Professional Documents
Culture Documents
Quasi Eksperimental
Quasi Eksperimental
Results from these designs (sometimes called preexperimental) are hard to interpret; they
should not be used if it is possible to use one of the better quasi-experimental or randomized
designs. Unfortunately, these poor designs are relatively common in clinical practice.
One-Group Posttest-Only
Design An example of this design, sometimes referred to as the oneshot case study, would be
some evaluations of a new treatment. The investigator introduces the treatment and uses some
outcome measure to determine the patient’s response to the treatment.
The design is diagrammed as follows: E: X 0
This diagram and those that follow indicate a time sequence. First, all participants are in one
group, the experimental group (E). Then the treatment (X) and, finally, a posttest (0) takes
place. The problem with the design is that it does not satisfy even the minimum condition for
a research problem, which is the investigation of a relationship or comparison. Note that the
intervention is not a variable because there is only one level. Does the one-group posttest-only
design have any value? If nothing else, it provides a pilot test of the feasibility and cost of the
treatment and the outcome measure. It can help get the “bugs” out of a procedure. The
investigator could compare the results to data from an earlier group that had not received the
treatment, but then the design would no longer be a one-group posttest-only design.
One-Group Pretest-Posttest Design This design is diagrammed as follows: E: 0, X 02 For this
design, an observation in the form of a pretest (0,) is given first, then the intervention is given
to all participants, and, finally, a second observation (03 or posttest is recorded. The problem
with this design is that the comparison is not with a second group, a control group. Instead, the
comparison is between the pretest and the posttest within the same group. Because there is no
comparison group, the design is susceptible to most of Cook and Campbell’s (1979) threats to
internal vuldiv. An example of this type of design might be a study of the effects of therapy on
pre- and posttest measures of motor performance on one group of children with physical
disabilities. The history threat means that environmental events are a possible problem in this
design. The lack of a control group prevents the investigator from knowing, for example,
whether other activities happening in the children’s school at the same time as the intervention
might be contributing to any changes. Maturation is also a possible threat to internal validity
because the children are getting older and may be better coordinated and stronger even without
the intervention. Canyovrr ficts are a possible problem in this and all pretest designs because
taking the pretest could influence the posttest. This design could be useful if previous research
had convincingly dcmonstrated that without treatment the problem behavior would not
decrease. In this situation, a successful intervention would be quite credible.