person dispensing the drug does not know which is
which. Nor can the participant tell because the two
kinds of pills are identical in size, shape, color,
weight, texture, taste, and so on.
Matching, random assignment of participants
to experimental and control groups, use of control
groups, and double-blind procedures are all meth-
ods of helping ensure that experiments display some
degree of internal validity.
External Validity. When it is discovered that the
results of an experiment cannot be generalized
beyond the narrow and exact conditions of the
experiment, the research does not show external
validity. If our depression study results apply only to
that specific primary care clinic, there is a problem. In
fact, most experimental research is done with the
hope of generalizing the results beyond the confines
of the immediate setting. Actually, it can be very
difficult to determine the external validity of a
given study. All too often, results produced in the
laboratory cannot be reproduced in real-life settings.
The worst mistake is merely to assume that because
certain results were produced in situation A, they
will automatically occur in situation B. Although
laboratory experimental research usually allows bet-
ter control of variables, its“artificial”nature may
prevent wide generalization.
Analog Research. The question of generalization
of results is particularly salient foranalog studies—
studies that are conducted in the laboratory,
where control is easier to exert, but whose condi-
tions are said to be“analogous”to real life. Most
often, they are used to highlight the nature of psy-
chopathology or therapy. For example, when
Watson and Rayner (1920) tried to show how Little
Albert could learn to be phobic for white rats, they
were constructing an analog of the way they
thought real-life phobias were acquired (see
Box 4-1). As another example, some researchers
study the correlates and effects of depression by
inducing depressive mood in non-depressed partici-
pants through the use of traditional mood induction
techniques (e.g., reading text that has been shown
to increase levels of dysphoria) or of slides and music
that induce certain mood states. These techniques
presumably create a depressive state analogous to
that seen in clinical depression.
In one sense, almost all experimental studies are
analog studies. But when severe practical or ethical
constraints prevent us from creating real-life condi-
tions, we must turn to analog situations. The
advantage of analog studies is that better internal
validity is possible because of the superior control
we can exert in the laboratory. The Achilles heel of
the method is the degree of similarity between the
analog and the real thing. For example, suppose we
decide to study the role of failure in creating
depression within adolescents. We do not want to
study adolescents with genuine depression for sev-
eral reasons. First, it is difficult to get a large-enough
sample who have similar backgrounds and who are
equally depressed. Second, would it really be ethical
to subject such adolescents to a strong, significant
failure experience and thereby risk plunging them
even deeper into depression? This, by the way,
illustrates a real dilemma with analog research. If
our experimental manipulations are truly effective,
they may be harmful (and therefore unethical). If
they are mild, the research becomes trivial and has
little external validity. Given the foregoing ethical
constraints and the fact that analog conditions will
allow us to control the degree and kind of failure,
the nature of the participants, and so on, the use of
analog studies can become an attractive alternative.
But when we use analog procedures, we may
pay a price. For example, in the previous hypothet-
ical experiment, how do we know that our parti-
cipants are really the same as participants
professionally diagnosed as depressed? Perhaps the
adolescents are recruits from a local school district
whose only claim to being depressed is a score
above some cutoff point on a questionnaire that
purports to measure depression. They are notclinical
cases of depression. In addition, we have no assur-
ance that an experience of failure on a problem in a
laboratory setting is at all the same as a“real”failure
in the everyday life of a person with depression.
Some have urged the use of animals for
research because they do not present some of the
methodological problems that exist in research with112 CHAPTER 4