Researching Abnormality 185The Therapy Dose-
Response Relationship
We have been discussing research
on treatment that uses experi-
mental designs, randomly assign-
ing participants to two or more
groups and then comparing the
results. Another approach re-
lies on correlational methods. By
defi nition, correlational studies of
treatment do not include random
assignment, nor do they have in-
dependent or dependent variables.
Rather, a correlational research
study determines whether changes
in one variable are associated with
changes in another variable.
Correlational research can an-
swer a question like this: Is more
treatment related to greater im-
provement? In other words, is a
higher “dose” of therapy (more
sessions) associated with a better
“response”? This association be-
tween dose and response is referred
to as the dose-response relationship,
and research suggests that the gen-
eral answer to this question is yes.
More sessions are associated with a
better outcome (Hansen, Lambert, & Forman, 2002; Shadish et al., 2000). In gen-
eral, patients improve the most during the early phase of treatment (see Figure 5.7),
and they continue to improve, but at diminishing rates, over time (Lutz et al., 2002).
There are individual exceptions to this general pattern, as when someone with a spe-
cifi c phobia has an extremely good response to a single session of exposure therapy
(Hellström & Öst, 1995; Öst, Grandberg, & Alm, 1997, Öst et al., 2001). However,
people with more severe or entrenched problems, such as schizophrenia or personality
disorders, may not show as much benefi t in the early stages of outpatient treatment but
rather tend to improve over a longer period of time.
Because the dose-response relationship is correlational, it does not indicate
whether the increased number of sessions causesthe increased response. It is pos-
sible that people who are feeling better during the course of treatment are more ea-
ger or more willing to attend additional sessions than those who are not responding
as well. If this were the case, the response would be “causing” the increased dose5.7 • The Dose-Response Relationship Comparing a more stringent defi nition of
“improvement” (Figure 5.7a) to a more liberal one (Figure 5.7b) makes it clear that the exact criteria for
improvemnt determine the particular height and shape of the curve for the dose-response relationship.
Source: Lambert et al., 2001. For more information see the Permissions section.Figure 5.7d 57 Th D R R l ti hi dfi f
(a) Time to recovery
100
90
80
70
60
50
40Percentage of patients30
20
100 5 10 15 20 25
Therapy session30 35 40 45Among 10,000 patients with various psychological
disorders, half attained clinically signifi cant
improvement (that is, recovery) by the 21st
session. Another 25% attained clinically signifi cant
improvement by the 40th session. However, most
people in RCTs, or in therapy in general, receive far
fewer than 40 sessions.(b) Time to improvement
100
90
80
70
60
50
40Percentage of patients30
20
100 5 10 15 20 25
Therapy session30 35 40 45Examining data using a lower standard of
improvement—any positive change that was stable
over time—and including patients who started
out able to function reasonably well despite their
disorder, yields different results: Half the patients
attained this lower standard of improvement by the
7th session, with another 25% “improved” by the
14th session (Lambert, Hansen, & Finch, 2001).A curious fi nding invites specula-
tion: People in the eastern part of the
United States remain in treatment
longer than those in the western part
(“Fee, Practice, and Managed Care
Survey,” 2000). This is merely a cor-
relation. One possible explanation is
that people in the eastern part have a
different defi nition of “improvement“;
that is, they continue in treatment until
their symptoms have improved more
than those of their counterparts in the
western part of the country. However,
there are other possible explanations;
PCL/Alamy Getty Images perhaps you can think of some.