Dawes, 1979, 1994; Dawes, Faust, & Meehl, 1989;
Garb, 1998; Goldberg, 1991; Grove et al., 2000;
Kleinmuntz, 1990; Meehl, 1986; Wiggins, 1973).
For example, in a comprehensive review of studies
pitting clinical versus statistical prediction, Grove et
al. (2000) reported, once again, that statistical pre-
diction was superior in roughly 50% of the studies,
whereas clinical prediction was superior in only a
small number of studies (approximately 6%). This
trend held true regardless of the judgment task (e.g.,
predicting psychotherapy outcome), type of judges
(physicians vs. psychologists), judges’amount of
experience, or types of data being combined.
Thus, because it is typically less expensive than clin-
ical prediction (primarily because of the personnel
costs involved), statistical prediction is preferred. As
stated by Meehl (1986):
There is no controversy in social science
that shows such a large body of qualita-
tively diverse studies coming out so uni-
formly in the same direction as this one.
When you are pushing 90 investigations
[this number is dramatically higher as of
2005], predicting everything from the
outcome of football games to the diagnosis
of liver disease, and when you can hardly
come up with a half dozen studies showing
even a weak tendency in favor of the cli-
nician, it is time to draw a practical con-
clusion. (pp. 373–374)
More recently, Garb (2005) reviewed findings
on the accuracy of clinical judgments in five differ-
ent areas: (a) the description of personality and psy-
chopathology; (b) diagnosis; (c) case formulation;
(d) behavioral prediction; and (e) decision making.
Although clinicians seem aware of the official diag-
nostic manual’s description of symptoms of psycho-
pathology and appear to be able to reliably describe
patients in these terms (especially if using structured
diagnostic instruments), they appear much worse at
reliably describing personality features of patients.
This limitation may be due to the relative lack of
structured instruments for personality that are used
in clinical practice, or simply that clinicians tend to
make these judgments based on their own internal,
but not well operationalized, implicit theories of
personality that are based on everyday experience
but not on scientific findings.
In general, clinical case formulations appear
highly subjective and unreliable; data for the case
formulations are often obtained unsystematically
from patients as well as informants. However, such
reports are known to be biased by memory heuristics
and biases that are influenced by the timing of events
and affect associated with events, for example.
The prediction of future behavior is a frequent
issue faced by clinical psychologists (e.g., what is the
likelihood this client will commit an act of vio-
lence?), but we do not appear to be particularly
good at it. Studies suggest that violence is often
overpredicted for certain genders (men) and certain
racial groups (African Americans).
Concerning treatment decisions, there is cur-
rently a great focus on evidence-based assessments
that have demonstrated validity in informing treat-
ment decisions (e.g., see Hunsley & Mash, 2008).F I G U R E 10-1 Paul Meehl is widely recognized as a
major proponent of the actuarial or statistical approach
to prediction.Courtesy of Paul Meehl292 CHAPTER 10