158 CHAPTER 5
Control group
A group of participants in an experiment
for which the independent variable is not
manipulated, but which is otherwise treated
identically to the experimental group.
Bias
A tendency that distorts data.
a comedy for the “loss-no helplessness” group. It could be the type of movie shown
fi rst—drama or comedy—and not differences between the groups that produces the
effects on the dependent variable. Confounds lead to ambiguous or uncertain results.
To minimize the possibility of confounds, the researcher should try to ensure that the
experimental manipulation alters only the independent variables (and does not inadver-
tently affect other variables, such as the order in which the movies are presented) and
that only those changes in the independent variables affect the dependent variable. It is
not always possible to ensure that this is the case, however, and thus the experimenter
must examine every reasonable hypothesis that might explain the effect.Control Groups and Conditions
A common method for examining the possible effects of confounds in an exper-
iment is to create a control group. The experimental group (or groups) and the
control group are treated identically throughout the experiment, except that the
independent variable is not manipulated for the control group. For example, say
that your hypothesis was confi rmed; the group of participants who had suffered an
early loss and experienced helplessness did in fact have a more negative mood after
watching a movie with the theme of loss. But here’s a potential confound: Maybe
these people are just very sensitive to depressing movies in general, not to themes
of loss specifi cally. To rule out this possibility, you would control for this factor by
testing another group of these people, showing them a depressing movie that has
nothing to do with loss. (This movie should be as similar as possible to the fi rst
one in every way, except that it does not have the theme of loss.) If your hypothesis
is correct, then you should fi nd a much larger drop in mood following the movie
about loss than following the movie that was merely depressing—which would pro-
vide evidence that it’s the theme of loss itself that is important.
To use a control group appropriately, it must be as similar as possible to the
experimental group. If the members of a control group differ signifi cantly from
members of the experimental group in terms of age, education, cultural background,
temperament, or any other characteristic, one or more confounding variables has
been introduced. Such a confound would cloud the interpretation of the results,
because a difference in the dependent variable could be attributed to the confound
instead of to the relation you intended to test. Perhaps the control group responded
less strongly to the depressing movie simply because most of the participants in that
group happened to be temperamentally placid.
You can imagine how hard it can be to match a control group perfectly to an
experimental group. This diffi culty often leads researchers to match the experimen-
tal group to the most similar possible control group: the experimental group itself!
Sometimes, rather than having two separate groups that are treated differently, re-
searchers have all participants take part in different conditions, or circumstances,
which correspond to the different ways that experimental and control groups would
be treated in a study that had both types of groups. For example, the same people
could watch a movie about loss and at a different time watch a movie that was
depressing but not about loss. However, when the same group of people take part
in more than one condition, you need to avoid a confounding that might be intro-
duced by the order of presentation of the conditions. For instance, if the “loss” fi lm
was always presented fi rst, it could be that the participants were more alert during
that part of the study, and that’s why they responded differently. To avoid this, you
wouldcounterbalance the order of exposure to each condition: Half of the par-
ticipants would watch the “loss” fi lm fi rst, and the other half would watch the
“non-loss/depressing” fi lm fi rst. This procedure would ensure that each condition
occurred equally often in each place in the order of presentation.Bias
The way a study is set up can affect the assignment of participants into groups
or infl uence the outcome of the experiment. For instance, suppose that you have
two groups: members of one group see a “loss” film and members of the other
see a “non-loss/depressing” fi lm. When you assign participants to the groups, you