Expectations, motivations and task demands (Reisberg, ch. 11, methods)
In most experiments, we're trying to study how people behave under natural
circumstances-how they behave when they're just being themselves. As a result,
it is crucial that we take steps to minimize the demand characteristics
of a study.
As the textbook chapter describes, the term experimental demand refers to
any cues in the experiment that might signal to participants how they "ought to"
behave. In some cases, the demand indicates to participants what results the
researcher hopes for (that is, results that would confirm the researcher's
hypothesis), and this may encourage participants to make the hoped-for response,
even if they're inclined toward some other option. In other cases, experimental
demand somehow suggests that certain responses are more desirable than others-so
that, for example, participants perceive some responses as indicating greater
intelligence, or greater sensitivity. And, of course, it's plausible that
participants will choose these responses to avoid appearing stupid or
insensitive.
What can we do to avoid these effects, so that we don't guide participants
toward some particular response, but instead observe participants as they
normally are? Researchers use several different strategies. First, we do all we
can to make sure that experimental demand never arises in the first place. Thus,
we make sure that the procedure contains no signals about what the hypothesis
is, or which group is receiving the experimental treatment and which group is
the control. Likewise, we do what we can to phrase our questions, and cast the
response options that are available, so that no responses seem preferable to any
others.
Second, if participants focus their attention on the experimental manipulation,
they may be too cautious or too deliberate in how they respond. Worse, they may
develop their own notions about how they're "supposed to" respond to this
manipulation. (See pp. 373-374 in Chapter 11 in the textbook for an example of
this.) To avoid these concerns, most experiments avoid drawing any attention to
the experiment's main manipulation; indeed, many experiments provide a "cover
story" about what the experiment is about, seeking to draw attention to some
other aspect of this procedure. In this way, we maximize the chance that during
the experiment, participants will behave naturally and spontaneously in response
to the manipulation.
Third, we do what we can to make sure that all participants in all conditions
receive exactly the same treatment. Thus, we make sure to encourage all
participants in the same way, give them similar instructions, and so on. In some
cases, we rely on a double-blind procedure, in which neither the
participant nor the person administering the procedure knows what the experiment
is about, or which comparison group the participant is in. This ensures that the
administrator won't be more encouraging, or more forceful, with one group in
comparison with the other; it also guarantees that all participants will have
the same expectations about the experiment.
Can we make certain that participants' expectations, goals, and hypotheses play
no role in shaping our data? Probably not, and this is one of the reasons
why replications (with other participants and other procedures) are so
important. Even so, we do what we can to minimize the contribution of these
factors, and so make it far more likely that our results can be understood in
the terms we intend.