Law and Human Behavior, Vol. 21, No.

2, 1997

The MacArthur Adjudicative Competence Study:

Development and Validation of a Research
Instrument
Steven K. Hoge,1 Richard J. Bonnie,2 Nonnan Poythress,3 John Monahan4
Maiiene Eisenberg,5 and Thomas Feucht­Haviar5

Assessment of competence to stand trial is a common evaluation that can have

substantial consequences for defendants and the criminal justice system. Despite a
voluminous literature, much remains unknown. An obstacle to progress in
understanding what is better termed "adjudicative competence" is the absence of
structured, standardized research measures for assessment of defendants. This article
presents the legal framework, assessment strategy, instrument description, psychometric
properties, and construct validation of the MacArthur Structured Assessment of the
Competencies of Criminal Defendants (MacSAC­CD). The measures meet or exceed
accepted indices of internal consistency, and interscorer agreement. Observed patterns
of correlations among measures support the underlying theoretical structure of
competence­related abilities. Moreover, the MacSAC­CD distinguishes groups of
competent and incompetent defendants; reflects changes in competence status; and
correlates positively with clinical judgments, negatively with psychopathology and
impaired cognitive functioning, and negligibly with cynicism toward the justice system

In 8%­15% of felony cases, defense attorneys have doubts about their clients* com­
petence to assist counsel and to participate in the process of adjudication (Hoge,
Bonnie, Poythress, & Monahan, 1992; Poythress, Bonnie, Hoge, Monahan, & Ober­
lander, 1994). It is not surprising, therefore, that the assessment of competence to
stand trial is one of the most frequently requested types of forensic evaluation in
the United States, accounting for an estimated 25,000 examinations per year (Stead­
man & Hartstone, 1983).6
1University of Virginia Schools of Law and Medicine, 580 Massie Road, Charlottesville, VA 22903-1789
(E-mail: [email protected]).
2University of Virginia School of Law.
3Florida Mental Health Institute, University of South Florida.
4University of Virginia School of Law
5Institute of Law, Psychiatry, and Public Policy, Univeisily of Virginia.
"Steadman and Hartstone (1983) based their estimate of the number of competence evaluations on the
number of defendants committed to public mental institutions for restoration of competence and

141
0147-7307/9W«00>141$12.50 C 1997 American Psychology-Law Society/Division 41 of the American Psychological Association
142 Hoge, Bonnie, Paythress, Monahan,Eisenberg,and Feucht­Haviar

While most evaluations occur on an "outpatient" basis (which includes jailed

defendants), approximately 7,000 defendants per year are involuntarily committed
to public mental institutions for evaluation of their competence to stand trial
(Steadman & Hartstone, 1983). These commitments entail significant deprivations
of liberty for defendants, who might otherwise be free before trial, considerable
financial costs for the public, and substantial delays in the criminal justice process.
Moreover, subsequent to evaluation, some of these defendants will be confined for
longer periods of time for restoration of competence at the price of even greater
loss of liberty, expense, and delay (Steadman, Monahan, Hartstone, Davis, & Rob­
bins, 1982).
The thoroughness and quality of forensic mental health professionals' evalu­
ation of competence has substantial ramifications. On the one hand, an examiner's
failure to identify significant impairments of the defendant's abilities to assist coun­
sel or to make rational, self­interested decisions could compromise the fairness of
the ensuing adjudication. Conversely, an examiner may erroneously conclude that
the defendant is incompetent, a determination that will lead, unnecessarily, to sus­
pension of the criminal proceedings and perhaps to a period of compulsory con­
finement and treatment.
Given the importance of assessments of competence to stand trial, and the
frequency with which courts refer competence questions to mental health profes­
sionals, it is not surprising that a voluminous legal and mental health literature has
been generated on the topic (for reviews, see Golding & Roescfa, 1988; Grisso,
1992; Nicholson & Kugler, 1991; Roesch, Ogloff, & Golding, 1993). Nonetheless,
much remains unknown about defendant competencies in the criminal adjudication
process, and a firm empirical base for the clinical assessment of competence has
yet to be established. For example, little is known about the range of relevant abili­
ties in the general defendant population, or about the relationship of mental dis­
orders to these abilities, or about ways in which the competence­related abilities
of defendants with mental disorder differ from those of nondisordered defendants.
Absent such information, clinical and judicial judgments of defendants' competence
will remain critically uninformed (Nicholson, Robertson, Johnson, & Jensen, 1988).

LIMITATIONS OF EXISTING MEASURES

The primary obstacle to progress in this area is the absence of adequate meas-
ures of competence-related abilities. Some may consider this a surprising claim in
light of the number and diversity of instruments that have been developed. A review
of the literature reveals no fewer than eight instruments, including two checklists
suggesting potential elements of a competence assessment (Bukatman, Foy, & De-
research on the ratio of evaluations to commitments. This estimate, which described state forensic
systems that had not been deinstitutionalized or decentralized, is now 15 years out of date. In a recent
50-state survey of forensic evaluation systems (Grisso, Cocozza, Steadman, Fisher, & Greer, 1994) state
forensic directors were asked to estimate the annual number of evaluations of competence to stand
trial, ranging from "low-end" to "high-end." These estimates summed to 24,000 and 39,000, respectively
(unpublished data, personal communication from Thomas Grisso).
MacArthur Adjudicative Competence Study 143

Grazia, 1971; Robey, 1965); three "screening" measures, including the Competency
Screening Test (CST; Lipsitt, Lelos, & McGany, 1971), the Georgia Court Compe­
tency Test (GCCT, Wildman et al., 1979), and the Computer Assisted Competence
Assessment Tool (CADCOMP; Barnard et al., 1991); two more comprehensive meas­
ures based on semistructured interviews, the Competency to Stand Trial Assessment
Instrument (CAI; Laboratory of Community Psychiatry, 1973) and the Interdiscipli­
nary Fitness Interview (IFI; Golding, Roesch, & Schreiber, 1984); and one measure
for use specifically with defendants who are mentally retarded, the Competence for
Standing Trial for Defendants with Mental Retardation (CAST­MR; Everington,
1990).
Despite this considerable array of instruments, each has shortcomings that limit
its utility. As a prelude to the introduction of a new research measure for evaluating
pretrial competence, we review briefly four key characteristics that an adequate
research measure, ideally, would have. The interested reader is referred to recent
reviews for more comprehensive discussions of the strengths and limitations of many
of these measures (Grisso, 1986,1991,1996; Melton, Petrila, Poythress, & Slobogin,
1987; Nicholson, 1993).

Content Derived from Legal Theory of Competence

Following Grisso's (1986) model, the starting point in the development of fo­
rensic assessment instruments is a careful consideration of the relevant legal con­
struct to ensure that important legal domains are captured by the instrument's
content. Each of the existing measures includes some relevant legal content; how­
ever, none was derived from a coherent theory of the legal construct of competence.
Particularly vulnerable to this concern are the screening measures, such as the CST
and GCCIJ which yield only single cutoff scores rather than quantitative indices
reflecting performance in discrete legal domains. Further, the "legal" content in
some of these measures is of dubious value; for example, 7 of the GCCT's 17 items
require a defendant only to identify "who sits where" by looking at a picture of a
courtroom. More comprehensive, interview­based measures such as the CAI or IFI
encourage more flexible and in­depth inquiries on legal issues, although even in
these measures the organization of such inquiries is not guided by a coherent legal
theory. And, as discussed below, the lack of standardized administration of these
interview­based measures means that the legal content assessed may vary from cli­
nician to clinician and from case to case.

Delineation of Competence-Related Abilities

Consideration of the legal construct of competence should give rise to a de-

lineation of discrete legally relevant functional abilities required in the legal do-
mains. Measures can then be constructed using items tailored to these discrete
competence-related abilities. Exemplary forensic assessment instruments in this re-
gard include Grisso's (1981) measures for assessing juveniles' competence to waive
144 Hoge, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar

Miranda rights, and the research measures for evaluating treatment competence
developed by Grisso and Appelbaum (1995).
None of the measures currently available yields quantitative indices of discrete
competence­related abilities. Most measures are limited to items that assess a de­
fendant's current cognitive understanding of various facets of his or her case or
the anticipated legal proceedings (usually trial proceedings); none systematically
poses problems relating to decisions that a defendant might face or assesses ability
to reason about these issues (e.g., weighing alternative courses of action) or assesses
the defendant's beliefs or perceptions concerning how he or she will be treated by
various authorities in the criminal justice system.

Assessment of Capacity versus Current Knowledge

As will be discussed in further detail below, the prevailing legal standard for
pretrial competence is based on a defendant's capacity. Existing measures, however,
encourage the administration of questions that assess current knowledge, e.g.,
"What does the judge do during trial?" (GCCT), or "Who is the only one at your
trial who can call on you to testify?" (CAI). Answers to such questions reveal the
knowledge that a defendant brings to the Interview situation, however that infor­
mation may have been obtained (e.g., watching L.A. Law, talking to a cell mate).
However, when poor answers are received, the researcher or clinician may not be
able to differentiate between a defendant who is capable (although currently igno­
rant) and one whose capacity is in some way impaired. In other words, current
knowledge is a satisfactory proxy for capacity only when a defendant is well in­
formed and can provide "good" answers. When answers to "current knowledge"
questions are inadequate, the question of capacity—which is the pertinent legal
query—remains unanswered. Ideally, a measure of pretrial competence­related
abilities should assess capacity.

Standardized Administration and Criterion­Based Scoring

Of the instruments currently available to researchers and clinicians, only the

screening measures and the CAST­MR involve standardized administration and cri­
terion­based scoring. Manuals for the more comprehensive measures, such as the
IFI and CAI, provide a thorough discussion of potential legal content and suggest
the kinds of questions that an examiner might ask; however, the questions actually
posed using these measures will vary from one clinician or researcher to another,
or across defendants examined by the same clinician or researcher. Similarly, these
measures use global rating scales to record the judged degree of impairment in
defendants' responses, although there are no firm objective criteria to anchor the
various points on the rating scales. Consequently, much is left to the discretion of
the person administering these measures, both in terms of the specific questions
posed to any particular defendant and in judging the adequacy of the responses
received. Thus, Grisso's (1992) summary of the state­of­the­art concluded:
MacArthur Adjudicative Competence Study 145

After two decades of research to improve CST [competence to stand trial] evaluations,
CST examiners still ace without any instrument offering standardized administration and
scoring (as contrasted with CST interview guides and subjective ratings) to assess the
domain of CST­related abilities for the general population of defendants who are referred
for CST evaluations. . . . Without an objective measure of the legally relevant abilities,
development of a research foundation for the field of CST assessment will continue to be
limited, (pp. 366­367; see also Grisso, 1996)

In this article we present a new forensic assessment research instrument, the

MacArthur Structured Assessment of the Competencies of Criminal Defendants (Mac­
SAC­CD), that is based on a comprehensive legal theory of competence and con­
tains the desirable structural and content features discussed above (i.e., derived
from legal theory; assesses multiple, discrete competence­related abilities; assesses
capacity; and utilizes standardized administration and criterion­based scoring). We
believe that the MacSAC­CD will enable researchers to conduct the kinds of em­
pirical studies needed to enhance the scientific basis for pretrial competence evalu­
ations. We present the legal framework, psycholegal assessment strategy, instrument
description, and data on the psychometric properties and construct validation of
this new research instrument.

LEGAL FRAMEWORK

The customary formulation of the construct of competence in the criminal

process was articulated by the Supreme Court in Dusky v. United States (1960):
The test will be whether [the defendant] has sufficient present ability to consult with his
lawyer with a reasonable degree of rational understanding—and whether he has a rational
as well as factual understanding of the proceedings against him.

While the legal standard of competence was established by the Supreme Court
more than 30 years ago, the interpretation of the Dusky text by lower courts and
legal commentators has been fluid. The lack of a shared understanding in the legal
community of what competence means in relation to the abilities of criminal de­
fendants is nowhere better illustrated than by the inadequacy of the vocabulary
judges and lawyers use to discuss the topic. An umbrella term, competence to stand
trial, is employed to refer to a diverse array of inquiries. Competence to stand trial
encompasses assessments to determine whether the criminal process should go for­
ward at all (often conducted immediately after arrest, before an attorney has been
appointed), as well as evaluations to determine whether the defendant can partici­
pate in the proceeding (often conducted on the eve of trial, or during the trial
itself).
The term "competence to stand trial" is misleading in a more fundamental
way as well. It implies that competence assessment is primarily directed at trial
participation. Yet it is well known that in virtually all American jurisdictions the
rate of guilty pleas exceeds 90%. Trial participation, therefore, is entirely hypotheti­
cal in the vast bulk of cases. Further, in Godinez v. Moran (1993) the Supreme
Court recently held that the tests for competence to stand trial and competence
to plead guilty are the same. For these reasons, the term adjudicative competence
is more appropriate than competence to stand trial.
 146 Hoge, Bonnie, Poythress, Monahan,Eisenberg,and Feucht­Haviar

Progress in basic research on the adjudicative competence of criminal defen­

dants is contingent on the articulation of a more coherent legal theory of what
competence means in the context of the social purposes that the legal rules for
determining competence are meant to further. Such a legal theory has recently been
elaborated by one of us (Bonnie, 1990, 1992, 1993). Briefly, concern about a de­
fendant's competence rests on three conceptually independent rationales—the dig­
nity of the criminal process requires that the defendant have a basic moral
understanding of the proceedings; the accuracy of adjudications is enhanced by
competent defendant participation; and client competence is & necessary predicate
for autonomous decision making. Given these rationales, the theory posits that the
adjudicative competence of criminal defendants is not unidimensional, but rather
consists of two related but separable components.
The first component of adjudicative competence refers to a foundational com­
petence to assist counsel. The minimum conditions legally required for participating
in one's own defense generally include (1) the capacity to understand the basic
elements of the adversary system, (2) the capacity to relate pertinent information
to counsel concerning the facts of the case,7 and (3) the capacity to appreciate
one's situation as a defendant in a criminal prosecution. The competence to assist
counsel component of adjudicative competence, as we have conceived it, serves the
dignity and accuracy rationales mentioned above, and the law clearly precludes any
adjudication adverse to a defendant who lacks the abilities required to assist in his
or her own defense. This is why we refer to competence to assist counsel as a
foundational component of adjudicative competence.
The second component of adjudicative competence is decisional competence
because a defendant who is competent to assist counsel may not be competent to
make specific decisions regarding the defense of his or her case that are encoun­
tered as the process of criminal adjudication unfolds. Decision making clearly in­
volves cognitive tasks in addition to those required for assisting counsel, but the
abilities required to establish decisional competence have not yet been definitively
established. The Supreme Court's decision in Godinez v. Moran (1993) acknow­
ledged the significance of decisional competence, holding that a defendant's trial
competence and competence to plead guilty should be addressed under a single
standard (the Dusky standard), and that the defendant's decision­making abilities
are encompassed within that standard. However, the Court did not articulate which
abilities are required for decisional competence in criminal adjudication. Existing
case law reflects four possible criteria that may be invoked in determining decisional
competence: (1) the capacity to understand information relevant to the specific de­
cision at issue, (2) the capacity to think rationally about alternative courses of ac­
tion, (3) the capacity to appreciate one's situation as a defendant confronted with
7In some instances, defendants who ate unable to perform this latter function may be deemed competent,
for example, defendants who are amnestic and may not recall any facts about the alleged offense.
Adjudication of competence in these cases necessarily requires assessment of factors extrinsic to the
defendant's cognition in order to determine if an adequate defense can be mounted, such as the
availability and adequacy of other sources of information about the alleged offense (see Wilson v. U.S.,
1968). The assessment included in the MacSAC-CD is intended to measure only the defendant's
capacities.
MacArthur Adjudicative Competence Study 147

a specific legal decision, and (4) the capacity to express a choice among alternatives
(Bonnie, 1993).
This reconceptualization of adjudicative competence has several advantages.
First, it provides a coherent explanatory framework for the sometimes inconsistent
state of the existing law. Second, it helps clarify the issues in areas where the law
is unsettled, such as the precise abilities required for some forms of decisional com­
petence. Third, it reveals the similarities between competencies in criminal defense
and competencies in other legal contexts, such as competence of patients to accept
or refuse mental health treatment (Appelbaum & Grisso, 1988,1995), and thereby
helps to link what have been discrete literatures in mental health law. Finally, be­
cause this approach is derived from an analysis of the social purposes that legal
rules on competence are attempting to further, it provides guidance for operation­
alizing the psycholegal abilities encompassed by the two dimensions of the construct.

PSYCHOLEGAL ASSESSMENT STRATEGY

Several factors influenced our strategy in the development of items for the
MacSAC­CD. Three major considerations were the primacy of content validity and
face validity in the development of forensic assessment instruments, the stand­
ardization of content, and the need to differentially assess capacity with respect to
discrete psycholegal abilities.

Primacy of Content and Face Validity

In his model for the development of forensic assessment instruments, Grisso

(1986) emphasized the importance of content validity and face validity:
There can be no more important criterion for face validity than the opinion of judges and
lawyers as prospective consumers of the results of one's.. .assessment instrument.... It
is possible that no amount of empirical validity will be able to overcome a judicial belief
in the face invalidity of an instrument's dimensions in relation to the legal construct, (p.
42).
Aware of this concern, our research group relied extensively on its own lawyer mem-
ber (Richard Bonnie) and reviews of successive drafts of the MacSAC-CD by mem-
bers of the MacArthur Network with particular legal expertise8 to ensure that the
instrument contained legal content of appropriate breadth and depth. Mental health
professionals with considerable experience and expertise in conducting adjudicative
competence assessments, both within and outside the Network, also reviewed item
content at varying stages of the instrument's development9
8These Network members included: Hon. Shirley Abrahamson, Supreme Court of the State of Wisconsin;
Pamela Hyde, J.D.; Stephen Morse, Professor of Law, University of Pennsylvania; and David Wexler,
Professor of Law, University of Arizona.
9Network members included Thomas Grisso, Ph.D., Paul Appelbaum, M.D., and Loren Roth, M,D.
Outside consultants who reviewed an earlier version of the MacSAC-CD during pilot testing included
Ronald Roesch, Ph.D., and Robert Nicholson, Ph.D.
148 Hoge, Bonnie, Poythress, Monahan, Eisenberg,andFeucht­Haviar

The Standardization of Legal Content

A major hurdle to the development of standardized measures of competence­

related abilities has been the concern that each defendant's criminal charges and
case are unique; this posed the challenge of developing a single set of competence
inquiries that are relevant to diverse cases.
In the MacSAC­CD, we elected to utilize vignette methodology to standardize
much of the item content. Although we recognize that some aspects of competence
assessment do involve inquiries that are necessarily case specific (e.g., defendant's
memory for events related to his or her case; relationship with counsel), the legal
system's expectations regarding competence are, to a significant degree, not tied
to the unique features of a defendant's case. For example, a defendant is expected
to understand the general role of the prosecutor; a defendant is not required, how­
ever, to know the tendencies or idiosyncrasies of the particular prosecutor who may
try his or her case. Similarly, the ability to decide competently whether to plead
guilty may be manifest by the capacity to weigh and consider the costs and benefits
of hypothetical plea offers; it is not necessary that a defendant know what specific
plea agreement he or she will be offered (if any) in order to evaluate his or her
reasoning abilities with respect to that kind of decision. Thus, meaningful measures
may be fashioned to examine some important aspects of defendants* competence­
related abilities in the context of a hypothetical case. The vignette used in the Mac­
SAC­CD is presented below.

Assessment of Discrete Competence­Related Abilities

In addition to the legal theory of competence in the criminal process articu­

lated by Bonnie (1990,1992,1993), and careful reviews by Grisso (1986,1992, 1996)
of existing criminal competence assessment instruments that emphasized the need
for structured assessments with standardized scoring procedures, we also relied on
the psycholegal assessment strategy pioneered by Appelbaum and Grisso (1988,
1995; Grisso & Appelbaum, 1991,1995; Grisso, Appelbaum, Mulvey, & Fletcher,
1995) in their research on the competence­related abilities of civil patients to accept
or refuse mental health treatment. The Grisso­Appelbaum work offered a model
for disaggregating the capacities necessary to make a legally effective decision and
provided measures of these capacities that could easily be adapted from treatment
decision making to the context of criminal adjudication. As we describe in greater
detail below, the general informed consent model, which involves the disclosure of
information and then testing for comprehension, formed the core strategy for the
assessment of capacity to understand legal information.

DESCRIPTION OF THE INSTRUMENT

Reflecting the theory of defendant competence in the criminal process sum­

marized above, our protocol of measures assesses capacities related to competence
MacArthur Adjudicative Competence Study 149

to assist counsel as well as capacities related to decisional competence. Table 1

displays the MacSAC­CD measures; each measure's acronym indicates the primary
legal domain (competence to assist counsel (CAC) or decisional competence (DC))
and the competence­related ability (understanding (U), appreciation (A), or rea­
soning (R)) that it assesses. The measures had a reading index at the eighth grade
level (Grander, 1978). (Note that the measures were read to the defendant by the
research assistant, with the defendant simultaneously reading along with a copy of
the text.) A description of each measure follows.

Component I: Competence to Assist Counsel

Competence to assist counsel, as described earlier, refers to the three minimum

conditions legally required for participating in one's own defense: (1) the capacity
to understand the charges, the nature and purpose of criminal prosecution, and
the basic elements of the adversary system; (2) the capacity to relate pertinent in­
formation to counsel concerning the facts of the case, and (3) the capacity to ap­
preciate one's situation as a defendant in a criminal prosecution (Bonnie, 1993).
Therefore, we created three measures, one to tap each of these capacities.

Competence to Assist Counsel: Understanding

Our measure of the defendant's capacity to understand the pertinent aspects

of a criminal prosecution, Competence to Assist Counsel: Understanding (CAC:U),
encompasses the defendant's understanding of (1) the basic characteristics of crimi­
nal prosecution and defense; (2) the role of the judge and the jury; (3) the nature
of criminal charges; (4) the nature of a guilty plea; and (5) the consequences of a
conviction.
As noted above, one limitation of existing instruments is that they measure
only a defendant's current knowledge, rather than a defendant's capacity to acquire
understanding from his or her defense attorney or other participants in the criminal
process (Grisso, 1986). These instruments do not provide defendants with standard

Table 1. MacSAC­CD Measures by Legal and Functional

Domains
Legal domain
Competence to Decisional
Functional domain assist council competence
Choice DC:C
Understanding CAC:U DCU­PG
DCU­WJ
Appreciation CAC:A DC:A
Reasoning CAC:R DC:R
150 Hogs, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar

disclosures of information about the criminal process and thus are not likely to
measure accurately a defendant's capacity to understand information that may be
disclosed later. The present instrument, in contrast, tests both the knowledge that
defendants bring to the interview, and their capacity to understand new information
that they are given.
The CAC:U employs a novel assessment sequence:
Step 1. The defendant is asked an open­ended question designed to elicit his
or her existing understanding (i.e., "actual understanding") of an area
of relevant information.
Step 2. Correct information, embedded in a vignette, is disclosed to the
defendant.
Step 3. The defendant is asked to demonstrate understanding of the informa­
tion by paraphrasing the disclosure.
Step 4. Finally, the defendant is asked a set of true­false questions based on
the disclosure.
The three response formats have been termed, respectively, Predisclosure (Step
1), Paraphrase (Step 3), and True­false (Step 4). The Predisclosure response meas­
ures the defendant's fund of knowledge before any disclosure is made and thereby
constitutes a measure of actual knowledge. The two response formats that are col­
lected following formal disclosure are measurements of defendants' capacity to un­
derstand disclosed information. We say this because a documented ability to
understand disclosed information is taken, in the aggregate, to be a measure of the
capacity to understand undisclosed information of similar complexity.10
We will illustrate this assessment sequence using a defendant's understanding
of the basic characteristics of criminal prosecution and defense as the target infor­
mation. To provide a plausible legal context for these items, the following vignette
is read to the defendant:
Two men, Fred and Reggie, are playing pool at a bar and get into a fight. Fred hits Reggie
with a pool stick. Reggie falls and hits his head on the floor so hard that he nearly dies.
Based on the story that you just read, I am going to ask you about how the legal
system works.
Step 1: (Predisclosure).
Let's say that Fred gets arrested and charged with a crime. Fred gets a lawyer. Fred's
lawyer is called the attorney for the defense. What is the job of the attorney for the
defense? [Subject response]
There is another lawyer involved in Fred's case who is called the prosecutor. What
is the job of the prosecutor? [Subject response]
Step 2. (Disclosure). The subject then reads the following information.
Fred's lawyer tells Fred how the legal system works. There are two sides. On one side is
Fred's lawyer, who is called the attorney for the defense. He will try to show that Fred
did not commit a crime. Also, as the case goes on, the defense attorney will tell Fred
what his choices are. On the other side is a lawyer called the prosecutor. The prosecutor

10For this reason, we did not include a defendant's Predisclosure responses of actual understanding in
the scoring of this measure, and do not report Predisclosure response scores here. Our primary reason
for soliciting Predisclosure responses was that doing so allowed the evaluation to proceed in a more
comfortable and conversational manner.
MacArthur Adjudicative Competence Study 151

will try to show that Fred did commit a crime and that there is no excuse for what Fred
did.
Step 3. (Paraphrase).
In your own words, tell me what Fred just found out about the legal system. [Subject
response]
Step 4. (True­False). The subject is asked to characterize statements as either true or false
based on what he or she has been told in the disclosure. The statements regarding the
defense attorney were as follows.
1. On one side is Fred's lawyer who is called the prosecutor.
2. As the case goes on, the defense attorney will tell Fred what his choices are.
3. The job of the defense attorney is to decide whether a person is guilty or not guilty.
4. The defense attorney tries to show that a crime has not been committed.

An analogous sequence is used to assess understanding the role of the judge

and the jury, the nature of criminal charges, the nature of a guilty plea, and the
consequences of a conviction.
The scoring range for each item in the Paraphrase and True­False response
formats is 2/1/0. For the Paraphrase format, two points are given for a response
that indicates full understanding of the relevant legal content expressed in the dis­
closure, i.e., both the advocacy and adversarial roles of the attorneys. One point is
given for a response that indicates a partial understanding of the legal content of
the disclosure, i.e., one of the roles of the attorney. Zero points are given for re­
sponses indicating lack of understanding. In the True­False format, two points are
assigned for four correct answers; one point is assigned for three correct answers;
and zero points are scored for two or fewer correct answers on the four questions
given. The Paraphrase and True­IMse response formats are summed to yield a score
on this item of 0­4. The CAC:U consisted of 7 items.

Competence to Assist Counsel­ Reasoning

Our measure of the defendant's capacity to reason, Competence to Assist Coun­

sel: Reasoning (CAC:R), requires defendants to determine which facts among those
presented are more relevant to the defense of a hypothetical criminal case. Defen­
dants are presented with a series of paired items of information and asked to ex­
plain why one item, rather than the other, should be related to an attorney. The
vignette used in this measure continues the story begun in the previous one, but
tells the defendant "a little more about Fred and how Fred got into trouble":
Fred has been working in the home construction business for 10 years. His main job is to
put roofs on houses. One day, after work, Fred decides to go to a baseball game with his
friend, Julie. After the game, they go to a bar. At the bar, Fred has a couple of beers
and he starts to play pool with a guy he just met named Reggie. Suddenly, Fred and
Reggie begin to argue about the pool game and begin to push each other. Then, Fred
swings at Reggie with a pool stick and hits him in the chest. Reggie falls down and hits
his head on the floor pretty hard. It looks like Reggie is hurt very badly. Fred runs to the
phone to call an ambulance. As it turns out, Reggie nearly dies.

Defendants are told that Fred has been charged with a crime (attempted mur-
der). They are then presented, in sequence, six pairs of facts—each on a card that
is also read aloud—and told "suppose that the information on both cards is accu-
152 Hoge, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar

rate." The defendant is asked to choose which item of information would be more
important to relate to an attorney. For example, one pair of cards reads:
Card 1: After Reggie pushed him, Fred thought that he saw Reggie reaching for a knife.
Card 2: Fred picked up a paycheck at work before he picked up Julie to go to a baseball
game.
The defendant is asked:
If Fred's lawyer asks Fred about his reason for fighting with Reggie and for hitting him
with a pool stick, which of these two pieces of information would be more important to
tell the lawyer—Card 1 or Card 2? What are your reasons for picking that card?

The scoring for the open­ended format is as follows: 2 points are awarded for
a correct choice (card 1) accompanied by an appropriate explanation (i.e., the per­
ception of the knife suggests self­defense or justifies the use of greater force—hit­
ting Reggie with the pool cue), 1 point for a correct choice alone, and zero points
for an incorrect choice. The CAC:R consists of 6 items.

Competence to Assist Counsel: Appreciation

Our measure of the capacity to appreciate one's situation as a defendant in a

criminal prosecution, Competence to Assist Counsel: Appreciation (CAC:A) assesses
the defendant's ability to appreciate his or her own legal predicament. The CAC:A
focuses on the defendant's own legal case, rather than a hypothetical vignette. Our
primary goal is to ascertain whether the defendant has delusional beliefs or other
symptoms that may affect his or her ability to appreciate the nature and seriousness
of his or her situation or role as a defendant in a criminal prosecution. The CAC:A
consists of 6 items assessing beliefs relating to (1) the criminal charges; (2) the
likelihood of conviction; (3) the impartiality of the adjudication process; (4) the
possible helpfulness of the defense attorney; (5) the possible benefits of disclosing
information to the defense attorney; and (6) the severity of punishment.
To illustrate, beliefs regarding the likelihood of conviction are assessed by tell-
ing the defendant the following:
Some people who get into trouble with the law are found guilty of a crime. Now I want
you to look at this card. It has five answers on it to indicate how likely you think it is
that you will be found guilty. I want you to pick the answer that comes closest to what
you think.
The five response options, which were read aloud to the defendant, were as
follows:
A. I definitely have a lower chance of being found guilty than most people charged with
[the offense that the defendant has been charged with].
B. I probably have a lower chance of being found guilty than most people charged with
[the offense that the defendant has been charged with].
C. I probably have about the same chance of being found guilty as others charged with
[the offense that the defendant has been charged with].
D. I probably have a higher chance of being found guilty than other people charged with
[the offense that the defendant has been charged with].
E. I definitely have a higher chance of being found guilty than other people charged with
[the offense that the defendant has been charged with].
MacArthur Adjudicative Competence Study 153

The subject is then asked to explain the reason for his response. The reason
given for each question is recorded and scored for facial plausibility: clearly plau­
sible, 2 points; of questionable plausibility, 1 point; and clearly implausible, zero
points. A "clearly plausible" score is assigned to a reason that rationally may explain
the defendant's initial response. For example, a defendant who indicated greater
likelihood of being found guilty (choice D or E) might indicate "The evidence
against me is very strong," or "I am Black and the system is prejudiced against
Blacks," or "My attorney is not interested in me." A "clearly implausible" score is
assigned to a reason that is irrational on its face; for example, a choice (A or B)
indicating less likelihood of being found guilty might be explained by "At the trial
the judge will hear the same voices that I hear and they will tell him to acquit
me."
If the plausibility of the defendant's reasoning is not clear (score = 1), then
a question is framed that denies the premise underlying the reason given for the
initial response and a second response and reason are elicited. For example, if a
defendant answered E "because I was being videotaped," the interviewer would
investigate whether a paranoid sense of being watched may be affecting the defen­
dant's perception of the case. Thus, the premise of the first reason might be negated
by asking "Suppose you were not being videotaped, then what would your choice
be?" The subject is then asked the reason for the second response.
The score for each item is derived as follows: If the score given to the initial
reason is based on a clearly plausible or clearly implausible reason (2 and zero
points, respectively), this score is assigned. When a defendant's initial reason is of
questionable plausibility (1 point), the plausibility scored assigned to the reason
given for the second response is the item score.

Component II: Decisional Competence

Decisional competence, as described earlier, refers to the ability to make the

specific decisions that are encountered in the process of criminal adjudication. Sev­
eral capacities may be relevant, depending on the governing legal standards: (1)
the capacity to understand information relevant to the specific decision at issue,
(2) the capacity to weigh and consider information to reach a decision; (3) the
capacity to appreciate one's situation as a defendant confronted with a specific legal
decision, and (4) the capacity to express a choice among alternatives (see Appel­
baum & Grisso, 1988). Because there is no consensus concerning the correct stand­
ard for decisional competence, we have constructed measures for all the listed
capacities. In doing so, we do not mean to claim that clinicians or judges must
consider each of these capacities in every determination of decisional competence;
the pertinent capacities depend upon the applicable legal standard. As noted ear­
lier, however, the law governing decisional competence is now unclear. Ultimately,
policy judgments regarding the necessary requirements for decisional competence
should turn, in part, on empirical data regarding the distribution of these capacities
among criminal defendants. As noted above, these data do not now exist.
154 Hoge, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar

The number and the nature of decisions that a defendant must make vary
from case to case. Previous epidemiological studies (Hoge, Bonnie, Poythress, &
Monahan, 1992; Poythress, Bonnie, Hoge, Monahan, & Oberlander, 1994) have in­
dicated that the two most commonly encountered decisions in criminal cases are
the decision whether or not to plead guilty, and the decision whether or not to
waive a jury and request a bench trial (explicitly discussed with the defendant in
87% of the cases in which the defendant pleads not guilty). Therefore, we con­
structed measures pertaining to these two decisions.

Decisional Competence: Understanding

Our measure of the defendant's capacity to understand information specific to

the decision whether or not to waive the constitutional right to trial by pleading
guilty is termed Decisional Competence: Understanding—Pleading Guilty (DGU­PG).
It consists of five items focusing on the defendant's understanding of (1) the ad­
mission of criminal conduct; (2) the waiver of legal rights; (3) plausible reasons for
pleading guilty rather than going to trial; (4) plausible reasons for going to trial
rather than pleading guilty, and (5) the prerogative to decide how to plead.
Similarly, our measure of the defendant's capacity to understand information
specific to the decision of whether or not to waive the constitutional right to a trial
by jury and instead request a bench trial is called Decisional Competence: Under­
standing—Waiving Jury (DC:U­WJ). It also consists of six items focusing on the de­
fendant's understanding of (1) the distinction between a trial by judge and a trial
by jury (two items); (2) the constitutional right to a jury trial and prerogative to
waive this right; (3) the participation by the defense in jury selection; (4) plausible
reasons for choosing a trial by judge and (5) plausible reasons for choosing a trial
by jury.
The same four­stage sequence used in the Competence to Assist Counsel: Un­
derstanding (CAC:U) measure, described above, is used in both the DC:U­PG and
DGU­WJ measures. That is, (1) the defendant is asked an open­ended question
designed to elicit his or her present understanding regarding the decision in ques­
tion; (2) correct information, embedded in a vignette, is disclosed to the defendant;
(3) the defendant is asked to demonstrate understanding of the information by
paraphrasing the disclosure; and (4) the defendant is asked a set of true­false ques­
tions based on the disclosure.

Decisional Competence: Reasoning

Our measure of the capacity to think rationally about alternative courses of

action in making decisions about criminal defense, Decisional Competence: Reason­
ing (DC:R), is modeled closely on the measure developed by Grisso et al. (1995,
pp. 134­136) to assess capacity to think rationally about mental health treatment.
The DC:R consists of 13 items addressing the defendant's capacity to (1) request
information needed to make a decision; (2) to conceive the primary legal effects
of alternatives; (3) to conceive the personal consequences of the alternative out­
MacArthur Adjudicative Competence Study 1SS

comes; (4) to compare alternative choices; (5) to assign relative values to alterna­
tives in a consistent way; (6) to think transitively (i.e., if A > B and B > C, then
A > C); and (7) to think using probabilities.
Assessment of several of the capacities (1­4 enumerated above) requires sub­
jects to respond to questions based on further elaboration of the vignette introduced
earlier. The defendant is told that "Fred's lawyer tells Fred that there are two ways
to plead. There are good and bad things about both. Fred will have to choose."
One choice is that Fred could plead not guilty and have a trial. The chances are Ugh that
Fred would be convicted of a crime. If Fred is convicted of attempted murder, it is likely
that he would be sentenced to 10 years in prison. Or, if Fred is convicted of simple assault,
it is likely that he would be placed on probation.
The other choice is that Fred could plead guilty. The prosecutor has talked with Fred's
lawyer and has made an offer: If Fred will plead guilty to the less serious charge of simple
assault, the charge of attempted murder will be dropped. Fred will get a sentence of six
months in jail. There would be no chance for Fred to be found innocent.
Now, you have heard about Fred's problem and about Fred's choices for dealing with
his problem. Let's say Fred is your friend and he wants you to give him some advice.

To assess the capacity to "request information needed to make a decision," for

example, the defendant would be asked: "What else would you want to know before
you advise Fred?"
Requests for new information, not previously disclosed but plausibly relevant,
result in a score of 2 points. Requests for previously disclosed, relevant information
are scored 1 point Zero points are awarded in the absence of a request for infor­
mation or to requests for irrelevant information.
In a similar fashion, subjects are asked questions to assess their abilities to
conceive the legal effects of alternative choices, to conceive the personal conse­
quences, and to compare the alternatives. Scoring follows the pattern described
above.
The remaining items of the DC:R do not relate to the vignette. The ability to
assign relative values to alternatives is measured by presenting subjects with a series
of activities such as "read a book" and "go to a movie," and asking them to indicate
their preferences. These activities are presented in multiple different pairings. The
scoring is based on subjects' ability to consistently rank their preferences across
various pairings. The abilities to think transitively and to use probabilities are as­
sessed by presentation of short questions with a multiple­choice response format.

Decisional Competence: Appreciation

Our measure of appreciation in decision making is the Decisional Competence:
Appreciation (DC:A). Like the CAC:A, DC:A assesses the defendant's ability to
appreciate his or her own situation. Our goal is to uncover whether the defendant's
ability to make the relevant decision in a self­interested manner may be affected
by delusional beliefs or other symptoms of mental disorder.
The DC:A is comprised of two items, one keyed to the decision whether to
plead guilty and one to the decision whether to request a waiver of jury. The struc­
ture of these items is similar to the CAC:A items, described above. First, the subject
is asked to make a probabilistic judgment about his or her decision to plead guilty
156 Hoge, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar

or waive a jury trial, on a scale ranging from "I definitely would [plead guilty or
have a jury]" to "I definitely would [plead not guilty or not have a jury]." The
subject is then asked to explain the reasons for the response. As on the CAC:A,
the reasons given are scored for facial plausibility: clearly plausible, 2 points; of
questionable plausibility, 1 point; and clearly implausible, zero points. If the plau­
sibility score is 0 or 2, this is the item score. As on the CAC:A, when the initial
reason given yields a plausibility score = 1, a question is posed that denies the
premise underlying the initial reason given and a second response and reason are
elicited and the plausibility score assigned to the second reason is the item score.

Decisional Competence: Choice

Our index of capacity to evidence a choice, Decisional Competence: Choice

(DC:C), merely requires that the subject be able to express a choice between "going
to trial" and "pleading guilty" when called upon to "give Fred advice" in the above
vignette. Subjects who choose either alternative score two points; subjects who are
unable to choose receive zero points.

PSYCHOMETRIC PROPERTIES OF THE MacSAC­CD

We have described elsewhere findings from a pilot study that used an extreme
groups design (Hoge et al., 1996). One group consisted of clearly incompetent de­
fendants recently admitted to state forensic hospitals for competence restoration;
the polar group consisted of pretrial defendants whose competence was not doubted
by their public defender. In the present investigation, we more thoroughly and more
rigorously examined the psychometric integrity of the MacSAC­CD. The number
and types of research participants were expanded to include a broader sample of
criminal defendants, and were not limited to just those presumed to lie at the ex­
tremes. Defendants hospitalized as incompetent, defendants being treated for men­
tal disorder in jail (but not referred for competence restoration), and pretrial
detainees from the general jail population were recruited without prescreening re­
garding level of competence or psychopathology to ensure that the full range of
competence and psychopathology found in defendant populations was represented
in the study. We present analyses describing three psychometric properties of the
MacSAC­CD measures: interscorer reliability, internal consistency, and correlations
among the measures.

VALIDATION OF THE MacSAC­CD

The current version of the Standards for Educational and Psychological Tests
(American Psychological Association, 1985) emphasizes that neat distinctions
among the separate forms of validity often mentioned in methodology texts (e.g.,
content validity, predictive validity) are overstated and that all are best seen as
MacArthur Adjudicative Competence Study 157

variants of "construct validity" (see Grisso, 1986, p. 48). This study attempted to
provide construct validation for the MacSAC­CD in five ways. First, we compared
the performance of three groups: (a) defendants with mental disorder who were
identified as incompetent to stand trial and were undergoing inpatient treatment
for restoration of competence (the "Hospitalized Incompetent" or HI group); (b)
defendants with mental disorder who were not identified as incompetent to stand
trial but who were being treated for their mental disorder in jail (the "Jail Treated"
or JT group): and (c) defendants from the general jail population who were not
identified as incompetent to stand trial and who were not receiving mental health
treatment. Because we did not screen the randomly chosen defendants in the last
group either for competence or for mental disorder, we refer to this group as the
"Jail Unscreened" or JU group. We hypothesized that performance would differ
among the three groups, with the JU group scoring highest (i.e., most competent),
and the HI group lowest. The JT group was predicted to Ml between these two
extremes: The scores of JT subjects were expected to show some impairments in
competence­related abilities because of mental illness, but not so extensive as to
have resulted in questions of incompetence having been formally raised.
Second, within the group of defendants with mental disorder who were iden­
tified as incompetent to stand trial and who received inpatient treatment for res­
toration of competence (the HI group), we compared defendants' performance
shortly after they had been found to be incompetent—when treatment to restore
competence was beginning—with the subsequent performance of the same defen­
dants at the time they were found to be restored to competence and returned to
court for the adjudication of their criminal cases. The MacSAC­CD should be sen­
sitive to changes in the capacities of mentally disordered defendants that occur as
the result of successful treatment to restore competence.
Third, again within the HI group, we compared performance 00 the MacSAC­
CD with clinical judgments of competence to stand trial provided by the defendants'
treating forensic clinicians. The MacSAC­CD should produce scores that correlate
positively and modestly with other measures of the same construct, and clinical
judgment is the most frequently used "measure" of competence to have one's case
adjudicated. However, some defendants in the HI group may have been judged
incompetent on the basis of case specific information or other facets of competence
not assessed by the MacSAC­CD. This, along with some restriction in the range of
scores for a hospitalized incompetent group, should result in correlations of only
modest magnitude.
Fourth, across all groups, we compared performance on the MacSAC­CD with
performance on measures of psychopathology and cognitive functioning. Much the­
ory on competence to stand trial has postulated, and much research on competence
to stand trial has found, that competence determinations are negatively related to
psychotic symptoms and to low intelligence (e.g., Grisso, 1992; Nicholson & Kugler,
1991; Roesch & Golding, 1980). Similar findings have been reported in studies of
competence to consent to mental health treatment (Grisso & Appelbaum, 1995).
The MacSAC­CD should produce scores that relate to indices of psychopathology
and cognitive functioning in a manner consistent with established theoretical and
empirical understandings of competence.
158 Hoge, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar

Finally, across all groups, we compared performance on the MacSAC­CD with

a measure of negative, cynical attitudes toward the criminal justice system. Easting
instruments to assess competence to stand trial have been criticized as mistaking
a defendant's exponentially based cynicism and distrust of police, courts, and law­
yers for impairments in competence­related abilities (Brakel, 1974). The MacSAC­
CD should produce scores that do not substantially correlate with a defendant's
skepticism toward the criminal justice system.
But for the injunction of the APA Standards (1985) not to reify subtypes of
validity, we would have termed these five approaches, respectively, criterion validity,
concurrent validity, convergent validity, theoretical validity, and divergent validity
(Foster & Cone, 1995)

METHOD

Participants

Data were collected from three groups of criminal defendants in Virginia and
Florida.11 Table 2 shows the contributions of participants by the two research sites.
All subjects were male12 and between the ages of 18 and 65 and were charged with
criminal offenses. Defendants with diagnoses of mental retardation or serious or­
ganic disorders, e.g., dementia, were excluded from the sample. Subjects who had
substantial cognitive impairment as indicated by a Verbal Cognitive Functioning
(VCF) score under 60 (see below) would have been excluded. However, no subject
in any group had a VCF score below 60, and only 8 subjects had a VCF score
below 70.
Among the two groups of consecutively admitted hospitalized defendants who
were approached for participation in the study, the refusal rate was 12% (9% in
Virginia and 16% in Florida). In the consecutively admitted jailed sample, the re­
fusal rate was 33.5% (33% in Virginia and 34% in Florida).

Hospitalized Incompetent Group

The Hospitalized Incompetent group (HI,n = 159) was recruited from criminal
defendants who had been committed to public sector forensic inpatient units in
Virginia and Florida for restoration of competence to stand trial. Subjects were
interviewed within two weeks of admission (M = 6.1 days, SD = 2.8 days).
11The states differed in two minor ways. In Florida, commitment for restoration of competence required
a formal judicial hearing. In Virginia, commitment for restoration of competence could be
accomplished via a formal judicial hearing, or via a less formal Criminal Temporary Detention Order.
In Florida, jailed defendants were randomly selected from among those represented by a public
defender. In Virginia, there was no such constraint on the random sampling of jailed defendants (but
fully 86% of the jailed defendants sampled from Virginia were nevertheless publicly represented).
12An independent study of competence­related abilities among female defendants has also been
conducted. A manuscript reporting the results of this study is in preparation.
MacArthur Adjudicative Competence Study 159

Table 2. Site and Demographic Characteristics of Hospitalized and Jailed Groups

Hospital Jail Jail
incompetent treated unscreened
Variables n = 159 n = 113 n = 94
Site
Virginia 84 58 41
Florida 75 55 53
Total 159 113 94
Age
M 34.2 32.4 29.9
SD 8.8 8.8 7.9
Percent non­White 37 32 28
Socioeconomic status (%)
Professional/managerial 2 4 1
Administrative/clerical 8 10 11
Skilled/semi­skilled manual labor 17 30 50
Unskilled/unemployed 73 55 38

Jail Treated Group

The Jail Heated group (JT, n = 113) was recruited from criminal defendants
in three jails, two in Florida (n = 55), the third in Virginia (n = 58). Subjects were
defendants who had been identified by jail personnel as mentally disordered and
who were currently receiving mental health treatment. The defendants in this group,
while mentally disordered, had not been identified as incompetent.

Jail Unscreened Group

The Jail Unscreened group (JU, n = 94) was recruited from the same three
jail facilities that were used to recruit the JT sample. The JU subjects were recruited
from among the pre­trial jail detainees who had not been identified by jail personnel
as mentally disordered, and who were not currently receiving mental health treat­
ment.

Measures of Predictor Variables

Chart Data. A combination of hospital charts and jail records provided infor­
mation on subjects' age, race, diagnosis, admission date, history of psychiatric treat­
ment, current medications, history of competence evaluations, and criminal history
including current charges.
Background Interview Data, A brief structured interview, modified from one
used in previous research (Grisso & Appelbaum, 1995), was used to obtain infor­
mation from all participants regarding psychiatric history, criminal history, previous
competence evaluations, contacts with defense attorneys, and current and past oc­
cupations.
160 Hoge, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar

Brief Psychiatric Rating Scale—Anchored. The BPRS (Overall & Gorham, 1962)
is a measure of the presence and severity of psychopathology based on an interview.
The anchored version of the BPRS used in this study has been demonstrated to
be a reliable measure of psychopathology (BPRS­A; Woerner, Mannuzza, & Kane,
1988). The severity of symptoms or signs of mental illness is rated on a 7­point
anchored Likert scale. The 18 items can be summed to provide a global measure
of the severity of the psychiatric condition. In addition, we calculated four subscale
scores that provide measures of psychoticism, depression, withdrawal, and hostility
(Overall & Porterfield, 1963).
Verbal Cognitive Functioning (VCF). Three subtests of the Wechsler Adult In­
telligence Scale—Revised (WAIS­R; Vocabulary, Similarities, Digit Span) were used
to generate a measure of subjects' cognitive capacities at the time of the research
interview. The VCF score was calculated using the WAIS­R norms to convert raw
scores to scale scores, summing the three scale scores, multiplying by 2, and using
the WAIS­R age­normed tables to convert the calculation to the equivalent of a
pro­rated Verbal IQ score. Research indicates that this procedure yields scores that
are highly correlated (r > .90) with WAIS­R Verbal IQ (see Grisso & Appelbaum,
1995, p. 154). However, among the acutely mentally disordered, cognitive functions
are likely to be impaired by the underlying disorder; thus, the prorated score is
better conceived as an index of current verbal cognitive functioning and is not pre­
sumed to be a measure of baseline intellectual functioning.
Clinical Judgment Regarding Competence. At the time of the research interview,
the responsible inpatient forensic mental health professional for each hospitalized
defendant was asked to rate subjects' adjudicative competence on a 6­point Likert
scale, which ranged from "clearly incompetent" (1) to "clearly competent" (6).
Perceived Criminal Injustice Scale (PCIS). The PCIS includes six items that as­
sess subjects' beliefs about the fairness of lawyers, police, juries, and judges. Each
item is scored on a 5­point Likert scale, which ranged from "strongly agree" (+2)
to "strongly disagree" (­2). The PCIS used in this study was adapted from a 10­item
scale developed by Hagan and Albonetti (1982). An example of an item is "Police,
lawyers, and judges treat people who commit crimes the same, whether they are
rich or poor." A response of "strongly disagree" (­2) would indicate perceived in­
justice. The internal consistency of our adaptation of the PCIS was a = ,66.

Procedure

Hospitalized Incompetent Group

Research assistants identified all eligible admissions to the forensic inpatient

units. As soon as possible after admission, the responsible clinicians were asked for
permission to approach patients to request study participation. Researchers ap­
proached the patients and explained the nature of the research study and described
the interview process, the length of time the interview would take, and the confi­
dentiality of their responses. Patients were offered $10 to participate and written
informed consent to the research study was obtained. Interviews were conducted
MacArthur Adjudicative Competence Study 161

in private offices or interview rooms. For each patient recruited into the study, a
treating clinician was asked to provide a clinical rating of competence.
Subjects were administered the VCF and the remaining instruments were ad­
ministered in the following order, the brief background interview, PCIS, CAC:U,
DC:U­PG, DC:A­PG, DC:U­WJ, DC:A­WJ, CAC:R, DC:R, CAGA, and BPRS.

Jail Treated and Jail Unscreened Groups

Pretrial jailed defendants were randomly selected and asked to participate in

the research study. The procedures were identical to those described above, includ­
ing subject payment.

RESULTS

We first describe our research samples, and then present data on the psy­
chometric properties of the MacSAC­CD. Finally, we address the construct validity
of the instrument.

Sample Description

The demographic characteristics of each group are summarized in Table 2. HI

subjects were more likely than either JT or JU subjects to be unemployed or un­
skilled, x2(52, n = 358) = 72.14, p < .01 and HI subjects were older than JU
subjects, F(2, 362) = 7.28, p < .001. There were no significant differences among
the groups in education or race.
There were few sites differences. In the JT group, Virginia subjects were more
likely than Florida subjects to be unskilled or unemployed (65% vs. 45%), x2(6, n
= 110) = 14.67, p < .02. Virginia subjects had more education (M = 12.3 vs. 11.2
years; t = 107) = 2.73, p < .01). In the HI group, Virginia subjects were more
likely than Florida subjects to be unskilled or unemployed (98% vs. 46%), x2(7, n
= 82) = 14.57, p < .05. There were no site differences on race or age.
Table 3 summarizes the group characteristics on background mental health and
criminal justice variables. The HI group is comprised largely of defendants with a
diagnosis of schizophrenia, and the JT group of defendants with a diagnosis of
affective disorder. While defendants in the JU group did not have hospital charts
from which diagnoses could be extracted, 9 of the 94 subjects in this group had a
past diagnosis recorded in jail records, but had not been identified by jail personnel
as mentally disordered or in need of treatment services at the time of data collec­
tion. Of these 9 defendants, 5 had a past diagnosis of substance abuse, and one
each had a past diagnosis of schizophrenia, psychotic disorder (unspecified), devel­
opmental disorder (unspecified), and antisocial personality disorder.
HI and JT subjects were more likely than JU subjects to be receiving psychiatric
medications at the time of the interview, F(2, 365) = 215.98, p < .001. Consistent
with their diagnoses, HI subjects were more likely than JT or JU subjects to be
162 Hoge, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar

Table 3. Group Characteristics on Mental Health and Criminal Justice Variables

Hospital Jail Jail
incompetent treated unscreened
Variables n = 159 n = 113 n = 94
Diagnoses (%)
Schizophrenia 65 24 2
Affective disorder 28 59 0
Other 7 17 7
Medicated (%) 83 87 0
Prior MH treatment (%)
Inpatient 79 64 10
Outpatient 77 64 15
Prior competence evaluations (%) 40 22 3
Age at first arrest
M 24.06 23.40 21.43
SD 6.95 8.2 6.2
Prior arrests, No.
M 1.7 1.7 1.8
SD 0.3 0.4 0.3
Ever in prison (%) 33 42 48
Current felony charges (%) 73 76 48
Prior felony convictions (%) 44 62 67

prescribed antipsychotic medication, and JT subjects were more likely to be pre­

scribed antidepressants, x2(6, n = 232) = 62.20, p < .001. JU subjects were less
likely than the HI or JT subjects to report prior inpatient, x2(2, n = 365) = 119.45,
p < .001, or outpatient mental health treatment, x2(2, n = 365) = 97.03, p < .001.
Subjects in the HI group were more likely than members of the other groups to
report evaluations of adjudicative competence prior to the adjudication of their
present case, x2(2, n = 363) = 42.08, p< .001.
Subjects in the JU group were less likely to be facing a current felony charge
than the HI and JT groups, x2(2, n = 364) = 22.12, p < .001. Subjects in the HI
group were less likely to have been convicted of a felony than the other groups,
x2(2, n = 360) = 14.75, p < .001. There were no group differences observed for
number of previous felony charges, previous jail and prison terms, age at first felony
arrest, number of previous misdemeanor charges, age at first felony conviction, age
at first misdemeanor conviction, number of times jailed, or number of times impris­
oned. Group characteristics on measures of psychopathology and cognitive function­
ing are summarized in Table 4. The JU group, as expected, had significantly lower
levels of psychopathology, as measured by the BPRS, than the other groups, F(2,
357) = 50.98, p < .001. There were no differences between the HI and the JT on
total BPRS scores. Subjects in the JU group had lower scores on the BPRS psy­
MacArthur Adjucadicative Competence Study 163

Table 4. Group Characteristics on Psychopathology and Cognitive

Functioning Measures
Hospital Jail Jail
incompetent treated unscreened
Variables n = 159 n = 113 n = 94
Brief Psychiatric
Rating Scale
(BPRS)
Total
M 37.02 38.95 28.72
SD 7.6 8.6 3.38
Psychoticism
M 6.60 5.71 3.74
SD 2.8 3.1 1.05
Hostility
M 6.58 5.79 4.41
SD 2.7 2.6 1.48
Depression
M 758 10.84 7.57
SD 2.71 2.64 3.18
Withdrawal
M 5.89 6.80 5.11
SD 2.5 3.0 2.54
VCF
M 87.97 92.37 90.39
SD 11.5 14.3 10.95

choticism subscale than did subjects in either the JT or HI groups, with subjects in
the JT group having lower scores than subjects in the HI group, F(2, 358) = 36.87,
p < .001. Comparison of scores on the BPRS hostility subscale produced the same
pattern of findings, F(2,358) = 23.38, p < .001. Conversely, subjects in the JT group
scored highest on the depression, F(2, 358) = 39.96, p < .001, and withdrawal
subscales, F(2, 357) = 9.96, p < .001. Subjects in the HI group bad a lower VCF
score than subjects in the JT and JU groups, F(2, 363) = 4.32, p < .01.

Psychometric Properties

We present analyses describing three psychometric properties of the MacSAC­

CD measures: interscorer reliability, internal consistency, and correlations among
the measures.

Interscorer Reliability
Nine research assistants were employed at four data collection sites over the
course of the study. To examine interscorer reliability, 40 completed protocols, 10
164 Hoge, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar

from each site, were selected. At sites with more than one research assistant, care
was taken to randomly select from the completed protocols of each RA to ensure
that each was represented in the sample. One particular senior research staff mem­
ber from the University of Virginia who had worked most closely with the research
group in the development of the measures and scoring criteria, and who was the
most experienced in administering and scoring the protocol, served as the "master
scorer" for the 30 protocols that had been administered by the other RAs.
Table 5 provides K coefficients and percent absolute agreement between scores
assigned by RAs and the "master scorer." K'S for the understanding and reasoning
measures range from .60 to .75, indicating satisfactory interscorer reliability. Percent
agreement between scorers for these measures ranges from 75% to 83% and sug­
gests satisfactory interscorer reliability. Results for the appreciation measures are
less clear. K'S for the CACA and DC:A were .48 and .36, respectively, which are
not particularly encouraging. However, the distribution of scores on these measures
was extremely skewed and contained little variance, which can depress K values.
Percent agreement for CAC:A and DC:A scoring was quite high, 88% and 75%,
respectively, and suggests that trained raters reach satisfactory levels of agreement
in scoring these measures.

Internal Consistency

Coefficient alpha and item­scale correlations were calculated for each of the
MacSAC­CD measures except DC:A. (Because DC:A consisted of only two Items,
a correlation coefficient was computed, r = .40, p < .001.) These are presented in
Table 6.
The highest ot's were obtained on the measures of understanding (CAC:U,
DCU­PG, DC:U­WJ). All a's obtained equal or exceed the generally accepted
value for research measures (r > .70; Nunnally, 1978, p. 245) but are slightly below

Table 5. Interscorer Reliability: K Coefficients and

Percent Agreement for MacSAC­CD Measures
Index of agreement with
"master scorer"
Ability/measure K Percent agreement
Understanding
CAC:U .60 75
DCU­PG .74 83
DC:U­WJ .75 83
Reasoning
CAC:R .60 82
DC:R .60 76
Appreciation
CAC:A .48 88
DC:A .36 75
MacArthur Adjudicative Competence Study 165

Table 6. MacSAC­CD: Internal Consistency

Measures
Range (and average)
of item­scale
a correlations
Measure
CAC:U .87 .46­.60 (.53)
CAC:R .68 30­.51 (.41)
CAC:A .74 .33­59 (.48)
DC:U­PG .88 .37­.65 (.54)
DC:U­WJ .88 .27­.64 (.50)
DC:R .71 .18­.61 (.32)

the recommended minimum (r > .90; Nunnally, 1978, p. 246) for decision making
in applied settings. For each measure, coefficient a could not be improved by the
deletion of any items.
Findings regarding the item­scale correlations were similarly encouraging.
Item­total correlations > .30 are usually considered good (Nunnally, 1978, p. 263),
and most items on all measures meet or exceed this level.

Correlations Among Measures

Correlations were computed between scores for each of the MacSAC­CD

measures. Although it was anticipated that these measures would correlate posi­
tively with one another, certain patterns of correlations would be expected if the
measures were tapping into distinct psycholegal abilities or constructs. Ideally, mea­
sures that tap one ability (e.g., "understanding") would correlate more highly with
one another than with measures that theoretically tap a different ability (e.g., "ap­
preciation"). Table 7 displays the correlations between MacSAC­CD measures.

Table 7. Correlations Between MacSAC­CD Measuresa

CAC:U CAC:R CAC:A DC:U­PG DC:U­WJ DC:R DC:A
CAC:U 1.00

CAC:R .61 1.00

CAC:A .36 .38 1.00

DC:U­PG .84 .59 .40 1.00

DC:U­WJ .84 .60 .41 .85 1.00

DOR .72 .63 .38 .78 .77

1.00
DC:A .54 .45 .45 .60 .57 .50 1.00
aAll correlations are significant at p < .01.
166 Hoge, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar

The pattern of correlations is generally consistent with the notion that differ­
ent competence­related abilities are being tapped by the MacSAC­CD measures.
All correlations between pairs of measures are significant. The three "under­
standing" measures (CAC:U, DC:U­PG, DC:U­WJ) correlate more highly with
one another (.84, .84, .85) than with any other measure. CAC:A, a measure of
"appreciation," correlates more highly with the other "appreciation" measure
(DC:A) (.45) than it does with any other measure; this is not true, however, for
DC:A, which has higher correlations with several other measures than with
CAC:A. Similarly, the competence to assist counsel reasoning measure, CAC:R,
is most highly correlated (.63) with DC:R, the decisional competence reasoning
measure; however, DC:R has higher correlations with the understanding measures
(.72­J8) than with CAC:R,

Construct Validity
Validity Question 1: Does Performance on the Research Instrument Reflect
Expected Differences Among Groups that Differ on Competence Status?

The three groups were compared in one overall multivariate analysis of co­
variance that controlled for the only four demographic or background variables that
significantly correlated with performance on the MacSAC­CD measures (age, SES,
prior contact with the criminal justice system, and prior contact with the attorney).13
The groups were significantly different, F(2,313) = 4.46, p < .001. Special contrasts
to test specific group differences were performed and are summarized below.
Competence to Assist Counsel. As shown in Table 8, the HI group scored sig­
nificantly lower than the JT and JU groups on the CAC:U ("Understanding") meas­
ure, f(l,n = 313) = ­3.92, p < .001. The JT group mean scores were intermediate
between the HI and JU groups' scores, Table 8 also shows that the same pattern
(HI < JT, JU) also obtained for the CAC:R ("Reasoning") scores, t (313) = ­4.70,
p < .001, and the CAC:A ("Appreciation") scores, t(313) = ­2.98, p < .005.
To illustrate the distributions of scores for the three groups on each of our
measures, Figure 1 shows the distributions of CAC:U scores. As can be seen, there
is significant overlap among the groups' performance. However, the HI group scores
are dearly skewed toward the lower end of the CAC:U range. An overlap among
the groups' performance similar to that shown in Figure 1 for CAC:U is also found
for each of the other measures.
Decisional Competence. Table 8 shows that the HI group obtained lower scores
than the JT and JU groups on both the DC:U­PG ("Understanding—Pleading
Guilty"), t(313) = ­6.66, p < .001, and the DC:U­WJ ("Understanding­Waiving
a Jury") measures, t(l, 313) = ­5.80, p < .001. There were no statistical differences
between the scores of the JT and JU groups. The same pattern holds true for the
DC:R ("Reasoning") measure, t(313) = ­5.51, p < .001, and the DC:A ("Appre­
ciation") measure, t(313) = ­5.24, p < .001.
13Correlation
matrices of demographic and background variables and performance on each of the
MacSAC-CD measures for each of the three groups are available from the authors upon request.
MacArthur Adjudicative Competence Study 167

Table 3. Performance on Research Measures by Group

Hospital Jail Jail
incompetent treated unscreened
Measures n = 159 n = 113 n = 94
Competence to assist council
CAC:U
M 18.92 21.20 22.60
SO 5.54 4.30 3.30
CAC:R
M 8.61 10.05 10.33
SD 2.93 1.98 1.68
CAC:A
M 10.45 11.10 11.55
SD 2.10 1.80 1.06

Decisional competence
DC:U­PG
M 13.30 16.64 17.26
SD 4.73 3.47 2.63
DC:U­WJ
M 16.77 20.04 20.49
SD 5.06 3.95 2.81
DC:R
M 11.74 13.96 14.05
SD 3.17 2.38 2.19
DC:A
M 3.31 3.80 3.92
SD 1.09 .60 31

FinaUy, the DC:C ("Choice") measure (not shown in Table 8) provided an in­
dex of defendants' capacity to choose among defense options. In the total sample
of 366 defendants, only 10 (2.7%) were unable to make a choice (i.e., scored a 0
on this item). Of these, 9 were in the HI group (5.6%) and 1 was to the JT group
(0.9%).

Validity Question 2: Does Performance on the Research Instrument

Reflect Changes in Competence Status?
A subgroup of the Hospitalized Incompetent (HI) group was created consisting
of those members of the HI group who had been, in the opinion of the responsible
mental health professional, restored to competence and were waiting for the formal
adjudication of their criminal cases to resume (n = 97). Defendants in this restored
subgroup were reevaluated on our research measures before being returned to
court. In order to be eligible for inclusion in this restored subgroup, HI subjects
had to have been treated for at least one week after the initial research interview
(actual M interval= 38.1 days, SD = 21.9). Attrition in this sample, i.e., from the
159 in the HI group to the 97 in the restored subgroup, was primarily due to rapid
discharges within the one­week interval following the initial research interview.
Rapid discharges commonly occurred when the responsible hospital clinicians dis­
168 Hoge, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar
MacArthur Adjudicative Competence Study 169

agreed with the original determination of incompetence. In some cases, defendants

appeared to respond quickly to treatment and were discharged within the week,
Finally, some subjects were lost to follow­up interview because they were returned
to court or discharged immediately following clinical determinations that compe­
tence had been restored and before researchers had an opportunity to approach
them for interview.
In comparison to defendants in the HI group who were not recruited for the
restored subgroup, those who were recruited had lower CAC:U scores at the time
of initial assessment (M = 16.2 compared to M = 18.9), t(159) = 2.98, p < .01,
and were less likely to be unemployed (50% compared to 64%), %2(7, n = 159) =
14.6, p < .05. There were no other differences found in comparisons of demo­
graphic, mental status, diagnostic, patienthood, or criminal justice history variables.
At the time of the second interview (i.e., after their competence had been judged
restored), the mean total BPRS score of defendants in the restored subgroup was
significantly lower (M = 29.6, SD ­ 5.5, versus M = 37.0, SD = 8.3), t(159) =
7.10, p < .001, than it had been when they began restorative treatment
The performance of the defendants in this subgroup after they had been re­
stored to competence compared to their initial performance on the MacSAC­CD
measures is summarized in Table 9. The subgroup means on all measures are sig­
nificantly higher on the second administration of the MacSAC­CD than they were
on the first, indicating improved performance as a function of changed competence
status. The second­administration scores of the restored subgroup are generally
similar to the scores found in the JT group of mentally disordered defendants who
were regarded as competent (compare withTable8).

Validity Question 3: Does Performance on the Research Instrument

Correlate Positively with Clinical Judgments of Competence?

Clinicians' judgments of competence were significantly correlated with Mac­

SAC­CD scores as summarized in Table 10. Statistically significant and moderately
strong (.29 to .46) correlations were found between clinicians' judgments and all
MacSAC­CD measures.

Validity Question 4: Does Performance on the Research Instrument

Correlate Negatively with Measures of Psychopathology and Impaired
Cognitive Functioning?

Table 11 displays the correlations between the MacSAC­CD competence to

assist counsel measures and measures of psychopathology and cognitive functioning
for defendants in the HI group. As can be seen, CAC:U scores were significantly
negatively correlated with total BPRS scores, and with BPRS psychoticism and with­
drawal subscale scores. CAC:U scores also correlated significantly positively with
VCF. CAC:R scores were significantly negatively correlated with BPRS psychoticism
subscale scores, and significantly positively correlated with VCF. CAC:A scores cor­
170 Hoge, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar

Table 9. Comparison Between First and Second

Administrations of Research Measures to Restored
Subgroup of the Hospital Incompetent Group (B = 97)
Administration
Measures First Second
Competence to assist council
CAC:U
It 16.19 19.47
SD 6.35 5.45
t,p ­3.87 p < .001
CAC:R
M 8.10 9.04
SD 3.21 2.84
t,p ­2.16 p < .03
CAC:A
M 9.90 10.83
SD 2.77 1.87
t,p ­2.70 p < .01
Decisional competence measures
DC:U­PG
M 12.47 15.19
SD 5.19 3.98
t,P ­4.08 p < .001
DC:U­WJ
M 15.59 18.96
SD 5.97 4.20
t,P ­4.55 p < .001
DC:R
M 11.20 12.57
SD 3.72 3.04
t,p ­2.78 p<.001
DC:A
M 3.19 3.58
SO 1.19 .99
t,p ­2.41 p < .02

Table 10. Correlations Between Experimental

Measures and Clinical Judgments of Competence­
Hospital Incompetents
Measures Clinical judgments
Competence to assist counsel
CAC:U .39**
CAC:R .41"
CAC:A .34**
Decisional competence
DC:U­PG .46**
DC:U­WJ .40"
DC:R .39"
DC:A .29**
* p < .05.
** p < .001.
*** p < .000.
 MacArtbur Adjudicative Competence Study 171

Table 11. Correlations Between Research Instruments and Psychopathology and Cognitive
Functioning Measures— Hospitalized Incompetent Croup
BPRS­ BPRS­ BPRS­ BPRS­ BPRS­
Measures total psychosis hostility depression withdiawal VCF
Competence to assist
counsel
CACU ­.28*** .44... .02 .07 ­.26" .60***
CACR ­.14 _39... .04 .14 ­.08 .40**
CACA ­.22" ­.41*** ­.12 .12 .07 .09
Decisional
competence
DCU­PG ­.33*** ­.46*** ­.01 .14 ­.27" .59*"
DC:U­WJ ­.31*** ­.47*** .00 ­.14 ­.33*** ­56*"
DC­.R ­.31*** ­.43*** ­.05 .10 ­.29***3 .56'"
DC:A ­.20* ­.40*** .01 .11 ­.12 31'"
•
• pS.05.
•• pS .001.
p••• S .000.

related significantly negatively with total BPRS scores, and with BPRS psychoticism
subscale scores. However, CACA was not significantly correlated with VCF.
Table 11 also shows the correlations between the MacSAC­CD decisional com­
petence measures and measures of psychopathology and cognitive functioning. Both
DC:U­PG and DC:U­WJ scores correlated significantly negatively with total BPRS
scores, and with BPRS psychoticism and withdrawal subscale scores. Both DOU­PG
and DCU­WJ scores correlated significantly positively with VCF. DC:R correlated
significantly negatively with total BPRS scores, and with BPRS psychoticism and
withdrawl subscale scores. DC:R correlated significantly positively with VCF. Fi­
nally, DCA scores were significantly negatively correlated with total BPRS scores,
and with BPRS psychoticism subscale scores. DC:A scores were significantly posi­
tively correlated with VCF.

Table 12. Correlations Between Research Instruments and Scores on

the Perceived Criminal Injustice Scale
Hospital Jail Jail
incompetent treated restored
Measures n = 159 n = 113 n ~ 94
Competence to assist council
CACU ­.07 ­.14 ­.20'
CACR ­.03 ­.;4 ­.04
CACA .12 ­.06 .08
Decisional competence
DCU­PG ­.02 ­.18 ­.06
DCU­WJ ­.02 ­21* ­.04
DC:R ­.02 ­.08 .05
DCA .08 ­.11 ­.15
• pS.05.
" p S .001.
"•p S .000.
172 Hoge, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar

Validity Question 5: Is Performance on the Research Instrument Uncorrelated

with Measures of Cynicism Toward the Criminal Justice System?

Table 12 displays the correlation between our research measures and an index
of cynicism toward the criminal justice, the Perceived Criminal Injustice Scale
(PCIS). The correlations are uniformly low, with only two of the 21 correlations
reaching statistical significance (CAC:U correlated negatively with PCIS in the JU
group, and DC:U­WJ correlated negatively with PCIS in the JT group).

DISCUSSION

We have presented here a set of new research measures for studying compe­
tence­related abilities in the criminal process, the MacArthur Structured Assess­
ment of the Competencies of Criminal Defendants. We believe that this set of
measures has several advantages over the existing measures fa the field. First, the
MacSAC­CD measures have superior content validity. The legal domains of the
MacSAC­CD reflect a current and coherent interpretation of the law of adjudicative
competence (Bonnie, 1993), one that encompasses a foundational requirement of
"competence to assist counsel" and a separate requirement of "decisional compe­
tence" that is amenable to contextual application. Second, these measures go be­
yond the testing of defendants' "understanding" the process of adjudication,
particularly the trial process, which is the focus of most current measures. Building
upon tests of capacity that have evolved in the treatment decision­making literature
(Appelbaum & Grisso, 1995), the MacSAC­CD permits the evaluation of other
competence­related abilities (appreciation, reasoning, and choice) not tapped by
most current measures. Third, "understanding" as measured by the MacSAC­CD
is not limited to assessing a defendant's existing or actual knowledge about the
legal process; rather, the first­disclose­and­then­test structure of the MacSAC­CD
understanding measures evaluates a defendant's capacity to understand new infor­
mation that might be disclosed to him by an attorney or others in the course of
preparation for adjudication. Finally, the MacSAC­CD offers the standardized ad­
ministration and criterion based, objective scoring that has often been called for in
this area (Grisso, 1992).

Summary of Main Findings

We have examined four basic psychometric properties of the MacSAC-CD, and

the findings are generally positive. The measures meet or exceed generally accepted
indices of internal consistency (a's > .70, item-scale correlations >. .30), and inter-
scorer agreement for all measures is satisfactory. The observed patterns of corre-
lations among individual measures are generally supportive of the theoretical
structure of competence-related abilities upon which the measures were con-
structed.
MacArthur Adjudicative Competence Study 173

We also have asked five questions about the construct validity of the MacSAC­
CD. Each question has been answered in the affirmative. Performance on the re­
search instrument reflects expected differences among groups that differ on
competence status. For each of the three measures of competence to assist counsel,
and for each of the four measures of decisional competence, defendants who are
hospitalized for restoration of competence to stand trial (the Hospital Incompetent,
or HI group) score significantly lower than mentally disordered defendants who
are being treated in jail, but who have not been identified as incompetent (the Jail
Treated, or JT group) and significantly lower than randomly chosen defendants from
the general jail population, who are not identified as incompetent to stand trial
and who are not receiving treatment (the Jail Unscreened, or JU group). For each
of the three competence to assist counsel measures and each of the four decisional
competence measures, the mean score of the JT group falls between those of the
HI group and the JU group, as predicted.
Performance on the research instrument reflects changes in competence status.
A subgroup of those defendants initially evaluated on the research measures upon
admission to the hospital for restoration to competence to stand trial (the HI group)
were later judged to have become restored to competence (the HR group). The
instrument was readministered while these defendants awaited the adjudication of
their criminal case to resume. For all three of the competence to assist counsel
measures and all four of the decisional measures, the mean scores are significantly
higher after restoration of competence than before. The scores of formerly incom­
petent defendants after their restoration to competence are similar to those of de­
fendants in the Jail Treated group.
Performance on the research instrument correlates positively and modestly with
clinical judgments of competence. The three competence to assist counsel measures
have a mean correlation of .38 with judgments of competence given by the defen­
dants' treating forensic clinicians. The four decisional competence measures have
a mean correlation of .39 with those clinical judgments. It should be noted that
more­than­modest correlations between our measures and clinical judgments of
competence are not to be expected, or necessarily desired. Clinical judgments are
not the "gold standard" for determining competence. To the extent that the abilities
we have identified in modern legal theories of adjudicative competence differ from
the abilities currently assessed by practicing forensic clinicians making competence
judgments, a high degree of correspondence between scores on our measures and
those clinical judgments would not be anticipated.
Performance on the research instrument correlates negatively with measures
of psychopathology and impaired cognitive functioning. All of the competence to
assist counsel measures and all of the decisional competence measures are signifi­
cantly negatively correlated with BPRS psychoticism subscale scores (ranging from
­.23 to ­.47). In addition, across all three groups, both the competence to assist
counsel and the decisional competence "Understanding" measures are highly posi­
tively correlated with verbal cognitive functioning (M r's of .55 and .56, respectively),
both competence to assist counsel and decisional competence "Reasoning" mea­
sures are somewhat more modestly correlated with verbal cognitive functioning (M
r's of .32 and .48, respectively), and both competence to assist counsel and deci­
174 Hoge, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar

sional competence "Appreciation" measures are poorly correlated with verbal cog­
nitive functioning (M r's of .15 and .22, respectively). As the "Appreciation" meas­
ures are the least "cognitive" of our research measures, their relative lack of
correlation with verbal cognitive functioning is not surprising.
Performance on the research instrument is generally uncorrelated with meas­
ures of cynicism toward the criminal justice system. The median correlation between
scores on the three competence to assist counsel measures in the three groups and
scores on the Perceived Criminal Injustice Scale is .10. The mean correlation be­
tween scores on the four decisional competence measures and scores on the PCIS
is .08.
In sum, the MacArthur Structured Assessment of the Competencies of Crimi­
nal Defendants appears to have high construct validity as a research measure of a
defendant's adjudicative competence: It can distinguish known groups of competent
and incompetent defendants; it can reflect changes in defendants' competence
status; and it correlates positively with clinical judgments of competence, negatively
with psychopathology and impaired intellectual functioning, and only minimally with
cynicism toward the justice system. In addition, it is firmly grounded both in legal
theory (Bonnie, 1993) and in psycholegal assessment theory (Grisso & Appelbaum,
1995). It is designed for standardized administration and objective scoring and is
psychometrically robust.

Limitations off Interpretation

Despite its advantages, the MacSAC­CD also has limitations. Most importantly,
the MacSAC­CD does not purport to assess all dimensions thought relevant to
"competence to stand trial." In particular, as a predominantly cognitive assessment
device, the MacSAC­CD does not attempt to assess a defendant's behavioral ability
to conform his or her demeanor to standards appropriate for a courtroom, and it
does not attempt to assess a defendant's interpersonal ability to cooperate with a
specific defense attorney. In addition, since the administration of the MacSAC­CD
is standardized, it does not take into account relevant contextual variables that may
arise in a given case, such as unusual factual or legal complexity (Grisso, 1986;
Roesch & Golding, 1980) that may be relevant to the determination of competence.
Rather, our aim has been to develop research measures that provide quantitative
indices of those competence­related abilities that are amenable to standardization
and objective measurement.
It is noted that our development of the MacSAC­CD deviates in some ways
from procedures used to develop conventional psychological "tests" (Foster &
Cone, 1995; Haynes, Richard, & Kubany, 1995). Rather than using empirical pro­
cedures to develop scales from initially large pools of potentially unrelated items,
we focused specifically on items that appeared directly responsive to legal criteria
and the functions served by the legal requirement of competence. For abilities re­
lating to understanding legal proceedings and communicating with counsel, our
point of departure was Dusky v. United States (1960) and other pertinent case law
and commentary. For abilities relating to decision making in criminal cases, we
MacArthur Adjudicative Competence Study 175

based our measures on those created by Grisso and Appelbaum (1995) for the
assessment of comparable competence­related abilities in the treatment decision­
making context. This strategy enabled us to sequence items in a way that preserves
the chronology of legal events, in the context of our hypothetical vignette, to which
a defendant is ordinarily exposed in the course of criminal prosecution. Although
we started with a somewhat larger item pool than is represented in the final version
of the MacSAC­CD and conducted pilot studies of earlier versions (Hoge et al.,
19%), decisions about item retention were driven more by considerations of face
and content validity than by purely empirical performance. In some instances items
that performed poorly (e.g., the item measuring Decisional Competence: Choice)
were nevertheless retained on considerations of face and content validity. As noted
earlier, and discussed in further detail below, we do not view the MacSAC­CD as
a psychological "test" in the classical sense, but as a tool for researchers to use to
evaluate systematically the relevant competence­related abilities of criminal defen­
dant groups of interest. Thus, our deviation from more classical test development
methods was intentional and, in our view, consistent with considerations appropriate
for the development of forensic assessment instruments.
We also note that the yield from the MacSAC­CD does not include a cate­
gorical index of "competence." That most of the currently available measures seek
to yield such a categorical determination is a testimony more to the political influ­
ences on their use in clinical practice (Grisso, 1987) than to their capacity to provide
a scientific basis for such categorical determinations. We emphasize that it was not
our goal to develop a "test" that would yield a value that could be transformed,
by applying a cut­off score, into an index of competence itself. In our view, a purely
objective test for competence determination is neither possible nor desirable. We
have noted in the introduction some of the limitations of existing competence mea­
sures, and in the discussion above additional limitations of the MacSAC­CD. Al­
though the MacSAC­CD offers significant improvement over existing measures, no
structured and standardized instrument, no matter how nuanced, can reflect all of
the variables relevant to a competence determination. Further, scoring of the Mac­
SAC­CD assigns equal weighting to each item in its various measures, an inherent
assumption about item values that may not be congruent with judicial reasoning.
Finally, "incompetence," like "insanity" or "dangerousness," is a conclusion that
involves social policy and value judgments that stand outside scientific measurement
(Morse, 1978). Thus, rather than aspiring to measure "competence" we have sought
the more modest goal of providing quantitative measures of discrete competence­
related abilities implicated in contemporary legal theory.
There are limitations to the particular research study reported here as well.
In the HI group, some subjects may have refused to participate in the research
study as a result of acute psychopathology. To the extent this is true, the findings
reported here may underestimate the deficiency in performance on the MacSAC­
CD of criminal defendants who are hospitalized as incompetent. Treatment effects
may also have contributed to an underestimation of impairments among hospital­
ized defendants. While HI subjects were approached shortly after admission, it may
be that some defendants responded to treatment provided in jail prior to hospital
admission or in the interval between admission and the administration of the in­
176 Hoge, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar

strument, and thus were less impaired on the MacSAC­CD than they were at the
time of referral for restoration.
Similar concerns may also be raised with respect to the performance of the
defendants receiving treatment at the jail (JT) for reasons other than competence
restoration; perhaps the stabilizing effects of psychotropic medications enhanced
their performance on the MacSAC­CD, resulting in an underestimation of their
(unmedicated) level of impairment. Although we cannot resolve this issue defini­
tively, we did examine the correlation between interval of admission to the jail and
MacSAC­CD scores. If medications or other treatment did function to enhance
their MacSAC­CD performance, then one might expect to find that those recruited
into the study after a lengthier interval of incarceration (and presumably, a longer
period of treatment) would perform better than those recruited soon after admis­
sion. However, the range of these correlations was from ­.07 to .11 (med = ­.04),
suggesting that treatment effects on performance in the JT group may have been
minimal.

Implications for Practice

The MacSAC­CD was designed as an instrument for use in empirical research

on adjudicative competence, and not for use in the actual practice of forensic evalu­
ation. As such, we have not offered cumulative percentile scores or other normative
data for the various groups of defendants in this study.
The instrument has 82 scored items, and the average time of administration
by a highly experienced research assistant was approximately two hours. While this
alone does not preclude its use in forensic assessment, the length of administration
is daunting in light of the 30­45­minute administration time required for the cur­
rently available measures. Other aspects of the MacSAC­CD that perhaps enhance
its use as a research measure also might discourage its use clinically. The Decisional
Competence: Understanding—Waiving Jury (DC:U­WJ) measure, for example,
might be superfluous in many evaluations where jury waiver is not being considered.
Others would question the clinical utility of data derived from the MacSAC­CD,
which are primarily nomothetic, in light of the usual emphasis on idiographic data
in determinations of adjudicative competence (Roesch, Hart, & Zapf, 1996).
Although the MacSAC­CD was not designed and is not being offered for fo­
rensic practice, we anticipate the development of a more streamlined, clinically port­
able and user­friendly instrument for clinical use. We view legal competence to be
a normative construct. Thus, normative data regarding a defendant's competence­
related abilities, although not likely to be dispositive in competence determinations,
may provide useful and relevant information to legal decision makers in individual
cases. We have completed preliminary field testing a 22­item measure, the Mac­
Arthur Competence Assessment Tool—Criminal Adjudication (MacCAT­CA), that was
distilled from the MacSAC­CD, using psychometric item­reduction techniques while
taking into account length of administration and, most importantly, concerns about
legal face and content validity. A multistate norming study, funded by the National
Institute of Mental Health, is now in progress to provide interpretive norms for
MacArthur Adjudicative Competence Study 177

the MacCAT­CA. It is anticipated that this instrument will provide legally relevant
normative data that will complement the more individualized data that forensic
clinicians commonly gather through social history inquiries, mental status examina­
tions, and clinical interviews that focus on case­specific details relevant to compe­
tence determinations.

ACKNOWLEDGMENTS

This work was supported by the Research Network on Mental Health and the
Law of the John D. and Catherine T. MacArthur Foundation. The authors would like
to thank members of the Network for their assistance in the conceptual development
of this project and for their comments on an earlier draft of this article: Shirley Abra­
hamson, Paul Appelbaum, Thomas Grisso, Pamela Hyde, Stephen Morse, Edward
Mulvey, Loren Roth, Paul Slovic, Henry Steadman, and David Wexler.

REFERENCES

American Psychological Association (1985). Standards for educational and psychological tests. Washington,
DC: American Psychological Association.
Anastasi, A. (1976). Psychological testing. New York: MacMillan.
Appelbaum, P. S., & Grisso, T. (1988). Assessing patients' capacities to consent to treatment. New England
Journal of Medicine, 319, 1635­1638.
Appelbaum, P. S., & Grisso, T, (1995). The MacArthur treatment competence study. I: Mental illness
and competence to consent to treatment. Law and Human Behavior, 19, 105­126.
Barnard, G. W., Thompson, J. W., Freemen, W. C., Robbins, L., Gies, D., & Hankins, G. C. (1991).
Competency to stand trial: Description and initial evaluation of a new computer­assisted assessment
tool (CADCOMP). Bulletin of the American Academy of Psychiatry and the Law, 19, 367­381.
Bonnie, R. J. (1990). The competence of criminal defendants with mental retardation to participate in
their own defense. Journal of Criminal Law and Criminology, 81, 419­446.
Bonnie, R. J. (1992). The competence of criminal defendants: A theoretical reformulation. Behavioral
Sciences and the Law, 10, 291­316.
Bonnie, R. J. (1993). The competence of criminal defendants: Beyond Dusky and Drope. University of
Miami Law Review, 47, 539­601.
Brakel, J. (1974). Presumption, bias, and incompetency in the criminal process. Wisconsin Law Review,
1974, 1105.
Bukatman, B. A., Foy, J. L., & Degrazia, E. (1971). What is competency to stand trial? American Journal
of Psychiatry, 127, 1225­1229.
Dusky v. United States, 362 U.S. 402 (1960).
Everington, C. T (1990). The competence for standing trial for defendants with mental retardation
(CAST­MR): A validation study. Criminal Justice and Behavior, 17, 147­168.
Foster, S., & Cone, J. (1995). Validity issues in clinical assessment. Psychological Assessment, 7, 248­260.
Godinez v. Moran, 113 S.Ct. 2680 (1993).
Golding, S. L, & Roesch, R. (1988). Competency for adjudication: An international analysis. In David
Weisstub (Ed.), Law and mental health: International perspectives, Vol. 4 (pp. 73­109). New York:
Pergamon.
Golding, S. L., Roesch, R., & Schreiber J. (1984). Assessment and conceptualization of competency to
stand trial: Preliminary data on the Interdisciplinary Fitness Interview. Law and Human Behavior,
9, 321­334.
Grisso, T. (1981). Juveniles' waiver of rights: Legal and psychological competence. New York: Plenum.
Grisso, T. (1986). Evaluating competencies. New York: Plenum.
Grisso, T. (1987). The economic and scientific future of forensic psychological assessment. American
Psychologist, 42, 831­839.
178 Hoge, Bonnie, Poythress, Monahan, Eisenberg, and Feucht­Haviar

Grisso, T. (1991). Clinical assessments for legal decision making: Research recommendations. In S. A.
Shah & B. D. Sales (Eds.), Law and mental health: Major developments and research needs (pp.
49­80). DDHS Publication No. (ADM) 91­1875. Rockville, MD: NIMH.
Grisso, T. (1992). Five­year research update (1986­1990): Evaluations for competence to stand trial.
Behavioral Sciences and the Law, 10, 353­369.
Grisso, T. (1996). Pretrial clinical evaluations in criminal cases: Past trends and future directions. Criminal
Justice and Behavior, 23, 90­106.
Grisso, T, & Appelbaum, P.S. (1991). Mentally ill and non­nentailv­ill patients' abilities to understand
informed consent disclosures for medication: Preliminary data. Law and Human Behavior, 15,
377­388.
Grisso, T., & Appelbaum, P.S. (1995). The MacArthur Treatment Competence Study. HI: Abilities of
patients to consent to psychiatric and medical treatments. Law and Human Behavior, 19,149­174.
Grisso, T., Appelbaum, P. S., Mulvey, E. P., & Fletcher, K. (1995). The MaeArthur Treatment Competence
Study. II: Measures of abilities related to competence to consent to treatment Law and Human
Behavior, 19, 127­148.
Grisso, T., Cocozza, J., Steadman, H. J., Fisher, W., & Greer, A. (1994). The organization of pretrial
forensic evaluation services: A national profile. Law and Human Behavior, 18, 377­394.
Grander, T. (1978). Two formulas for determining the readability of subject consent forms. American
Psychologist, 33, 773­775.
Hagan, J., & Albonetti, C. (1982). Race, class, and the perception of injustice in America. American
Journal of Sociology, 88, 329­355.
Haynes, S. N, Richard, D. C. S., & Kubany, E. S. (1995). Content validity in psychological assessment:
A functional approach to concepts and methods. Psychological Assessment, 7, 238­247.
Hoge, S. K., Bonnie, R. J, Poythress, N., & Monahan J. (1992). Attorney­client decision making in
criminal cases: Client competence and participation as perceived by their attorneys. Behavioral
Sciences and the Law, 10, 385­394.
Hoge, S. K., Poythress, N., Bonnie, R., Eisenberg, M., Monahan, J., Feucht­Haviar, T., & Oberlander,
L. (1996). Mentally ill and non­mentally­ill defendants' abilities to understand information relevant
to adjudication: A preliminary study. Bulletin of the American Academy of Psychiatry and the Law,
24, 187­197.
Laboratory of Community Psychiatry (1973), Competency to stand trial and mental illness (DHEW
Publication No. ADM77­103). Rockville, MD: Department of Health, Education, and Welfare.
Lipsitt, P., Lelos, D., & McGarry, A. L. (1971). Competency for trial: A screening instrument. American
Journal of Psychiatry, 128, 105­109.
Morse, S. (1978). Law and mental health professionals: The limits of expertise. Professional Psychology,
9, 389­399.
Nicholson, R. A, (1993). Current methods for assessing criminal competencies. Paper presented at the
Annual Conference of the American Psychiatric Association, San Francisco.
Nicholson, R. A., Briggs, S., & Robertson, H. (1988). Instruments for assessing competency to stand
trial: How do they work? Professional Psychology, 19, 383­94.
Nicholson, R., & Kugler, K. E. (1991). Competent and incompetent criminal defendants: A quantitative
review of comparative research. Psychological Bulletin, 109, 355­370.
Nicholson, R., Robertson, H., Johnson, W., & Jensen, G. (1988). A comparison of instruments for
assessing competency to stand trial. Law and Human Behavior, 12, 313­321.
Nunnally, J. C. (1978). Psychometric theory. New York: McGraw­Hill.
Overall, J., & Gorham, D. (1962). The Brief Psychiatric Rating Scale. Psychological Reports, 10,799­812.
Overall, J., & Porterfield, J. (1963). Power vector method of factor analysis. Psychometrika, 28, 415­422.
Poythress, N., Bonnie, R. J., Hoge, S. K, Monahan, J., & Oberlander, L. (1994). Client abilities to assist
counsel and make decisions in criminal cases: Findings from three studies. Law and Human
Behavior, 18, 435­450.
Robey, A. (1965). Criteria for competency to stand trial: A checklist for psychiatrists. American Journal
of Psychiatry, 122, 616­623.
Roesch, R, & Golding, S. L. (1980). Competency to stand trial. Urbana, IL: University of Illinois Press.
Roesch, R., Hart, S. D, & Zapf, P. A. (19%). Conceptualizing and assessing competency to stand trial:
Implications and applications of the MacArthur treatment competence model. Psychology, Public
Policy, and Law, 2, 1­18.
Roesch, R., Ogloff, J., & Golding, S. (1993). Competency to stand trial: Legal and clinical issues. Applied
and Preventive Psychology, 2, 43­51.
Steadman, H. J., & Hartstone, E. (1983). Defendants incompetent to stand trial. In J. Monahan & H.
J. Steadman (Eds.), Mentally disordered offenders (pp. 39­64). New York: Plenum.
MacArthur Adjudicative Competence Study 179

Steadman, H., J. Monahan, J., Hartstone, E., Davis, S., & Robbing, P. (1982). Mentally disordered
offenders: A national survey of patients and facilities. Law and Human Behavior, 6, 31­38.
Wildman, R., Batchelor, E., Thompson, L., Nelson, E, Moore, J., Patterson, M., & deLaosa, M. (1979).
The Georgia Court Competency Test. Newsletter of the American Association of Correctional
Psychologists, 2, 4 (abstract).
Wilson v. U.S., 391 E2d 460 (1968).
Woerner, M. G., Mannuzza, S., & Kane J. M. (1988). Anchoring the BPRS: An aid to improved
reliability. Psychopharmacology Bulletin, 24, 112­118.