (Kirk

& Kutchins, 1992, p. 44). Transforming reliability into a

technical problem gave Spitzer and his colleagues a great

deal of power when it came to interpreting the results of

reliability studies used in developing DSM-III (Kirk &

Kutchins, 1992).

Finally--and perhaps most importantly--since no

standards for interpreting the kappa statistic were ever

established, issues of diagnostic reliability were subject to

widely disparate interpretations, depending on the particular

author's goals. Kirk and Kutchins (1992) argue that the

researchers involved in the development of DSM-III

interpreted kappa statistics in ways that supported their

view of DSM-III as substantially more reliable than DSM-II,

interpretations that Kirk and Kutchins (1992) dispute. Kirk

and Kutchins (1992) contend that the DSM-III field trial

kappa scores purported to indicate enhanced reliability for

DSM-III categories were not that different from those Spitzer

& Fliess (1974) previously reported as indicating only

satisfactory reliability in DSM-II. In other words, Kirk and

Kutchins' (1992) central argument is that the DSM-III Task

Force interpreted kappa scores in a way which was favorable

for DSM-III; despite the acceptance of DSM-III and DSM-III-R

as reliable, Kirk and Kutchins (1992) still see reliability

as a problem in DSM-III, as well as in DSM-III-R. They claim

that the degree to which the DSM's diagnostic reliability was

taken for granted by both supporters and critics after DSM-

III is dramatically represented by the fact that DSM-III-R, a

substantial revision of DSM-III, was accepted as reliable

even though no field trials on interrater agreement were ever

conducted (Kirk & Kutchins, 1992).

Eysenck's critique. Eysenck et al. (1983) presented one

of the more thorough assessments of DSM-III's reliability and

validity. Like Adamson (1989), Eysenck et al. (1983) feel

that a dimensional approach is preferable to a categorical

one. They criticize DSM-III for being highly unreliable, but

contend that all forms of diagnosis, even those in more

traditional areas of medicine, have low levels of

reliability. Eysenck et al. (1983) argue that DSM-III's

biggest problem is not reliability, but validity. They

question focusing primarily on reliability because "it is the

absence of any indication of validity which is far more

critical, as is the absence of any appreciation of the

importance of scientific proof for schemes of this kind"

(Eysenck et al., 1983, p. 189). Eysenck et al. (1983) feel

that psychiatry's aspirations have expanded beyond its

scientific abilities. As a result, both validity and

reliability have suffered. Eysenck et al. (1983) go so far

as to state that "DSM-III is primarily a professional manual

and only secondarily, if at all, a scientific manual" (p.

184). In their view, DSM-III is neither reliable nor valid.

Reliable but not valid. Wakefield (1992b) further

develops the idea that DSM-III and DSM-III-R suffer from

validity problems. He argues that DSM-III-R's definition of

mental disorders does not effectively distinguish disorders

from non-disorders. For example, Wakefield (1992b) contends

that DSM-III-R holds that "a mental disorder is a mental

condition that (a) causes distress or disability and (b) is

not a statistically expectable response to external events"

(p. 238). According to Wakefield (1992b), this definition is

simultaneously too broad and too narrow. He observes that it

is too broad when considering DSM-III-R's V codes. The V

codes, which DSM-III-R indicates are conditions not

attributable to mental disorder but which are treated by

clinicians, fit the definition of mental disorder but are not

considered disorders; using this example, Wakefield observes

that the definition of mental disorder is often too

inclusive, applying to behaviors most clinicians do not

perceive to be disorders. Likewise, Wakefield (1992b) points

out that many behaviors classified as disorders are

expectable (e.g. post traumatic stress disorder, anaclitic

depression), even though DSM-III-R's definition implies that

disordered behavior is statistically unexpectable. While

Wakefield (1992b) believes that post traumatic stress

disorder constitutes dysfunctional behavior and is therefore

legitimately classified as a disorder, he contends that DSM-

III-R's definition of mental disorder implies that post

traumatic stress is not a disorder because it involves

behavior that, following a traumatic incident, can often be

expected. In this case, Wakefield (1992b) sees DSM-III-R's

definition of mental disorder as too narrow.

Wakefield (1992b) believes that the reason that DSM-III-

R's definition of mental disorder lacks validity is because

so much attention has been focused on diagnostic reliability.

He asserts that enhanced reliability can actually decrease

validity because "one need only operationalize criteria in

ways that do not correspond to what one is trying to measure"

(Wakefield, 1992, p. 241). Wakefield (1992b) feels that DSM-

III-R sacrifices validity through its strategy for enhancing

reliability. This strategy "attempts to increase reliability

by eliminating the inference to internal mechanism failure

and relying for diagnostic judgments on the unexpectable

nature of the harms alone" (p. 242). Wakefield (1992a,

1992b) defines a dysfunction as the failure of an internal

mechanism to function as it is supposed to. He

differentiates the term "dysfunction" from the term

"harmful," noting that the former is the failure of an

internal mechanism to operate appropriately and the latter is

derived from social norms (Wakefield, 1992a). He proposes

that in order to improve diagnostic validity, DSM-III-R

criteria need to be altered so that they better isolate

disorders within the person, rather than unexpectable

behaviors that are "due to environmental stresses, lack of

education, or other specific nondysfunctional causes"

(Wakefield, 1992b, p. 243). Wakefield (1992a, 1992b)

believes that DSM-III-R can improve its validity if its

authors emphasize internal mechanism dysfunctions.

Validity and cynicism. Brown (1987) conducted a study

that looked at the ways in which DSM-III was used in a

community mental health center. Often, staff felt the

diagnoses assigned to clients were insufficient in describing

client problems, and therefore not particularly valid as

diagnostic constructs. For example, when assessing one

particular client, a resident and his supervisors found the

diagnoses available inadequate. They saw the client's

behavior "as the central concern, and diagnosis was exactly

what the resident believed it to be: the necessity of

'putting some DSM thing on paper'" (Brown, 1987, p. 39).

Labels were used more for bureaucratic reasons than

anything else. In fact, clinicians generally regarded DSM-

III labels "as imprecise and/or not very helpful" (Brown,

1987, p. 40). Brown (1987) concludes that DSM-III's claims

of validity are exaggerated, given the limitations it puts on

clinicians and the negative attitudes many clinicians have

towards it. He states that "the inaccuracy of diagnosis, in

its restricted DSM-III sense. calls into question the

validity of diagnosis, therefore producing and heightening

diagnostic uncertainty, ambivalence, and conflict" (Brown,

1987, p. 49). Of course, whether or not Brown's findings are

due to cynicism on the part of community mental health

workers or genuine problems with DSM-III as a diagnostic tool

cannot be easily determined from this particular study.

Transitive Diagnosis

Kelly (1991b) outlined a new approach for understanding

client problems. This approach, which he called transitive

diagnosis, emphasizes understanding the construct systems

used by clients in order to best comprehend their presenting

problems. Kelly tried to differentiate transitive diagnosis

from more traditional, Kraepelinian styles of diagnosis. For

example, he made clear that transitive diagnosis emphasizes

the process of client change, rather than static categories;

one of Kelly's biggest criticisms of traditional diagnostic

approaches is their emphasis on static labels. In transitive

diagnosis, assessment is continuous, because the client's

construct system is always evolving. For Kelly, categorical

diagnosis has a hard time emphasizing process because it

results in a specific diagnosis which then is maintained

throughout treatment. Kelly's transitive diagnosis involves

a continual reevaluation of client constructs. Below,

transitive diagnosis is described as Kelly presented it.

Then, Bannister's constructivist critique of traditional

diagnosis is briefly outlined. Finally, Faidley and

Leitner's (1993) more recent elaboration of transitive

diagnosis is discussed.

Kelly on Transitive Diagnosis

Kelly (1991b) described transitive diagnosis as

"concerned with transitions in the client's life" (p. 153).

He felt that traditional approaches to diagnosis "impose

preemptive constructions on human behavior. Diagnosis is all

too frequently an attempt to cram a whole live struggling

client into a nosological category" (p. 154). Kelly (1991b)

emphasized that people are always in the process of changing

and reconstruing events, and strongly objected to

"nosological diagnosis and all the forms of thinking which

distract our attention from the fact that life does go on and

on" (p. 154). Transitive diagnosis is Kelly's alternative

approach to understanding clients and their concerns. Unlike

DSM-III-R diagnosis--which usually is completed early in

therapy and used to determine treatment--transitive diagnosis

is a process which continues throughout therapy. Leitner (in

press) describes assessment in transitive diagnosis as

involving "psychological description of the client's

constructs, psychological evaluation of the client's

construct system, and an analysis of the milieu in which

adjustment is sought." Below, all of the elements that Kelly

felt were important in making a transitive diagnosis are

summarized. Transitive diagnosis is so different from

traditional notions of psychiatric diagnosis that Raskin and

Epting (1994) suggest that the term "transitive

understanding" (rather than transitive diagnosis) might

better describe what it entails.

Formulating client problems. The first step in

transitive diagnosis involves identifying the client's major

problem areas. Leitner (in press) is quick to point out that

this does not involve assigning the client to a particular

diagnostic category. Rather, it involves understanding a

client's constructions of events in order to understand where

these constructions are causing the client difficulties.

Kelly (1991b) pointed out that clinicians need to consider

their own norms when trying to understand client problems.

Recent research by Soldz (1989) further illustrates the need

for therapists to be aware of their own constructions when

evaluating clients. Soldz's (1989) dissertation results

indicate that therapists employ different constructs when

describing clients than they do when describing

acquaintances; in describing clients, constructs relating to

emotional stability are employed more often.

In addition to awareness of one's own constructs,

awareness of a client's cultural context, personal-social

context, job patterns, and family situation are all important

in understanding his or her presenting problem (Kelly,

1991b). Finally, considering the gains and losses that the

client experiences as a result of his or her presenting

problem is quite important (Kelly, 1991b). At this stage of

diagnosis, evaluation of the losses and gains that client

behavior produces ought to be tentative. Kelly (1991b)

observed that

in transitive diagnosis the clinician seeks a
preliminary overview of these gains and losses in
relation to the client's symptoms. He follows this up
in a later stage with a more intimate appraisal of these
gains and losses in the light of the client's personal
constructs. (Kelly, 1991b, p. 168-169)

Describing client constructs. This stage of the

diagnostic process has three main components. The first

involves understanding clients' own constructions of their

problems. While Kelly (1991b) observed that a client's

construction of the problem may not always be the most

rewarding construction available, it is very important for

the clinician to respect and understand the client's

construction. The second task of this stage involves

learning what clients feel others think the problem is.

Finally, clinicians need to evaluate what clients construe

their own life roles to be. Kelly (1991b) notes that "the

clinician should keep in mind that his client is bound only

by the consistent interpretations he places upon his

experience" (p. 170). Realizing that client roles are played

out in relation to expectations of others can be useful in

coming better to understand clients and their problems.

Evaluating the construct system. In evaluating client

construct systems, transitive diagnosis focuses on

aggression, anxiety, and constriction (Kelly, 1991b).

Aggression occurs when people actively test their constructs

in an attempt to confirm them; often, clients experience

hostility, wherein they try to force events to fit with their

constructs rather than acknowledging that their constructs do

a poor job accounting for events (Kelly, 1991b). In

transitive diagnosis, awareness of areas of hostility and

aggression can help clinicians identify areas of the

construct system requiring reevaluation (Kelly, 1991b).

Anxiety results when one's construct system does not

accommodate events; if the clinicians can identify areas of

anxiety, they can help clients focus on developing ways to

elaborate their systems so as to reduce anxiety (Kelly,

1991b). Finally, constriction involves limiting one's

perceptual field in order to avoid impending disconfirmation

of one's constructs (Kelly, 1991b). Awareness of ways in

which clients constrict their awareness in order protect

particular constructions can be quite critical in helping

clients during therapy.

Treatment. DSM-III-R diagnosis involves evaluating

clients and determining what disorders they suffer from;

treatment then proceeds based on the category of disorder a

client has. As alluded to earlier, transitive diagnosis

takes a somewhat different approach. Kelly (1991b)

considered transitive diagnosis to be an ongoing process; the

therapist should always be evaluating and reevaluating client

constructions. This is important in the treatment phase of

therapy. Kelly felt that clinicians need to approach each

client individually. Treatment plans should be created with

the particular client in mind. By understanding client

constructions, developing an individualized treatment plan is

made easier. As new material is uncovered, clinicians are

expected to revise and reevaluate their own constructions of

the client. Further, as the client changes, clinicians need

to reassess client constructs. Because Kelly felt that the

therapy process involves helping clients reconstrue events,

therapists need to constantly attend to changes within a

client's construct system. In is in this regard that

transitive diagnosis does not stop after the initial

assessment phase of therapy; it continues throughout the

course of counseling.

Eliciting constructs. An important part of transitive

diagnosis involves eliciting specific construct dimensions

from the client. A construct dimension usually consists of

an adjective or brief phrase (for example "good," "happy,"

"always in a rush," and "tough as nails") and its perceived

opposite (for example, "bad," "working," "lazy," and "wishy-

washy"). Construct dimesnions are idiosyncratic to the

persons using them; the opposite of "good" for one person may

be "bad," but for another person it may be something quite

different--such as "sad" or "irresponsible."

Kelly (1991b) introduced the Role Construct Repertory

Test (or Rep Test) as a scientific way to elicit personal

constructs. Various forms of the Rep Test have been

developed, but the basic premise behind each version is the

same. Client's are asked to list important figures in their

lives. Then subjects are asked to take three of these

figures at a time and indicate how two of them are the same

and different from the third. In so doing, clients produce

construct dimensions consisting of bipolar opposites (Kelly,

1991b). This can easily be done with clients, resulting in a

set of specific construct dimensions that can be evaluated in

a number of ways. Of course, other methods of construct

elicitation have been developed. Leitner (in press)

catalogues construct elicitation techniques, and provides an

excellent overview of their advantages and disadvantages.

The pyramid procedure, self-characterization sketch,

laddering techniques, and interviewing methodologies are all

described in detail (Leitner, in press). Any or all of these

various techniques for eliciting constructs can be used as

part of transitive diagnosis.

Recent Developments

Don Bannister did a significant amount of early work on

diagnosis from a personal construct theory perspective (Agnew

and Bannister, 1973; Bannister, 1968; 1985; Bannister,

Salmon, and Leiberman, 1964). However, most of Bannister's

writings use PCT as a means for critiquing traditional

psychiatric diagnosis. For example, Bannister, Salmon, and

Lieberman (1964) tried to demonstrate empirically that

nosological diagnosis fails to dictate treatment choices.

They studied 1000 psychiatric hospital patients, for whom

they recorded the first diagnosis and the first treatment.

After analyzing their results, Bannister et al. (1964)

tentatively concluded that the study's "findings were not

consistent with the notion that each particular diagnosis

leads logically (or habitually) to a particular treatment"

(p. 731).

A later study by Agnew and Bannister (1973) employed the

Rep Test to investigate the usefulness of traditional

diagnosis. Eight psychiatric consultants were used as

subjects. Each chose twenty patients they knew well; these

patients were divided into two groups of ten (group A and

group B). Agnew and Bannister (1973) asked the consultants

to rank order patients in group A along eight psychological

constructs (generous, obstinate, considerate, reserved,

unreliable, likeable, mature, submissive) and to rank order

patients in group B along eight psychiatric diagnoses

(neurotic depression, personality disorder, schizophrenia,

anxiety state, psychotic depression, hysteria, brain damage,

and obsessional neurosis). One month later, the consultants

repeated this procedure, only this time they rank ordered the

patients in group A along the psychiatric diagnoses and the

patients in group B along the construct dimensions. They

rated the diagnoses and constructs according to interjudge

agreement, stability over time, intercorrelations between the

constructs and diagnoses, and patterns of interrelationships

between terms. Agnew and Bannister concluded that their

"results indicate that the psychiatrists are no more stable

and have no greater interjudge agreement in using diagnostic

terms than they achieve with everyday language" (p. 73). In

their view "psychiatric nosology is not a true specialist

language" (p. 73).

Bannister (1985) incorporated the sociality corollary

into a study that examined how the degree to which

psychiatrists take into account the patient's point of view

impacts on the patient's recovery. The sociality corollary

holds that "to the extent that one person construes the

construction processes of another, he may play a role in a

social process involving the other person" (Kelly, 1991a, p.

66). In other words, the roles people play in relation to

one other are influenced by the expectations that people have

about one another. Bannister (1985) found that when the

hospital agreed with patients about the cause of patient

problems, both the patient and hospital tended to agree about

the appropriate treatment. Type of treatment was not related

to positive or negative patient outcome, but agreement about

cause of the problem was. Bannister (1985) also found that

hospitals generally attributed patient problems to internal

causes. Consistent with the above mentioned finding, when

patients attributed their problems to internal causes the

outcome of their hospitalization was more likely to be

positive. Bannister (1985) concludes that it is very

important for mental health professionals to keep in mind the

patient's explanation of his or her problems because

understanding these problems from the patient's perspective

appears to greatly impact on treatment effectiveness.

Keeping with Bannister's findings, transitive diagnosis

emphasizes understanding the patient's constructions of the


Bannister's work critiqued traditional forms of

diagnosis, but little research or elaboration has been done

on Kelly's original formulation of transitive diagnosis. A

number of articles have called for personal construct

therapists to develop and use transitive diagnosis in

evaluating clients (Epting & Leitner, 1992; Raskin & Epting,

1993). In fact, Faidley and Leitner (1993) have recently

made an effort to expand transitive diagnosis beyond Kelly's

original formulations. Faidley and Leitner (1993) critique

DSM-III-R, arguing that it is often inconsistent with

particular theoretical orientations. Further, they argue

that thinking about client behaviors in terms of their

personal meanings is more fruitful than thinking about client

behavior in terms of observable symptoms and concrete

categories. Focusing on client constructions, client

behaviors, process variables, and the relationship between

client and therapist are all part of Faidley and Leitner's

(1993) approach. Important elements of Landfield and Epting

(1987) and Faidley and Leitner's (1993) work which add to

Kelly's (1991b) original formulation of transitive diagnosis

are briefly summarized below.

A transitive diagnosis framework. Kelly (1991a)

cautioned against placing too great an emphasis on

reliability; a highly reliable instrument may simply be one

that is insensitive to change (Landfield & Epting, 1987). In

other words "the construct scientist seeks the primary

consistency of his tests at the level of whole dimensions of

personal meaning rather than in how persons specifically

apply their dimensions to self and others at a particular

moment in time" (Landfield & Epting, 1987, p. 92). Faidley

and Leitner (1993) echo this sentiment by moving away from a

primary focus on diagnostic reliability and discussing what

they see as the critical components of transitive diagnosis.

They believe that in order to be valid, transitive diagnosis

should provide useful and rich information, be susceptible to

change, emphasize process over category, and accurately

reflect the client's meanings. Again, in order to meet these

demands, assessment becomes a continual process; it does not

end with the assignment of a disorder.

Construct interpretation. Landfield and Epting (1987)

posited twenty construct interpretation guidelines. Faidley

and Leitner (1993) incorporate fifteen of these guidelines

into the process of transitive diagnosis. Several of these

guidelines warrant mention here in order to give the reader a

feel for how constructs can be interpreted and evaluated in

transitive diagnosis.

Looking at a client's construct dimensions as

contrasting part of his or her personality is one way to

interpret constructs. Faidley and Leitner (1993) observe

that "contrasts that appear incredulous to the clinician may,

in fact, be opposites for the client" (p. 61). As an example

of this, they discuss a man who possessed the construct

dimension of "depressed vs. irresponsible." Most people

would not see "irresponsible" as the opposite of "depressed."

However, in this man's construction of reality, such a

dimension was very powerful; when he was promoted at work,

the perceived responsibility led him to attempt suicide

(Faidley and Leitner, 1993). Landfield and Epting (1987) do

a nice job presenting unique construct dimensions that people

hold; in so doing, Landfield and Epting (1987) clarify the

idiographic nature of personal constructions.

Constructs can also be used to hypothesize about how

clients will behave in the future. For example, Faidley and

Leitner (1993) mention a client who holds the construct

"passive/murderous." The dominant pole of the construct may

be "passive," but the therapist can use the "murderous" pole

as a barometer of what to expect if the client shifts away

from the "passive" pole; this construct dimension alerts the

therapist "to the possibility of violent, even murderous,

behavior should the client become less 'passive'" (Faidley &

Leitner, 1993, p. 62).

Observing how often a client overuses a construct can

provide therapeutic insight, as can attention to how often

client's describe social interactions (Landfield & Epting,

1987; Faidley & Leitner, 1993). In addition, therapists

should focus on how open constructs are to accommodating new

experiences; people who lack constructs of change are likely

to be less amenable to change. Identifying constructs that

deal with death, destruction, depression, and maladjustment

can also be useful to therapists, as can identifying

constructs that indicate alienation--such as "don't care,"

"uninterested in life," and "self-rejecting" (Landfield &

Epting, 1987; Faidley & Leitner, 1993). Attention to

descriptive modifiers that clients use is also important.

Constructs modified by words like "always," "never,"
"very," "absolutely," or "perfectly" may be ones that
represent more significant and core understandings. An
excessive use of extreme modifiers may signal a
constricted approach to relating or problem solving.
(Faidley & Leitner, 1993, p. 66)

Likewise, overuse of modifiers such as "sometimes," "a bit,"

and "somewhat" may suggest decision-making difficulties

(Faidley & Leitner, 1993).

If a client only supplies one pole of construct

dimension, it may indicate that he or she is trying to avoid

something; in such cases, the therapist can interpret the

client's behavior as indicating some kind of threat.

Finally, if therapists try to imagine client constructs as

their own, they are often able to better comprehend their

clients. Faidley and Leitner (1993) observe that looking at

the world through the client's constructions "may be the most

important for the therapist interested in understanding the

client's struggles" (p. 67).

Case studies. Faidley and Leitner (1993) devote four

chapters to actual case studies that employ transitive

diagnosis. Each chapter presents a series of personal

constructs produced by a particular client, and then presents

a number of clinical hypotheses regarding those constructs.

Further, interpersonal process analyses (wherein the

therapist assesses the client's body language, manner of

speaking, and other methods of communication) are undertaken

for each client, and discussed in relationship to the

constructs elicited. Finally, the fates of the various

therapeutic hypotheses are presented; as therapy progressed,

Faidley and Leitner (1993) explain the ways in which these

hypotheses were confirmed, disconfirmed, dismissed, or

altered. By presenting these case studies, Faidley and

Leitner (1993) provide concrete examples of the process of

transitive diagnosis.

Common Criticisms of Psychiatric Diagnosis

Issues of reliability and validity remain central to

many of the debates over psychiatric diagnosis. However,

many of the researchers involved in the debate over

reliability and validity--even those who feel DSM-III and

DSM-III-R are lacking in this regard--share an underlying

belief in the value of diagnosis. Nevertheless, numerous

critics have attacked what they see as the underlying

theoretical flaws of psychiatric diagnosis. Because DSM has

been the dominant diagnostic scheme during the past forty

years, many of the criticisms of psychiatric diagnosis have

also involved criticisms of DSM (for an interesting debate on

the advantages and disadvantages of DSM-III, see Klerman,

1984; Vaillant, 1984; Spitzer, 1984a; 1984b; and Michels,

1984a, 1984b). Even so, common concerns about psychological

assessment are relevant for both DSM and transitive diagnosis

approaches. Therefore, some of the more common controversies

involving psychiatric diagnosis are presented below; specific

DSM-III or DSM-III-R research is incorporated into the

discussion where it is relevant in explicating these


Difficulty Identifying the Mentally Ill

Rosenhan (1973) wondered whether or not staff members

could distinguish pseudopatients from other, presumably real,

patients. He had eight pseudopatients--three females, five

males--present themselves for admission at twelve different

mental hospitals. During admission interviews,

pseudopatients complained about hearing voices that said

"empty," hollow," and "thud" (Rosenhan, 1973). None of the

pseudopatients were detected as imposters. All were admitted

to the hospital, and all but one were diagnosed with

schizophrenia. The one pseudopatient not diagnosed with

schizophrenia was diagnosed with manic-depressive psychosis

(Rosenhan, 1973).

Once hospitalized, pseudopatients "ceased simulating any

symptom of abnormality" (Rosenhan, 1973, p. 251). They

simply behaved normally. However, they were still not

detected by staff members as imposters, even though none of

them had any previous history of psychopathology.

Interestingly, the only people who seemed to realize that the

pseudopatients were imposters were the actual patients.

Pseudopatients were hospitalized "from 7 to 52 days, with an

average of 19 days" (Rosenhan, 1973, p. 252). Thus,

pseudopatients did not go undetected because staff lacked

time to observe them. Even upon release, none of the

pseudopatients were detected as imposters. All eleven of the

pseudopatients admitted with a diagnosis of schizophrenia

were released with a diagnosis of schizophrenia "in

remission" (Rosenhan, 1973).

Rosenhan (1973) offers a potential explanation for why

the staff were unable to distinguish the pseudopatients from

real patients, suggesting that the staff's failure was due to

a bias towards Type II errors, wherein staff felt it was

safer to "err on the side of caution, to suspect illness even

among the healthy" (p. 252). However, a second study by

Rosenhan disconfirms this explanation. Rosenhan (1973)

informed the staff at a research and teaching hospital that

he would send one or more pseudopatients to the hospital over

the course of the next three months. During that time,

hospital staff members rated all patients presenting

themselves for admission on a 10 point scale. This scale

assessed the confidence of staff members that a patient was a

pseudopatient. Rosenhan's (1973) results are quite


Forty-one patients were alleged, with high confidence,
to be pseudopatients by at least one member of the
staff. Twenty-three were considered suspect by at least
one psychiatrist. Nineteen were suspected by one
psychiatrist and one other staff member. Actually, no
genuine pseudopatient .presented himself during
this period. (p. 252)

The results of Rosenhan's (1973) second study indicate that

Type II errors were not the primary cause of staff being

unable to distinguish real patients from pseudopatients.

Staff members were just as easily influenced towards

diagnosing real patients as pseudopatients as they were

towards diagnosing pseudopatients as real patients.

Essentially, hospital staff members were unable, in either of

Rosenhan's studies, effectively to distinguish the mentally

ill from "normal" people. While Rosenhan's (1973) results do

not necessarily generalize to DSM-III-R or transitive

diagnosis, and alternative explanations for Rosenhan's (1973)

results are possible, his study is important because it makes

clear the need to have a diagnostic system that distinguishes

disordered from normal people.

Mental illness As Myth

One of the most vociferous critics of mental illness and

psychiatric diagnosis during the past three decades has been

Thomas Szasz (1960, 1963, 1974, 1987, 1991). Szasz attacks

medical model approaches to abnormal behavior, arguing that

"mental illness" is a metaphorical term, since the mind

cannot become physically sick. According to Szasz, an

illness requires a corresponding biological dysfunction. The

term "mental" does not correspond to any physical structure

of the body. Thus, Szasz (1974, 1987) argues that one cannot

actually have a mental illness; that is, mental illness, in a

literal sense, does not exist. In Szasz's (1974) words:

"'Mental illness' is a metaphor. Minds can be 'sick' only in

the sense that jokes are 'sick' or economies are 'sick'" (p.


Szasz maintains that people are either physically sick

or not. In many cases, the etiology of particular categories

of mental illness is not known. Szasz argues that many

mental illnesses of unknown etiology are probably not

illnesses at all in that they are not biologically caused.

Rather, they constitute "problems in living." It is

irresponsible, in Szasz' (1974, 1987) view, to insist that

misunderstood and socially undesirable behaviors are known

mental illnesses, when these behaviors are not necessarily

illnesses at all. While Szasz is perhaps the best known

critic of the medical model, others have expressed

dissatisfaction with applying the medical model in a

psychological context (Breggin, 1991; Laing, 1965, 1967).

Szasz (1974, 1987) is concerned with psychiatric

diagnosis based upon the medical model. Because Szasz feels

that mental illness is a metaphorical term, he argues that a

medical model approach in diagnosing deviance is

inappropriate. For Szasz, most abnormal behavior cannot be

successfully approached from the standpoint of the medical

model. He does not deny that people often exhibit behavior

patterns which deviate from societal norms, are difficult to

understand, and are sometimes a source of unhappiness for

those engaging in them. Instead, he argues that the

categories used to describe these behavior patterns, though

called illnesses, are often merely descriptions (Szasz, 1974,

1987). Szasz makes this clear in his critique of DSM

diagnosis, which he sees as theoretically tied to the medical

model of abnormality.

According to Szasz (1960, 1974, 1987), the problem is

not simply the creation of diagnostic categories; the problem

is that these metaphorical illness labels have ceased to be

viewed as descriptors and have come to be viewed as

biological disease entities. Furthermore, Szasz (1987)

maintains that DSM, in all of its versions, is not truly a

scientific enterprise. DSM categories do not describe

scientifically valid diseases; they are categories devised,

and continually revised, by committees. Controversial

diagnoses--such as homosexuality, paraphilic rapism,

premenstrual syndrome, and self-defeating personality

disorder--are included or excluded from each successive

version of DSM more for political than scientific reasons

(Szasz, 1987; for an excellent chronicling of the complicated

political battles surrounding some of these diagnoses, see

Kirk & Kutchins, 1992). Szasz (1987) angrily complains that

instead of a scientifically legitimate diagnostic system, in

DSM "we have psychiatric democracy" (p. 80), where diagnostic

categories are based more on political consensus than

scientific discovery. Robert Spitzer, head of the DSM-III

and DSM-III-R task forces, echoes Szasz's complaint when he

laments that expert consensus, rather than empirical data,

will be used to make final decisions about DSM-IV (Spitzer,


Labeling Theory and Mental Illness

Concern over the stigma of mental illness is not new.

Goffman (1961, 1963) looked at stigma in general, and

specifically at the way mental patients become stigmatized

during the process of hospitalization. His analysis

addressed the impact "total institutions" (such as mental

hospitals) have on inmates, and was quite influential in the

deinstitutionalization movement. Addressing diagnosis more

specifically, Thomas Scheff (1966) developed labeling theory-

-which argues that being labeled as mentally ill has powerful

consequences. Scheff speaks of "residual deviance," which

involves deviant behaviors for which society lacks a

specific, legally sanctioned response. Most of the time,

residual deviance is not addressed by society. When society

does respond to residually deviant behavior, labeling of the

deviant person occurs. For Scheff (1966), deviant persons

are often labeled as mentally ill and given psychiatric

diagnoses. These diagnoses influence the way others perceive

and respond to labeled persons. As a result, societal forces

impact on labeled individuals, causing them to alter their

self-perceptions and behaviors. Ultimately, labeled

individuals come to behave in ways that are consistent with

the diagnoses they are assigned. This has potentially

important consequences for categorical diagnosis which, if

labeling theory is correct, results in stigmatization for

those diagnosed.

Research on Scheff's theory has produced mixed results.

For comprehensive reviews of labeling theory research, see

Gove (1980) and Link and Cullen (1990). Here, discussion of

a few studies will have to suffice. One early study used to

support labeling theory was performed by Phillips (1963), who

found that seeking psychiatric help for mental illness

results in rejection by others. Farina, Gliha, Boudreau,

Allen, and Sherman's (1971) research indicated that mental

patients are affected by other people being aware of their

status as mental patients. In the research of Farina et al.

(1971), mental patients who were informed that a confederate

knew of their mental illness status had a significantly

harder time performing a task than control subjects.

Further, observers rated subjects who believed others knew of

their mental illness status as more tense and less well

adjusted than they did control subjects. The findings of

Farina et al. (1971) lend support to the idea that

psychiatric labels influence the behavior of those labeled.

More recently, Socall and Holtgraves (1992) found that

people were less willing to interact with a person when the

person was labeled as mentally ill than when that same person

was labeled as physically ill. Gallop, Lancee, and Garfinkel

(1989) found that nurses respond to hypothetical statements

about borderline personality disorder patients differently

than they do to schizophrenic patients; nurses react in a

more emotionally involved way in response to schizophrenic

patients, but tend to be demeaning and contradictory in

response to borderline patients. This suggests that

different disorders can result in different labeling effects.

Not all research has fully supported labeling theory.

For example, studies by Chassin, Presson, Young, and Light

(1981) and Holman and Caston (1987) provide only slim to

moderate support for labeling theory. Warner, Taylor,

Powers, and Hyman (1989) conducted a study that disconfirmed

one of labeling theory's most central predictions; their

study's results showed that acceptance of a mental illness

label does not produce lower levels of functioning. In other

words, subjects in the study by Warner et al. (1989) who

accepted their diagnoses did not exhibit more deviant

behavior than subjects who rejected their diagnoses. This

contradicts Scheff's (1966) idea that a person assigned a

mental illness label will come to behave in accordance with

the label. However, the results of Warner et al. (1989) did

confirm certain aspects of labeling theory. It was found

that acceptance of a mental illness diagnosis results in

lower self-esteem when high levels of stigma are associated

with it. Further, Warner et al. found that acceptance of a

mental illness label is positively correlated with an

attribution of external locus of control. Thus, the study by

Warner et al. (1989) provides mixed support for labeling


Rabkin (1984) contends that while stigma regarding

mental illness should not be overlooked, labeling approaches

overestimate the degree to which people are rejected based on

an identification as mentally ill. She discusses a field

study in Toronto, a telephone survey in New York City, and a

survey in Oklahoma that all examined the degree to which a

mental illness label results in stigma. The three studies

found that people often do not know that they live near

psychiatric facilities, but are not upset about it when they

do know. Rabkin (1984) concludes that deviant behavior is

more important in rejection of the mentally ill than are

labels; further, she sees societal attitudes towards the

mentally ill changing in a positive direction. Brockington,

Hall, Levings, and Murphy (1993) found that certain factors

were correlated with more positive responses to the mentally

ill. In their study, association with a mentally ill person

enhanced tolerance of the mentally ill. Brockington et al.

(1993) also found being young, having more education, and

having a higher level managerial job were associated with

greater tolerance. In addition, they found that people in a

town served by a mental hospital were slightly less tolerant

of the mentally ill than people in a town served by a

community-based mental health service.

Gove (1980) provides a sophisticated critique of

Scheff's approach, one which questions the basic premises

behind labeling theory. He states that the labeling

perspective fails to see "the deviant as someone who is

suffering from an interpersonal disorder, but instead as

someone who, through a set of circumstances, becomes publicly

labeled a deviant and who is forced by societal reaction into

a deviant role" (p. 85). Gove (1980) rejects this idea. He

cites evidence showing that, of those identified as mentally

ill, "a substantial majority have a serious disorder quite

apart from any secondary deviance" (p. 85). To Gove (1980),

deviant behavior, rather than a deviant label, is what causes

people to respond negatively. He criticizes labeling theory

research for failing to distinguish what it is that makes

people respond negatively to labeled persons. Link and

Cullen (1990), in their review of labeling theory research,

cite a number of studies which indicate that deviant

behavior, not labeling, is primarily responsible for

rejection of the mentally ill. In addition, Link, Cullen,

Frank, and Wozniak (1987) present a table summarizing twelve

previous studies that investigated whether labeling or

deviant behavior was more important in influencing

perceptions of mental patients. Ten of the studies found

deviant behavior to be more influential than labels (Link et

al., 1987). Such findings support Gove's (1980) criticism of

labeling theory. However, a recent study by Burk and Sher

(1990) found that both mental health professionals and peers

rated persons labeled as children of alcoholics more

negatively than those not labeled as such, regardless of

whether the labeled person was described as socially

successful or having behavior problems.

In response to Gove's (1980) criticisms, Bruce Link and

his colleagues have developed a revised labeling theory,

which holds that socialization creates a set of beliefs about

mental patients, and that when someone becomes a mental

patient these beliefs become salient. In this revised

formulation, labeling has negative consequences when labeled

individuals believe that others will discriminate against

them based on their status as mental patients (Link, Cullen,

Struening, Shrout, and Dohrenwend, 1989). This revised

approach challenges Gove's (1980) assertion that behaviors,

not labels, are most critical in stigmatizing mental

patients. In a vignette study, Link et al. (1987) found an

interaction between beliefs about how dangerous the mentally

ill are and labeling. That is, they found that people who

perceive the mentally ill as dangerous are more likely to

respond negatively to a person with a mental illness label.

Further, Link et al. (1987) found that being perceived as

dangerous was actually more important in predicting negative

responses to a labeled person than behavior. This study

confirms a revised labeling approach's hypothesis that labels

matter, if they activate certain expectations about the

labeled person.

Other studies by Link and his colleagues have supported

their revised labeling theory formulations. For example,

Link (1989) found that labeled persons who believe that

mental illness diagnoses will lead to discrimination and

devaluing are more likely to report loss of income,

unemployment, and feelings of demoralization than nonlabeled

persons with the same beliefs. Again, this seems to indicate

that beliefs about the consequences of a label influence the

way such a label affects self-perceptions and behavior. Link

et al. (1989) found similar results; their research indicates

that both mental patients and untreated community residents

believe that mental patients are rejected by most people.

Patients supported using withdrawal, secrecy, and education

to cope with expectations of rejection (Link et al., 1989).

More ominously, Link, Mirotznik, and Cullen (1991) found that

patients who try to escape labeling effects by keeping their

patient status a secret, avoiding situations potentially

leading to rejection, and teaching others about their

disorders fail to avoid the negative consequences of

labeling. In fact, the results of Link et al. (1991) showed

that trying to avoid labeling effects was more harmful than


Other Labeling-Related Research

Other studies that have investigated the impact of

psychiatric diagnoses on perceptions of persons have yielded

interesting results. While related to the labeling theory

research, the studies presented below are not explicitly

rooted in a labeling theory approach but are still highly

relevant for determining the impact of diagnostic labels on

perceptions of self and others.

Mindfulness. Langer and Abelson's (1974) research is

most directly related to labeling theory research. Langer

(1989) distinguishes between mindfulness and mindlessness.

Mindlessness occurs when we rigidly adhere to the categories

we have created. Mindfulness involves incorporating new

information into what is already known, as well as the

acceptance of multiple views of the same event (Langer,

1989). Langer and Abelson (1974) investigated the ways in

which labels impact on clinicians. Two groups of clinicians

participated in the study. One group was trained according

to a psychoanalytic model, which included acceptance of

traditional approaches to psychiatric diagnosis and labeling;

the other group was trained according to a behavioral model,

which "typically includes severe skepticism about the utility

of diagnostic categories and labels" (Langer & Abelson, 1974,

p. 5). All clinicians watched a fifteen minute videotape of

a young man discussing his past work experiences. When

labeled a "job applicant," clinicians generally rated the

young man as fairly well adjusted, theoretical orientation

notwithstanding. However, when labeled a "patient," analytic

clinicians rated the young man as significantly more

disturbed than when he was labeled a "job applicant." By

comparison, behavioral clinicians rated the young man as

fairly well adjusted regardless of his label. Thus,

theoretical orientation may influence the ways in which

clinicians apply any given diagnostic system.

The impact of the young man's label becomes even more

striking when assessments of him by the analytic clinicians

in the two experimental conditions are compared. In the "job

applicant" condition, analytic clinicians described the young

man as "'attractive and conventional looking'; 'candid and

innovative'; 'ordinary, straightforward'; 'upstanding,

middle-class citizen type, but more like a hard hat';

'probably of lower or blue-collar origins'; 'middle-class

protestant ethic orientation; fairly open--somewhat

ingenious'" (Langer & Abelson, 1974, p. 8). In the "patient"

condition, analytic clinicians described the same young man

as a "'tight, defensive person conflict over

homosexuality'; 'dependent, passive-aggressive'; 'frightened

of his own aggressive impulses'; 'fairly bright, but tries to

seem brighter than he is impulsivity shows through his

rigidity'; 'passive, dependent type'; 'considerable

hostility, repressed or channeled'" (Langer & Abelson, p. 8).

Further, when asked whether or not the young man's life

outlook was realistic, analytic clinicians in the "patient"

condition generally seemed to view the young man's outlook as

less realistic than analytic clinicians in the "job

applicant" condition. This study's results suggest that

labels and theoretical preconceptions can indeed bias the way

clinicians perceive clients. Langer and Abelson (1974)

question the very necessity of psychiatric diagnosis as

currently practiced, noting that "if the therapist were not

given the patient label, he would see a very different range

of behaviors or attribute the given behaviors to factors

other than the patient's 'illness'" (p. 9).

A second study by Langer further investigated the impact

of labels, this time on laypersons. The study showed that

even mindful perceptions can be inappropriate when it comes

to assessing deviance. In this study, subjects viewed a

videotape of a person who was identified as either an ex-

mental patient, a homosexual, a divorced person, a cancer

victim, or a millionaire (Langer & Imber, 1980). Subjects

provided with such information were considered mindful

because the novelty of the label was expected to increase the

accuracy with which they perceived the videotaped person. A

sixth mindful group of subjects was asked to attend to the

videotaped person's physical appearance. A seventh group of

subjects constituted the mindless control group; these

subjects were not given any information about the videotaped

person, nor were they told to attend to anything in

particular about the subjects

Results indicated that subjects from all of the mindful

groups, including those asked to attend to the physical

characteristics of the videotaped person, more accurately

perceived the videotaped person's characteristics than the

mindless subjects (Langer & Imber, 1980). However, mindful

subjects also rated the videotaped person's characteristics

as significantly more extreme and atypical than the mindless

subjects. The very distinctiveness that is seen in others

when one mindfully attends to them is often interpreted as

deviance. Langer and Imber's (1980) research suggests that

in making psychological diagnoses of any kind, it is

important to be aware of tendencies to evaluate distinctive

behavior as more deviant than it actually is. Ridding

ourselves of underlying biases is also necessary, since

inappropriate mindful evaluations "can be used to confirm

hypotheses that may be used as justifications for

differential treatment. Furthermore, mindful evaluations of

nondeviants may lead to assumptions of deviance also

resulting in prejudicial behavior" (Langer & Imber, 1980, p.


Suggestion effects. Temerlin (1968, 1970) conducted a

study in which the effects of diagnostic suggestion were

evaluated. Clinical psychology graduate students, clinical

psychologists, and psychiatrists were used as subjects; all

were shown a videotape of a patient. Three clinical

psychologists had previously viewed the tape, and judged the

patient mentally healthy. However, prior to watching the

videotape, experimental subjects were told by a prestigious

confederate that the patient "was 'a very interesting man

because he looks neurotic, but actually is quite psychotic'"

(Temerlin, 1968, p. 350). Four control groups were used. In

none of these control groups was it suggested to subjects

that the videotaped man was psychotic. After watching the

videotape, subjects in all groups were asked to diagnose the

patient from a counterbalanced list of various psychoses,

neuroses, and miscellaneous personality disorders, including

"normal or healthy personality." The prestigious

confederate's suggestion had a significant impact on

experimental subjects' diagnoses. None of the subjects in

any of the four control groups diagnosed the patient as

psychotic. However, in the experimental group, 60 percent of

the psychiatrists, 28 percent of the clinical psychologists,

and 11 percent of the graduate students diagnosed the patient

as psychotic (Temerlin, 1968). In thinking about DSM-III-R

and transitive diagnosis, it is important to consider the

degree to which clinicians can objectively evaluate patients;

further, consideration of factors that might influence the

diagnostic process--such as the impact of diagnostic

suggestion--is quite important.

Problem-centered approach. Two studies performed in the

early 1960s by Paul Rothaus and his colleagues compared

reactions to mental patients who describe their

psychopathology in problem centered terms to mental patients

who describe their psychopathology in mental illness terms.

In the first of these studies, patients in a Veterans

Hospital engaged in a role play which simulated a job

interview (Rothaus and Morton, 1962). Some of the patients

were assigned the role of interviewer and the other half

assigned the role of job applicant. Each interviewer-

applicant team role played the interview twice; one time the

applicant described his reason for being in the hospital as

mental illness, and the other time he described it as a

problem in getting along with people. Results indicated that

job interviewers responded more positively to the applicant

when the applicant's problem was presented in problem-

centered terms rather than mental illness terms (Rothaus and

Morton, 1962).

A second study replicated the first, except this time

thirty-two real job interviewers from the Texas Employment

Commission were used (Rothaus, Hanson, Cleveland, and

Johnson, 1963). Sixteen mental patients were trained in

presenting themselves in mental illness terms and problem-

centered terms. Patients met separately with two

interviewers, using the problem-centered approach in one

interview and the mental illness approach in the other. Job

interviewers in the problem-centered condition received

written information about the patient that explained the

patient's entry into the hospital as due to "problems in his

everyday interpersonal relationships" (Rothaus et al., 1963,

p. 86). Job interviewers in the mental illness condition

received written information about the patient that explained

the patient's entry into the hospital as due to "trouble with

his nerves" (p.86). In the problem-centered condition,

patients described themselves as having had difficulty in

social relationships and explained that their hospitalization

had taught them how to solve their problems. In the mental

illness condition, patients described themselves as having

had a nervous breakdown and explained that their

hospitalization and medication had resulted in a cure

(Rothaus et al., 1963). Interviewers rated the degree to

which they felt they could place each patient in a job.

Interviewers rated each applicant twice, once after reading

the written description and once after the personal


Results indicated that job interviewers rated patients

more positively, and therefore easier to place in jobs, when

patients explained their hospitalization in problem-centered

terms as opposed to mental illness terms. Further, in both

the mental illness and problem-centered conditions, Rothaus

et al. (1963) found that interviewers rated applicants more

positively after having met with them in person. In fact,

differences in ratings of patients in the mental illness and

problem-centered conditions only occurred after the in person

interview; no differences in interviewer ratings of the job

applicant were found based on mental illness or problem-

centered written descriptions alone. Rothaus et al. (1963)

hypothesize that "getting to know" someone with a

psychological problem often helps dispel preconceptions.

Further, their study's results suggest that the way patients

present themselves and their problems to others has an effect

on the way others react to them. In thinking about

psychiatric diagnosis, it seems important to weigh the

different impressions people receive from various ways of

describing particular disorders.

Sexism and Diagnosis

DSM-III and DSM-III-R have often been accused of being

biased against women. For example, Loring and Powell (1988)

found that both sex of client and sex of psychiatrist

influenced the way diagnoses were assigned when using DSM-

III-derived criteria. They found that male clinicians are

more likely to diagnose white women patients as depressed and

black women patients as having paranoid personality disorders

than are female clinicians (Loring & Powell, 1988). In

addition, Loring and Powell (1988) found that when diagnosing

white women patients, white women psychiatrists tend to

assign less severe diagnoses. Despite this one exception for

white women clinicians diagnosing white women patients,

psychiatrists tended to diagnose a patient most accurately

when the patient's sex and race were identical to the

psychiatrists. Further, Loring and Powell (1988) found that

psychiatrists' diagnoses had a higher degree of reliability

and accuracy when made without knowledge of a patient's sex

or race. The important question that remains to be answered

is whether DSM-III-R itself is sex biased, or whether DSM-

III-R categories are fairly objective and unbiased but

assigned in a sex biased manner in real world scenarios.

Russell (1985) argues against DSM-III because she sees

its psychiatric diagnoses as oppressing women in a number of

specific ways. She critiques DSM-III, which she feels

contains subjective value judgments regarding pathology which

are often sexist. She cites the now classic study by

Broverman, Broverman, Clarkson, Rosenkrantz, and Vogel

(1970), which found that clinicians' conceptions of a

typical, mentally healthy adult conformed to expectations of

a typical, mentally healthy male but were significantly

different from conceptions of a typical, mentally healthy

female--implying that women are likely to be seen as less

well adjusted than men. Russell (1985) contends that Western

conceptualizations of pathology are similar to Western

conceptualizations of the traditional female role. For

example, dependent personality disorder is described as

involving the impairment of independent functioning, an

inability to take responsibility for major life areas, the

subordination of one's own needs, and low self-confidence.

Russell (1985) comments:

All this has a familiar ring. It is very close to what
we are expected to be as women. If we succeed in our
role as women then we may be diagnosed as having a
personality disorder. If we do not succeed in our role
as women, we will be punished in other ways. (p. 303)

DSM-III, in Russell's view, reflects the sexism inherent in


There has been a fair amount of empirical research

regarding sexism in diagnosing abnormality. Rosenfield

(1982) provides evidence that assessments of deviance are

influenced by sex role norms and that they can be harmful for

both men and women. She examined the records of patients

entering a psychiatric emergency room and found that patients

who exhibited "deviant deviance" by engaging in abnormal

behaviors which ran counter to traditional sex role norms had

significantly higher rates of hospitalization than patients

whose abnormal behaviors were consistent with traditional sex

role norms (Rosenfield, 1982). More specifically, Rosenfield

(1982) found that males diagnosed with disorders

predominantly identified in females--such as neurotic or

psychotic forms of depression--were significantly more likely

to be hospitalized than were similarly diagnosed females. On

the other hand, females diagnosed with disorders

predominantly identified in males--such as personality and

substance abuse disorders--were significantly more likely to

be hospitalized than were similarly diagnosed males. No

significant differences in rate of hospitalization were found

for subjects diagnosed with disorders in which neither sex is

typically predominant. Rosenfield (1982) concludes that

"males are more often hospitalized for 'feminine' types of

disorders than are females. Females are more often

hospitalized for 'masculine' types of disorders than are

males" (p. 22).

Rosenfield (1982) observes that the findings might be

due to the fact that males and females exhibiting abnormal

behavior inconsistent with sex role norms are generally more

severely disturbed than others, thus explaining their higher

rates of hospitalization. While further studies seem

necessary in order fully to determine whether or not this is

so, Rosenfield negates this explanation to some extent by

noting that the number of subjects exhibiting "deviant

deviance" who were diagnosed as severely disturbed did not

differ significantly from the number of subjects whose

abnormality was consistent with sex role norms. Thus,

Rosenfield feels that severity of impairment does not explain

the results of this particular study.

Brown (1990) reports that female military veterans are

less likely to be diagnosed with posttraumatic stress

disorder than their male counterparts; such women are often

misdiagnosed with personality disorders. Similarly, Caplan

(1992) notes that learning disabilities are diagnosed far

more often in boys, even though recent research indicates

that such disorders occur equally in both boys and girls.

While diagnostic bias does happen, it is unclear how much

DSM-III-R is to blame. An objective manual in a sexist

society is bound to be used in sex biased ways.

Caplan (1992) does not believe that DSM-III-R is

objective; she sees a great deal of sexism in it. She

expresses particular concern over two new diagnoses contained

in DSM-III-R that she believes are sexist: self-defeating

personality disorder and late luteal phase dysphoric

disorder. Self-defeating personality disorder applies to the

battered wife who remains within an abusive relationship.

Caplan (1992) feels this diagnostic category is dangerous

"because it leads both therapists and the women so diagnosed

to believe that the problems come from within, that the women

have a sick need to be hurt, humiliated, unappreciated, etc."

(p. 75). Late luteal phase dysphoric disorder involves

emotionality resulting from premenstrual syndrome. Caplan

(1992) argues that classifying premenstrual syndrome as a

mental disorder can easily be used "to justify keeping women

out of well-paying, responsible jobs" (p. 76). She adds that

extremely masculine behaviors are not considered disorders;

for example, there is no "Macho Personality Disorder" or

"Testosterone-Based Aggressive Disorder" (Caplan, 1992). In

Caplan's (1992) view, DSM-III-R diagnosis is extremely biased

against women. She concludes that "the human and clinical

applications of biased categories need rapid and radical

transformation" (p. 78).

Racism and Diagnosis

Controversy over the effect of race on diagnosis has not

been resolved, since interpretations of the studies done in

this area vary. Fernando (1993) points out that the history

of psychiatry has been influenced by racist trends in Western

culture, and Worthington's (1992) review of empirical

research conducted between 1965 and 1991 does seem to

indicate that diagnosis is often biased along racial lines.

Rosenfield (1984) conducted a study that investigated the

impact of racial bias in diagnosis and treatment of the

mentally ill. She found that non-white males are

involuntarily hospitalized more often than their white

counterparts. However, she concluded that whites are less

stigmatized by mental disorder diagnoses and hospitalized

less often because "the more coercive conditions under which

nonwhites enter treatment determine the more severe responses

to nonwhites' deviance" (Rosenfield, 1984, p. 21). Thus,

racial biases within the DSM-III and DSM-III-R diagnostic

systems may not be the primary reason that whites and non-

whites are treated differently with regard to diagnosis and

hospitalization; rather, accurate DSM-III-R diagnoses may

simply be applied in a racially biased manner.

Loring and Powell's (1988) previously mentioned study

directly addressed the impact of race on diagnosis of mental

disorders, and found that more severe diagnoses are often

assigned to African Americans. More specifically, they found

that race had an effect on the diagnoses assigned to African

Americans, especially African American males. For example,

Loring and Powell (1988) discovered that African American

males are more likely to be diagnosed with a paranoid

schizophrenic disorder than are other patients. Further,

African Americans, regardless of sex, are more likely to be

diagnosed with paranoid personality disorders than are white

patients. In general, Loring and Powell (1988) found that

"clinicians appear to ascribe violence, suspiciousness, and

dangerousness to black clients even though the case studies

are the same as the case studies for white clients" (p. 18).

In their study, diagnostic bias according to race occurred

even when the clinicians making the diagnoses were African

Americans; that is, even African American clinicians were

more likely to diagnose African American males as suffering

from paranoid schizophrenic disorders. Again, whether the

occurrence of racial bias is due to DSM diagnosis itself or

the effects of societal biases on the way diagnoses are

assigned is unclear.

Other research supports Loring and Powell's (1988)

findings. Pavkov, Lewis, and Lyons (1989) conducted a

logistic regression analysis on the types of diagnoses

assigned severely disturbed people, and found that being

African American predicts a schizophrenic diagnosis. Lawson,

Hepler, Holladay, and Cuffel (1994) examined census data from

the Tennessee Department of Mental Health for 1984 and 1990.

Results indicated that African Americans were overrepresented

in public inpatient facilities compared to their proportion

in the population. Further, Lawson et al. (1994) "found that

African Americans were overrepresented among patients with a

diagnosis of schizophrenia" (p.73) and that "African American

inpatients were overrepresented in most other diagnostic

categories as well" (p. 73). Lawson et al. (1990) point out

that DSM-III-R was throughout Tennessee in 1990, implying

that it was used in a racially biased manner, if not racially

biased itself.

Neighbors, Jackson, Campbell, and Williams (1989)

reviewed literature on racism and diagnosis, and concluded

that such research makes one of two contradictory

assumptions. Some researchers assume that African American

and Caucasian persons exhibit similar symptoms when afflicted

with a disorder, but that these symptoms are perceived as

different by clinicians. Other researchers assume that

African American and Caucasian persons actually do exhibit

symptoms differently, but that clinicians lacking cross-

cultural knowledge are not sensitive to these differences.

Neighbors et al. (1989) suggest that structured interview

research might help clarify controversy surrounding race and

diagnosis. However, they observe that the DSM-III-R

diagnostic criteria used in structured interviews were

"compiled by clinicians working with predominantly white

patient populations. Thus, one could argue that we are

merely replacing one biased system with another" (Neighbors

et al., 1989). Despite this concern, Neighbors et al. (1990)

believe that epidemiological research on mental disorders and

how they affect both African Americans and Caucasians can

better help researchers to understand how race issues impact

on diagnostic inferences.

Culture bias. Rothblum, Solomon, and Albee (1986) argue

that DSM diagnosis fails to consider sociocultural factors,

and therefore takes behaviors that are socially deviant and

transforms them into diseases. For example, Rothblum,

Solomon, and Albee (1986) feel that men who rape women or

abuse children are often diagnosed with personality

disorders. However, they contend that sociocultural factors

influencing these men's behavior, such as the sexist nature

of American society, are overlooked in DSM-III. The issue of

whether DSM-style diagnosis can be generalized across

cultures is a controversial one. Brown (1990) argues that

client assessments often fail to take culture into account.

However, research by Martin and Grubb (1990) implies that

taking culture into account does not necessarily lead to

avoidance of culture biased diagnoses. They found that white

juvenile offenders are often perceived as having a

psychological disorder, whereas black juvenile offenders are

often perceived as behaving in culturally expected ways.

Cohen and Carlin (1993) also found culture to be important in

diagnosis. Their research indicated that African Americans

and Caucasians diagnosed with dementia often show differences

"in behavioral and cognitive symptomatology" (p. 383). Cohen

and Carlin (1993) appear to perceive such differences as due

more to sociocultural influences on the way dementia is

manifested than in diagnostic bias within DSM-III-R.

Criticisms of Diagnosis in Relation to the Present Study

While criticisms of psychiatric diagnosis as sexist,

racist, and culturally biased can appear quite convincing at

times, it is important to remember that these criticisms may

simply indicate that the society from which current

diagnostic manuals came is sexist, racist, and culturally

biased in ways of which it is often unaware; based on the

contradictory studies in this area, whether or not these

biases have influenced DSM in its various forms is not

entirely clear. Those who favor DSM-style diagnosis

convincingly argue that if racial, cultural, and gender

biases can be eliminated, then a truly objective, reliable,

and valid diagnostic manual can be created. Further, it can

be argued that DSM is not biased, but simply applied in

biased ways. On the other hand, those who question DSM-III-R

diagnosis often wonder whether or not researchers in favor of

categorical diagnosis can free subsequent revisions of DSM

from societal biases and produce a truly objective diagnostic


The ability to produce an objective, categorical system

of diagnosis is a crucial area of debate between those in

favor of DSM-III-R diagnosis and those who oppose it; firm

consensus on this issue has yet to be reached. Those who

develop alternatives to DSM, such as Faidley and Leitner

(1993) need to be aware of the many complex issues involved

in creating any diagnostic system. Sexism, racism, culture

bias, labeling effects, and underlying theoretical

assumptions all impact on the system of diagnosis that one

creates. If each of these issues is not addressed in a

cogent fashion, the resulting diagnostic scheme is bound to

suffer. Herein, an initial attempt is made to investigate

labeling effects by assessing the impact of DSM-III-R and

transitive diagnoses on perceptions of a person with a

psychological problem.

Hypotheses and Overview of the Study

A vignette study, in which three DSM diagnoses were

presented in DSM form as well as translated and presented in

transitive diagnosis form, was performed in order to compare

the reactions of laypersons to DSM-III-R diagnoses and

transitive diagnoses. Laypersons were used as subjects

rather than mental health professionals because laypersons

generally have little knowledge about either DSM-III-R

diagnosis or transitive diagnosis. Had mental health

professionals been used, they would generally have been

familiar with DSM-III-R diagnosis and unfamiliar with

transitive diagnosis. This might have influenced their

responses. Layperson subjects, by comparison, provided the

opportunity to examine the ways in which DSM diagnosis and

transitive diagnosis impact on people who possess little or

no knowledge of either.

While the research literature is inconclusive regarding

the effects of labels, a substantial number of studies

indicate that labels influence the ways in which we perceive

others (Burk & Sher, 1990; Kus & Carpenter, 1991; Langer &

Abelson, 1974; Langer & Imber, 1980; Link, 1987; Link &

Cullen, 1990; Link et al., 1987; Link et al., 1989; Link et

al., 1991; Scheff, 1966). In the present study, it was

predicted that DSM diagnosis and transitive diagnosis would

create different perceptions of a diagnosed person. DSM

diagnosis, as a result of its emphasis on labeling and

treatment, was expected to make people desire greater social

distance from those diagnosed than is transitive diagnosis.

Hypothesis 1: Subjects who read a brief description of a
man will desire greater social distance from the man
when the man is described as having a past psychological
problem and given a DSM-III-R diagnosis than when the
man is described as having a past psychological problem
and given a transitive diagnosis. Further, it is
predicted that subjects will desire the least social

distance from the man when the man is not described as
having a psychological problem at all.

Using a series of bipolar adjectives, Burk and Sher (1991)

found that mentally ill teenagers were generally rated as

more sad, weak, rebellious, abnormal, nervous, rejected,

rough, bad, unfriendly, inactive, and unable to control their

behavior than "typical" teenagers. In the present study, it

is hypothesized that a similar pattern will occur. DSM

diagnosis is expected to result in more negative ratings of a

diagnosed person than is transitive diagnosis.

Hypothesis 2: Subjects who read a brief description of a
man will perceive the man differently along a series of
bipolar dimensions when the man is described as having a
past psychological problem and given a DSM-III-R
diagnosis than when the man is described as having a
past psychological problem and given a transitive

In keeping with the first hypothesis, wherein it is predicted

that subjects will desire greater social distance from a

person when that person is described in DSM diagnosis terms

than when described in transitive diagnosis terms, it is also

predicted that subjects will see themselves as less similar

to a person when the person is DSM diagnosed than when

transitively diagnosed. This is expected because people will

desire greater distance from a "sick" person with a "mental

illness" than they will from a person whose thoughts and

feelings are leading to maladaptive constructions of the


Hypothesis 3: Subjects who read a brief description of a
man will rate the man as more dissimilar from themselves
on a series of bipolar dimensions when the man is
described as having a past psychological problem and
given a DSM diagnosis than when the man is described as

having a past psychological problem and given a
transitive diagnosis. Further, it is predicted that
subjects will rate the man as least dissimilar from
themselves when the man is not described as having a
psychological problem at all.

Further, because of the way in which the DSM diagnostic

labels are expected to impact on subjects, subjects are

predicted to perceive a diagnosed person as less similar to a

typical person when the diagnosed person is described in DSM

terms rather than transitive diagnosis terms.

Hypothesis 4: Subjects who read a brief description of a
man will rate the man as more dissimilar from a typical,
normal person on a series of bipolar dimensions when the
man is described as having a past psychological problem
and given a DSM-III-R diagnosis than when the man is
described as having a past psychological problem and
given a transitive diagnosis. Further, it is predicted
that subjects will rate the man as least dissimilar from
a typical person when the man is not described as having
a psychological problem at all.

The DSM disorders that will be presented in both DSM form and

a translated transitive diagnosis form are adjustment

disorder, dysthymia, and schizophrenia. The hypotheses above

are expected to hold true for all three of these disorders.



Subjects were 364 introductory psychology students. All

were recruited through the University of Florida subject

pool, and were tested in small groups of 5-12 subjects.

Subjects signed up for the study by responding to a notice

placed on the psychology department bulletin board. When the

subjects arrived for the study, they were randomly assigned

to one of three groups; this was accomplished by randomly

distributing questionnaire booklets to subjects. Since

subjects could receive one of three versions of the

questionnaire booklet, their group was determined by the

booklet they randomly received. Of the 364 subjects, 258

were white, 28 were African-American, 49 were Hispanic, and

29 were Asian. Mean age of subjects was 18.49 years.


A brief vignette of a man, adapted from Link et al.

(1987), was used. The vignette described a man named Jim

Johnson. While brief, it provided detailed information about

his salary, personal hygiene habits, work behavior, job

aspirations, and plans to meet young women (Link et al.,

1987). Seven versions of the vignette were used. In six of

the seven versions, the vignette began with a paragraph

describing Jim Johnson as having had a past psychological

problem. Six versions described Jim with a past

psychological problem because three different psychological

problems were described, each in two different ways (once

using DSM-III-R diagnosis terminology and once using

transitive diagnosis terminology). In the first version of

the vignette, Jim's psychological problem was described as an

adjustment disorder. In the second version, his

psychological problem involved the behaviors of an adjustment

disorder, but was described in transitive diagnosis terms.

In the third version of the vignette, Jim's psychological

problem was described as dysthymia. In the fourth version,

his psychological problem encompassed dysthymic behaviors,

but was described in transitive diagnosis terms. In the

fifth version of the vignette, Jim's psychological problem

was described as histrionic personality disorder. In the

sixth version, his psychological problem involved histrionic

personality disorder behaviors, but was described in

transitive diagnosis terms. In the seventh version, Jim was

not identified as having a past psychological problem, and no

diagnosis accompanied the vignette. The seven variations of

the Jim Johnson vignette are reproduced in Appendix A.

Certain similarities existed between all three of the

DSM-III-R diagnosis descriptions attached to the beginning of

the Jim Johnson vignette. First of all, each one presented

Jim Johnson with a DSM-III-R diagnostic label, and then

proceeded to describe what the disorder specified by that

label involves. Descriptions of disorders were adapted from

DSM-III-R descriptions; the only changes were to simplify the

descriptions' language in order to make them more accessible

to lay readers. Second, at the end of each DSM-III-R

description was the following statement: "Once Jim Johnson

received treatment he felt much better." This statement was

included in order to maintain consistency with DSM-III-R's

self described goal, in which a DSM-III-R diagnosis is the

initial step "leading to the formulation of a treatment plan"

(APA, 1987, p. xxvi). Further, this statement is consistent

with the medically oriented language of DSM-III-R and the

fact that virtually all DSM-III-R categories are contained in


Each of the three transitive diagnosis descriptions

attached to the beginning of the Jim Johnson vignette

presented the same information as its corresponding DSM-III-R

diagnosis description. For example, the behaviors in the

transitive diagnosis description of dysthymia corresponded to

the dysthymic behaviors listed in both DSM-III-R and the DSM-

III-R dysthymia description used in this study. That is, the

same behaviors that are part of a DSM-III-R diagnosis of

dysthymia were included in the transitive diagnosis

translation of dysthymia. The same was true for the

transitive diagnoses of adjustment disorder and

schizophrenia. All three transitive diagnoses included

statements about the way Jim Johnson's problem "was caused by

the way he construed events in his life." Further, all three

noted that his behaviors constituted "responses to the way he

thought and felt about events in his life." All ended with

the following statement: "Once Jim learned to construe events

differently he felt much better."

It was considered important to keep the information in

all three transitive diagnoses consistent with information

about the corresponding DSM-III-R disorders on which they

were based. In this way, study results reflect differences

in perceptions of the same information presented in two

different ways. However, it is important to note that true

transitive diagnoses are even more idiographic, detailed,

descriptive, and process oriented than the ones employed in

the current study (see Faidley & Leitner, 1993). However, in

order to effectively compare transitive diagnosis to DSM-III-

R diagnosis, transitive diagnoses needed to be created which

would reflect specific DSM-III-R disorders and which could be

easily and quickly presentable to subjects.

In order to assess reactions to the seven versions of

the Jim Johnson vignette, a measure of social distance (Link

et al., 1987) was used. This measure consists of seven items

which assess the degree of social distance from Jim Johnson

desired by respondents. Link et al. (1987) reported an

internal consistency reliability rating of .92 for this

measure (using Cronbach's a). Each item was scored using a

four-point Likert format (definitely unwilling, probably

unwilling, probably willing, definitely unwilling), with "0"

being "definitely unwilling" and "3" being "definitely

willing." An overall social distance measure was obtained by

adding together a subject's responses to each item, and then

dividing the total by seven. All seven social distance items

are included in their entirety in Appendix B.

Reactions to the seven versions of the vignette were

also assessed using a series of 11 bipolar adjectives. These

specific bipolar adjectives were first used by Chassin and

her colleagues (Chassin, Eason, & Young, 1981; Chassin,

Presson, Young, & Light, 1981) in a pair of studies of the

effects of labeling on institutionalized adolescents, and

later used by Burk and Sher (1990) in a study of labeling

children of alcoholics. Because of the bipolar adjectives'

usefulness in providing information about perceptions of

deviance in these studies, they seemed particularly

appropriate for use in the current study. The 11 bipolar

adjectives were placed on seven point scales, modeled on

those employed by Osgood, Suci, and Tannenbaum (1969), and

were used to have subjects evaluate themselves, a "typical"

person, and Jim Johnson. The adjectives employed, as well as

the descriptors below each number on the seven-point scale,

are contained in Appendix C.

Prior to the actual study reported herein, a 23 subject

pilot study that used the procedures outlined above was

conducted; at the conclusion of this pilot study, a "funnel

questionnaire" (see Page, 1969, 1971, 1973; Page & Kahle,

1976; Page & Scheidt, 1971) was given to subjects in order to

insure that subjects were not influenced by the demand

characteristics of the study (e.g. guessing the hypotheses,

expecting particular responses are desired, etc.). With a

funnel questionnaire, subjects begin by answering broad,

open-ended questions about the study (e.g. "What was the

purpose of this study?"), and proceed to more and more

specific questions about the study until they are told the

hypothesis and asked directly whether or not they figured it

out while participating in the study. None of the pilot

study subjects who completed the funnel questionnaire were

able to guess any of the hypotheses of the experiment. The

funnel questionnaire that was used in the pilot study can be

found in Appendix D.


Subjects were divided into seven groups of subjects.

Each group was asked to read one version of the Jim Johnson

vignette. The group that read the description of Jim Johnson

containing no psychological problem or diagnosis served as a

control group. The remaining six groups of subjects

constituted experimental groups.

After reading one of the seven vignettes, subjects were

asked to fill out the social distance items. In addition,

they were asked to evaluate themselves, Jim Johnson, and a

"typical" person along each of the 11 bipolar adjective

dimensions. In order to control for order of presentation

effects, the order in which the social distance items and 11

bipolar adjectives were presented was counterbalanced; half

of the subjects filled out the social distance items before

completing bipolar adjective ratings, while the remaining

subjects completed these tasks in reverse order. The order

in which subjects rated themselves, Jim Johnson, and a

"typical" person along the 11 bipolar adjective dimensions

was also counterbalanced.

Once subjects read the vignette and completed the social

distance scale and bipolar adjective ratings, they were

thoroughly debriefed, and any questions they had were


Two difference scores for each of the 11 bipolar

adjective dimensions were calculated. The first score,

herein called the "self difference" score, measures how

differently subjects rated themselves from Jim Johnson; for

each of the 11 bipolar dimension, this score was calculated

by counting the number of points on the seven point scale

between ratings of self and ratings of Jim Johnson. The

second score, herein called the "typical difference" score,

measures how differently subjects rated a "typical person"

from Jim Johnson; for each bipolar dimension, this score was

calculated by counting the number of points on the seven

point scale between ratings of a "typical" person and ratings

of Jim Johnson.

Subjects' difference scores were used to calculate two

"overall dissimilarity" scores. The first of these, the

"overall self dissimilarity" score, indicates subjects'

rating of dissimilarity between themselves and Jim Johnson

across all 11 bipolar dimensions. The "overall self

dissimilarity" score was derived by summing each subject's 11

"self difference" scores. The second of these, the "overall

typical dissimilarity" score, indicates subjects' rating of

dissimilarity between a "typical person" and Jim Johnson

across all 11 bipolar dimensions. The "overall typical

dissimilarity" score was derived by summing each subject's 11

"typical difference" scores.


The purpose of the study was to compare layperson

reactions to Jim Johnson when he was described according to

either transitive diagnosis or DSM-III-R diagnosis. It was

hypothesized that when Jim Johnson was described according to

DSM-III-R diagnosis, he would experience greater social

rejection. In other words, it was predicted that when Jim

Johnson was described in DSM-III-R terms, subjects would

desire greater social distance from him, rate him as more

dissimilar from themselves on a series of bipolar dimensions,

and rate him as more dissimilar from a "typical" person on a

series of bipolar dimensions than when he was described in

transitive diagnosis terms. Table 1 provides descriptive

statistics for all of the dependent measures for which there

were significant differences based on type of diagnosis (DSM-

III-R, transitive, or none).

Before the social rejection hypotheses were tested, a

principal components factor analysis was performed in order

to determine whether the three dependent measures (social

distance, overall self dissimilarity, and overall typical

dissimilarity) were independent of each other, and were each

measuring a particular attribute of social rejection. The

factor analysis suggested only two factors, determined using

the criterion of minimum eigenvalue greater than or equal to

1. A scree plot also indicated that a 2 factor solution was


Table 1. Descriptive Statistics for Dependent Measures
Yielding Significant Results For Type of Diagnosis

Social Distance
DSM-III-R 155 1.142 .450
Transitive 157 1.159 .471
No Diagnosis 52 .648 .047
Negative Outward Impression
DSM-III-R 155 3.462 .991
Transitive 157 3.425 .915
No Diagnosis 52 2.658 .105
Unacceptable Behavior
DSM-III-R 155 5.475 1.049
Transitive 157 5.461 .963
No Diagnosis 52 5.987 .679

Overall self dissimilarity and overall typical

dissimilarity both loaded highly on Factor 1; social distance

constituted Factor 2 (see Table 2 for a summary of oblique

factor loadings). Thus, in statistically analyzing the

degree to which subjects socially rejected Jim Johnson, two

measures were employed. The first measure collapsed overall

self dissimilarity score and overall typical dissimilarity by

averaging them together into an "overall dissimilarity"

score. The second measure was social distance rating.

Table 2. Oblique Factor Loadings for Principal Components
Factor Analysis of Social Rejection Measures

Overall Self Dissimilarity .864 .061
Overall Typical Dissimilarity .881 -.035
Social Distance Rating -.003 .999

A one-way MANOVA was employed in order to test whether

subjects rejected Jim Johnson based on the kind of diagnosis

he was assigned. The independent variable of interest was

type of diagnosis (DSM-III-R, transitive, or none). The

dependent variables were the social distance rating and the

overall dissimilarity score. MANOVAs testing the main and

interaction effects of sex of subject and order of social

distance scale presentation (i.e. whether or not subjects

filled out the social distance scale rating for themselves or

a "typical" person first) were also performed in order to

test whether these variables influenced the results. No

effect was found for sex of subjects, Wilk's lambda = 1.00,

F(2, 352) = .900, p > .05, or for order of presentation,

Wilk's lambda = 1.00, E(2, 352) = .196, p > .05. Further,

there were no significant interaction effects for sex and

type of diagnosis, Wilk's lambda = .98, E(2, 352) = 13.394, p

> .05, or for sex and order, Wilk's lambda = 1.00, E(2, 352)

= .774, p > .05. There also was no interaction effect for

type of diagnosis and order, Wilk's lambda = .99, F(2, 352) =

.675, p > .05. Finally, there was no three way interaction

effect for type of diagnosis, order, and sex, Wilk's lambda =

.99, F(2, 352) = .941, p > .05. Based on these statistics,

it is safe to assume that sex and order of presentation

variables did not confound the study's results.

The MANOVA testing the main effect of type of diagnosis

was highly significant, Wilk's lambda = .86, E(2, 352) =

13.876, p < .0001. In other words, subjects responded

differently to Jim Johnson based on the type of vignette they

read about him. In order to isolate the effect of type of

diagnosis, two one-way ANOVAs were performed in order to

determine whether type of diagnosis had a greater impact on

social distance ratings or overall dissimilarity ratings.

The ANOVA testing the effect of type of diagnosis on overall

dissimilarity ratings did not produce significant results,

E(2, 361) = 1.90, p > .05. However, the ANOVA testing the

effect of type of diagnosis on social distance did produce a

significant result, (2, 361) = 28.45, p < .0001.

The Tukey-Kramer Honestly Significant Difference (HSD)

procedure was used to further isolate the effect of diagnosis

type on social distance scale ratings of Jim Johnson. Tukey-

Kramer's procedure was chosen because it assumes that cell

sizes are either equal or extremely different from each

other. In the present study, cell sizes in the transitive

and DSM-III-R diagnosis groups were extremely similar, but

differed substantially from the no diagnosis group. Thus,

Tukey-Kramer, which accounts for extremely different cell

sizes, seemed appropriate. Results of the Tukey-Kramer

procedure, which made all pairwise comparisons, indicated

that subjects in the no diagnosis condition provided social

distance ratings of Jim Johnson that were significantly

different from those given by subjects in either the

transitive diagnosis or DSM-III-R diagnosis conditions. That

is, no diagnosis subjects desired significantly less social

distance from Jim Johnson than transitive diagnosis subjects

or DSM-III-R diagnosis subjects. There was no significant

difference in social distance ratings of Jim Johnson when

comparing transitive diagnosis subjects to DSM-III-R

subjects; both the transitive and DSM-III-R vignettes

resulted in subjects desiring approximately equal amounts of

social distance from Jim Johnson. Table 3 presents the

results of the Tukey-Kramer comparisons in their entirety.

See Figure 1 for a line graph of the way subjects in each of

the three groups rated Jim Johnson on the social distance


Table 3. Pairwise Comparisons of Type of Diagnosis for
Social Distance Ratings Using Tukey-Kramer Procedure

No diagnosis
vs. DSM-III-R .494 .168*
vs. Transitive .511 .168*
vs. Transitive .017 .119
*indicates significant at the .05 level

Because three different DSM-III-R and transitive

diagnosis vignettes were used, a secondary analysis was

undertaken in order to ascertain whether or not results were

confounded by the type of diagnosis vignette. This analysis

involved a one-way MANOVA testing whether there was an

interaction between type of diagnosis and type of vignette.

This analysis was done separately because, had type of

vignette been included in the original series of MANOVAs, it

would have negated the effect of type of diagnosis by causing

the same data to be included in both categories. The MANOVA

testing the interaction between type of diagnosis and type of

vignette yielded no significant results, Wilk's lambda = .98,

E(2, 352) = 1.127, p > .05. This MANOVA indicated that there

was no interaction between the form of diagnostic vignette

(histrionic, adjustment, or dysthymic) and the method of

diagnosis (transitive or DSM-III-R).

1.2 -



&; .8

DSM-III-R Transitive None

Type of Diagnosis

Figure 1. The Effects of Type of Diagnosis on Social
Distance Ratings.

The study hypothesized that subjects would evaluate Jim

Johnson differently along the bipolar dimensions depending

upon which type of diagnosis (DSM-III-R, transitive, or none)

he received. Prior to testing this hypothesis, a principal

components factor analysis was performed. The factor

analysis suggested inclusion of only three factors,

determined using the criterion of minimum eigenvalue equal or

greater than 1. As in the previously reported factor

analysis, a scree plot was also performed. The scree plot

confirmed that a 3 factor solution was appropriate. Happy-

sad, strong-weak, normal- abnormal, relaxed-nervous, and

accepted-rejected all loaded highly on Factor 1. Obedient-

rebellious, gentle-rough, and good-bad loaded highly on

Factor 2. Active-inactive constituted Factor 3. Can control

behavior-can't control behavior and friendly-unfriendly did

not load highly on a single factor, and were therefore

eliminated from subsequent analyses. Table 4 presents a

summary of oblique factor loadings for the eleven bipolar


Table 4. Oblique Factor Loadings for Principal Components
Factor Analysis of Eleven Bipolar Dimensions

Happy-Sad -.706 -.000 .142
Strong-Weak .600 .097 -.246
Obedient-Rebellious -.045 .813 -.018
Normal-Abnormal -.672 .274 -.214
Relaxed-Nervous .856 -.013 .233
Accepted-Rejected .598 -.165 -.149
Gentle-Rough .047 .828 .011
Good-Bad -.090 .808 .202
Can control-behavior-
Can't control behavior .461 -.465 .043
Friendly-Unfriendly .023 -.484 -.686
Active-Inactive .104 -.087 .898

Subjects' ratings of Jim Johnson on the items that

loaded on Factor 1 were averaged into a single Factor 1

score. Because all Factor 1 items seemed to revolve around

the impact of one's outward appearance, this averaged score

was named the "negative outward impression rating," with