Section Sixscores, in which each individual participant’s pretest score is
subtracted from the posttest score. Mean values of gain scores
are computed within the experimental group and the control
group. The significance of the difference in means of the gain
scores between the two groups can then be tested with a t-
test. In realistic company settings, a control group should not
be too difficult to obtain. Members of the control group can be
employees, who may even participate in the same training at
some point in the future, after its effectiveness has been
determined. When assigning employees to control groups,
evaluators must be able to provide justifications as to why
control group members are not being trained when they know
that another group is receiving training. Another aspect of
control groups, of which evaluators should be aware, is the
group’s knowledge of the training. In work settings, it is often
likely that members of the control group may be receive
substantial information about the training through discussions
with other employees.^57
The second experimental design is the posttest-only
control group design. This design also utilizes an experimental
and control group with random assignment of participants.
Unlike the pretest–posttest control group design, there is no
pretest. With randomization and sufficient sample size, the two
groups may be assumed to have equivalent characteristics,
such as ability and experience. This design is useful where a