5. Describe each of the following types of designs, explain its logic, and why t

Question

5. Describe each of the following types of designs, explain its logic, and why the design does or does not address the selection threats discussed in Chapter 10 of Trochim, Donnelly, and Arora (2016): a. Non-equivalent control group pretest only b. Non-equivalent control group pretest/posttest c. Cross-sectional d. Regression-Discontinuity Please show the references and web address used for this question.

Trochim, W., Donnelly, J., & Arora, K. (2016). Research methods: The essential knowledge base (2nd ed.). Mason, OH: Cengage.

Explanation / Answer

Trochim and Donnelly (2006):
a) Non-equivalent control group posttest only. The nonequivalent control group posttest only design has at least two nonrandomly selected groups. One of the groups is the experimental group and will be administered the new treatment, while the other group is the control group and will have no treatment, or the standard treatment. If the two groups can be shown to be relatively equivalent the results of this design are very similar to a true experiment, and the comparison between the groups on the posttest should reflect the effect of the treatment (Jackson, 2012). The problem with this design is that it allows very little within the design to determine if the groups are roughly equivalent in the first place, meaning that “you can never be sure the groups are comparable” (Trochim & Donnelly, 2008, p. 211). This means that the design is especially susceptible to selection bias since the groups are not randomly selected, and the groups may not be equivalent. It is possible using this design that the two groups could have different history or maturation during the duration of the program that could impact the posttest, but none of the other multiple-group threats to internal validity are applicable for this design. If participants of each group are not isolated from each other there is also the possibility of the social interaction threats to internal validity; diffusion or imitation of treatment, compensatory rivalry, or resentful demoralization (Trochim & Donnelly, 2008). b) Non-equivalent control group pretest/posttest. The nonequivalent control group pretest/posttest design has at least two nonrandomly selected groups. One group comprises the experimental group, and receives treatment, while the other group is the control group, receiving no treatment or the standard treatment. Each group receives a measure regarding the dependent variables at the beginning of the study, and then another equivalent measure regarding the dependent variables at the end of the study. The addition of the pretest allows the researcher to compare the groups with respect to the dependent variables to determine whether the groups are roughly equivalent. If the groups are equivalent the internal validity of the posttest comparison is increased, and if they are not the posttest scores can be statistically adjusted based on the pretest scores.   
c) cross-sectional. A cross-sectional design studies multiple strata of a population at the same time. For example, a cross-sectional design focused on writing development may sample from 7th, 9th, and 11th grade students. The logic behind a cross-sectional study is to collect as much data regarding the stratum as quickly as possible. This study often does not adequately address selection bias, since the subjects are by definition stratified around some criteria that distinguishes them from the other strata.
   d) regression-Discontinuity. The regression-discontinuity design begins by administering a pretest to all participants, and then relegates them to control or experimental groups based on a specific cutoff score. Those with the greatest need for treatment are administered the treatment, while the other group serves as a control. Unlike randomized experiments this design makes no attempt to equalize the groups. While in other designs nonequivalence is damaging to internal validity, in a regression-discontinuity design it serves to strengthen the internal validity. The reason for this is that the groups are intentionally designed to not be equivalent, but instead have a linear relationship with each other. If during the posttest there is still a linear relationship between the groups, the treatment had no effect. However, if there is a discontinuity between the groups at the point of cutoff, a gap between each group’s linear representations of scores, this indicates the effect of the treatment. This design addresses all threats to internal validity from multiple groups, but can still be affected by social interaction threats that will tend to diminish the effect size.

Related Questions

Navigate

• Browse (All)
• Browse 5
• Subjects

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.