1. What are the three preexperimental designs? These designs are preexperimental because they do not allow for randomization. They are: (1) the one shot case study, (2) the one group pretest-posttest design, and (3) the static group comparison. These designs are called preexperimental because they are non-interpretable, and they are not desirable for our class project. The one shot case study has no pretest, the treatment, and then the posttest. The intent is to show that the posttest is affected by the treatment. But without the pretest, the researcher cannot know for sure. The one group pretest-posttest design has a pretest, then the treatment, and then the posttest. Assuming that the group is being tested concerning low self-esteem, the researcher cannot positively conclude that the treatment led to favorable results. Other things could have caused the subjects' self esteem to rise, such as a historic event, the passage of time, or maturation. The static group comparison is the best way to get around the above problems because it involves the inclusion of a control group. Without randomization, no pretest, then treatment to the experimental group but not to the control group, and then a posttest to both shows that the experimental group ended much better than the control group.

2. What makes the above designs non-interpretable? On the one shot case study, the lack of a pretest keeps the researcher from knowing if the treatment worked. With the one group pretest-posttest design, the researcher cannot positively conclude that the treatment alone led to favorable results. Other things could have caused it. On the static group comparison, the researcher still cannot say the treatment was the true reason for change because of the inability to randomize.

3. What are the characteristics of quasi-experimental designs? The researcher wants to control extraneous variables and to randomize all groups, but this is not always possible. In educational settings, one cannot always randomize. Quasi-experimental design and true experimental design are distinguished by not being able to randomize.

The characteristics of this kind of design are: (1) an inability to randomly assign, (2) one cannot fully control extraneous variables because of the intact groups (educational classroom is hard to control), and (3) it is difficult to do with human studies in educational settings.

4. What are the major quasi-experimental designs? The major designs are (1) nonrandomized control group, pretest-posttest design, (2) counterbalanced designs, (3) one group time series design, (4) control group time series design, and (5) single subject experimental designs.

5. What are the strengths and weaknesses associated with each design? Concerning the nonrandomized control group, pretest-posttest design, this is giving the pretest to the experimental group and the control group, implementing the treatment on the experimental group, and then giving the posttest to both groups. The researcher will, then, look at the differences before and after the treatment.

Concerning the counterbalanced designs, there are four experimental treatments and four groups. Each group gets each treatment at different times. The researcher can see the effectiveness of each treatment with respect to the others. One problem can exist because of a combination of treatments and whether or not they are randomized and in different sequences. This type of design can be quick but also very complicated.

Concerning the one-group time series design, the researcher can get a clear picture of the attribute to be measured, and maturation can also be ruled out. With this design, there are 3 pretests, the treatment, and then 3 posttests. The weaknesses are: (1) there is no control although the subjects serve as own control, (2) mortality (people can die, drop out, or move away), (3) people's scores can change because they become insenstive to the tests over time.

Concerning the control group time series design, there are 3 pretests and 3 posttests treatment is not given to everyone. If the results of all 6 tests are the same and the ones getting treatment show improvement, then the researcher knows that the treatment is good.

Concerning the single-subject experimental design, the researcher uses handicaps to improve certain behaviors. They get a baseline of what is sharing before reward/treatment. Then, they give reward/treatment and measure new level of sharing among children. Then, they remove the treatment/reward to see how that affects sharing. Finally, they return to the baseline to see if the children have learned anything about sharing. If the child has responded above the baseline, then the lesson of sharing has been learned. The strengths of this design are: (1) the subject serves as own control, (2) if reward is not the only motivation, then the researcher should be able to see it, and (3) maturation can be ruled out. The weaknesses are: (1) testing over and over again and (2) mortality.

				Tom of Bethany

Back To TLEE's Home Page

Index to Selected Essays And Book Reviews

Lesson 21. Experimental Designs (Part II)

Send email to: tlee6040@aol.com