Comparative Studies
Therapy research has become increasingly sophisti-
cated since Eysenck’s critique. More and more
studies do not just assess outcomes but also compare
two or more techniques in terms of efficacy. How-
ever, several critics have questioned whether some
designs are powerful enough (have enough study
participants) to detect differences between alterna-
tive treatments (Kazdin & Bass, 1989).
The Temple University Study. A major com-
parative study was conducted by Sloane, Staples,
Cristol, Yorkston, and Whipple (1975a, 1975b).
More than 90 outpatient neurotics were assigned
to (a) behavior therapy, (b) short-term psychoana-
lytically oriented therapy, or (c) a minimal treat-
ment waiting list. The groups were matched in
terms of sex and severity of symptoms; otherwise,
assignment to the different therapeutic conditions
was random. Treatment was carried out by three
behavior therapists and three analysts, all of whom
were highly trained and experienced. Before-
and-after measurements included psychological
tests, a target symptoms technique (a measure ori-
ented toward individualized treatment criteria), a
standard interview, informants’ reports (people
who had known the patients for an average of
12 years), and ratings by the therapist, the patient,
and an independent rater. Note that these proce-
dures correspond closely to several of the recom-
mendations made earlier in the methodology
section of this chapter.
All three groups had improved at a 4-month
posttesting point. However, the psychoanalytic
therapy (PT) and behavior therapy (BT) groups
improved more than did the waiting list (W)
group. The independent rater could find no differ-
ences with respect to improvement in target symp-
toms between the BT and PT groups. General
estimates, such as improvement in work or social
situations, also failed to discriminate between the
PT and BT groups. The rater’s global outcome
assessment indicated that 80% of both the BT
group and the PT group improved, whereas only
48% of the W group did. For general adjustment,
93% of the BT group and 77% of both the PT
group and the W group showed improvement.
Although the comparative change between the
BT group and the PT group varied depending on
the specific criterion, in general there was a slight
trend in favor of the BT group. There was no evi-
dence of deterioration effects, and all three groups
maintained their improvement after 1 year. It
should be noted, however, that there was a trend
for the improvement to continue and for patients in
the W group to approach or equal those in the BT
and PT groups. This could suggest that what ther-
apy does is accelerate change, rather than produce
more change, as compared to no treatment. Psy-
choanalysts produced better outcomes with less dis-
turbed patients, whereas the level of patient
disturbance made no difference for behavior thera-
pists. This suggests that behavior therapy may have
greater versatility, perhaps because of the greater
flexibility and eclecticism of the techniques
employed by behavior therapists.Meta-analyses. In 1977, Smith and Glass pub-
lished a review of nearly 400 psychotherapy out-
come studies. Their verdict was that the evidence
is convincing in showing the effectiveness of psy-
chotherapy. What is unique about their review is
their use of a method calledmeta-analysis. Using this
method, they analyzed all therapy studies that dealt
with at least one therapy group and one control
group, or with two therapy groups, and averaged
the effects of therapy across all these studies. In their
analysis, they definedeffect sizeas the mean differ-
ence between treated and control participants’
scores on relevant outcome measures divided by
the standard deviation of the control group.
Earlier, we mentioned the meta-analysis of Smith
et al. (1980). Their survey included 475 studies
involving 25,000 patients treated by some 78 thera-
pies for an average of 16 sessions. Outcome compar-
isons showed effect sizes that averaged a .85 standard
deviation. As previously mentioned, this means“the
average person who receives therapy is better off at
the end of it than 80% of the persons who do not”
(Smith et al., 1980, p. 87). Using a somewhat stricter
definition of therapy and removing“placebo therapy”
and“undifferentiated counseling”from the data, the336 CHAPTER 11