4 - 14
value for a significant outcome. However, excellent statistical techniques are now available for assessing
multiple dependent variables.
Two, three, or more dependent variables are recorded and analyzed in a single procedure referred to
as a multivariate analysis. Multiple measures may vary together, thus suggesting a common underlying
process for all measures. When they do not vary together, different processes are suggested. Both
theoretical and practical considerations determine the number of measures used. A discussion of statistical
procedures for analyzing concurrently several dependent measures is too advanced for a first course in
research methods. Suffice it to say there are advantages to this type of analysis.
Aside from statistical advantages, however, there are other reasons for recording two or more
dependent variables. It may be that under the conditions of your experiment, a single dependent measure
may not show any systematic relationship to your independent variable. Your measure may be too
insensitive or too variable. If you record other, different measures, your chances of finding a systematic
relationship may be increased. In addition, recording more than a single dependent variable will allow
you to evaluate the relationship among them. You have little to lose and much to gain by recording more
than a single dependent variable, unless doing so is inconvenient, time consuming, or expensive.
Response Classes of Dependent Variables
The number of dependent measures recorded by researchers is determined by both theoretical and
practical considerations. Investigators studying behavior, whether in a laboratory or an applied setting,
generally use three major classes of responses. These three classes of responses are motor responses,
physiological measures, and self-report measures. Whatever measure is used, great care must be taken
when measuring and recording the response. It is not uncommon for researchers to record different
classes of responses within the same experiment. Each has advantages and disadvantages associated with
its use.
Motor responses involve the skeletal muscle system in some way. These responses may vary in
terms of accuracy, frequency, latency, duration, or intensity. Some examples of motor responses are
walking, talking, drinking, eating, crying, fighting, running, smiling, studying, smoking, gambling,
freezing, jumping, bar pressing, playing, key pecking, and choosing. When motor responses are
automatically recorded, mechanically or electronically, errors due to the human observer are virtually
eliminated. However, only some motor responses can be automated; observers must record other
responses. Relying on observers to note and record our dependent variables is a serious issue for a
considerable amount of research being done in psychology today. The issue is that human observers are
not perfectly reliable at the task and thus represent an imperfect measuring instrument. (This issue will be
discussed in more detail in Chapter 6.)