manipulated by the investigators—one group
received it, the other group did not. The dependent
variable was the participant’s self-reported levels of
depression. Good experimental procedure was also
followed bymatchingsubjects on important variables
that might have affected the outcome of the
research (e.g., gender, age, preexisting levels of
depression). Also, participants must beassigned ran-
domlyto experimental and control groups. The idea
is that the only significant feature different for the
two groups should be the experience of bibliother-
apy. Hence, the lower depression scores in the
experimental group can be assumed to be caused
by the addition of bibliotherapy to treatment.
Between- and Within-Group Designs. Our
depression treatment study just described is also an
example of the between-group design. In abetween-
group design, we have two separate sets of partici-
pants, each of which receives a different kind of
treatment or intervention. Take, for example, a tra-
ditional study of therapy efficacy. The question to
be addressed is whether a particular form of therapy
is better than no treatment at all. In its simplest
form, an experimental group (receiving some
form of treatment) is compared to a control group
(receiving no treatment at all). Ideally, patients
would be randomly assigned to each group. Some
set of measures (e.g., level of anxiety, interview
impressions, or test data) is taken from all patients
in both groups prior to treatment (or no treatment),
after treatment, and perhaps at a follow-up point 6
months or 1 year after treatment is concluded. Any
differences between the two groups either at the
conclusion of treatment or during the follow-up
are assumed to be a function of the treatment that
was received by the experimental group.
In awithin-group design, comparisons might be
made on the same patient at different points in
time. To illustrate how this procedure works, sup-
pose we are interested in the effects (e.g., level of
distress) of being on a waiting list. We might decide
to place every patient on a 6-week waiting list but
carry out a variety of assessment procedures before
doing so (point A). Six weeks later, these patients
would be reassessed just prior to beginning
treatment (point B). At the conclusion of treatment
(point C), the patients would be assessed for a third
time, and they might also be followed up later
(point D). Any changes taking place between points
A and B (while on the waiting list) could be com-
pared to the changes that take place between point
B and point C or D. These more complex analyses
of changes would give us a better view of the effi-
cacy of treatment relative to merely being on a
waiting list.
There are many variations of the within-group
design. However, a major advantage is that it
requires fewer participants. Indeed, as we shall see
later in the case of single-subject designs, we can
determine whether or not a specific intervention
has an effect merely by observing one participant.Internal Validity. Sometimes an experiment is
notinternally valid. That is, we are not sure that
the obtained outcome is really attributable to our
manipulation of the independent variable. Some
studies do not even include a control group for
comparison with an experimental group. In this
case, any observed changes could be the result of
some other variable. For example, suppose we did
not have a control group in our study of the effects
of bibliotherapy on depression in addition to anti-
depressants. Even though the experimental group
showed decreases in depression, perhaps this was
not at all due to bibliotherapy. Maybe it was due
to being on an antidepressant for a longer period of
time. Or maybe it was due to talking to someone
on the phone (albeit briefly) once a week. Without
a control group that also experienced these other
events, one can never be sure. In short, when extra-
neous variables are not controlled or cannot be
shown to exist equally in experimental and control
groups, these variables mayconfoundthe results.
We shall see later that sometimes in studies of
therapy effectiveness, one group of patients receives
a new form of therapy. A second group of patients
is matched with the therapy group (or patients are
randomly assigned to the two groups) and then
assigned to a waiting list. The assumption is that
the only difference between the waiting-list sub-
jects and the therapy subjects is that the latter110 CHAPTER 4