Yet a mental health professional may feel strongly that
the patient could benefit from admission (or perhaps
has nowhere else to go). What should be done? The
“humanitarian”choice often seems to be to alter a
diagnosis, or at least to“fudge”a bit. The patient
with alcohol dependence suddenly is diagnosed with
something else. Similarly, an insurance company may
reimburse a clinic for the treatment of patients with
one diagnosis but not another. Or perhaps one diag-
nosis permits six therapy visits but another allows as
many as fifteen sessions. Therefore, a diagnosis may be
intentionally or unintentionally manipulated.
These examples may lead us to believe that
diagnostic unreliability is the rule and not the
exception. However, Meehl (1977), for example,
feels that psychiatric diagnosis is not nearly as unre-
liable as it is made out to be. Specifically, Meehl
argues that if we confine ourselves to major diag-
nostic categories, require adequate clinical exposure
to the patient, and study well-trained clinicians who
take diagnosis seriously, then interclinician agree-
ment will reach acceptable levels.
The field of psychopathology has begun to
address these concerns about reliability by develop-
ing structured diagnostic interviews that essentially
“force”diagnosticians to assess individuals for the
specificDSMcriteria that appear in the diagnostic
manual. For example, there are now several struc-
tured interviews that assess features of Axis I disor-
ders, and a number of structured interviews for Axis
II disorders exist as well. Interestingly, the overall
level of diagnostic reliability reported in empirical
studies has increased greatly following the introduc-
tion of these structured interviews. It is clear that
adhering to the structure and format of these inter-
views has led to a significant increase in diagnostic
reliability. Table 5-6 presents a brief section from a
structured interview. We will discuss structured
interviews in more detail in Chapter 6.
However, even with the use of structured inter-
views, reliability is not equally good across all catego-
ries. The presence versus absence of some disorders
(e.g., generalized anxiety disorder) may be particu-
larly difficult to judge. Further, there is some question
as to whether or not busy clinicians will devote the
time and effort necessary to systematically evaluate
the relevant diagnostic criteria. Reliability coefficients
never seem as high in routine, everyday work settings
as they are in structured research studies.Validity. Reliability will directly affect the valid-
ity of a diagnostic system. As long as diagnosticians
fail to agree on the proper classification of patients,
we cannot demonstrate that the classification system
has meaningful correlates—that is, has validity.
Important correlates include prognosis, treatment
outcome, ward management, etiology, and so on.
And without predictive validity, classification
becomes an intellectual exercise devoid of any
really important utility. However, if we can dem-
onstrate that categorization accurately indicates eti-
ology, course of illness, or preferred kinds of
treatment, then a valid basis for its use has been
established.
The predominant method for establishing the
validity of a diagnostic construct was outlined in a
classic article by L. N. Robins and Guze (1970).
They proposed that establishing the diagnostic
validity of a syndrome is a five-stage process:
(a) clinical description, including a description ofT A B L E 5-6 SIDP-IVQuestions Used to
Assess a Dependent Personality
Disorder Criterion
CRITERION 1-DEPEN: HAS DIFFICULTY MAKING EVERY-
DAY DECISIONS WITHOUT AN EXCESSIVE AMOUNT OF
ADVICE AND REASSURANCE FROM OTHERS
Questions to Ask
Some people enjoy making decisions and other people
prefer to have someone they trust tell them what to do.
Which do you prefer?
Do you often turn to others for advice about everyday
decisions like what to have for lunch or what clothes to
buy?
Scoring
0 not present
1 subthreshold
2 present
3 strongly present
Adapted from the Structured Interview for DSM-IV Personality (SIDP-IV:
Pfohl, Blum, & Zimmerman, 1994). Copyright © 1989, 1994. Used with
permission.150 CHAPTER 5