Rasmussen College Quasi Experimental Design and Single Case Evaluation Discussion
1) Create a nonequivalent comparison groups design for evaluating the effectiveness of a parent education program for parents at high risk for child abuse. What would you do to assure the readers of your study that the threat of selection bias seems remote? Provide a description of the dependent variable, and when and how it would be measured.
Rubin & Babbie (2017, p. 273) explain that in a nonequivalent comparison design, the researcher is unable to randomly assign participants to groups, so a comparable group that appears similar to the research group is used as a comparison. This group is referred to as the comparison group instead of the control group.
For the example above, the most comparable group to parents participating in the group (the research group) would be parents deemed at risk that are not participating in the group (comparison groups).
While differences in motivation would undoubtedly exist in the groups, Rubin & Babbie (2017, pp. 274-275) discuss the importance of removing the threat of selection bias through various measurement options. One measure that would be a valid option for the example above by using the switching replication design. This design would allow for both groups to receive the treatment option (an ethical concern) but evaluated separately to determine the effectiveness of the intervention.
In the example above, the dependent variable would be the score on a personality assessment tool that explicitly measures the ability to control actions and emotions when angry. The assessment would be administered to both groups as a pretest. The intervention would be presented to the research group only, and both groups would again take the assessment (posttest for the research group). The intervention would then be given to the original comparison group. Finally, a third assessment would be administered to both groups that would act as a posttest for the comparison group. If both posttests show positive results after the intervention (parenting education group), it is reasonable to conclude that the intervention is effective.
2) Imagine you have been hired by the Florida Department of Corrections as a consultant to evaluate one of its prisoner reentry programs in which inmates are sent to this residential program before being released to the community. You are not able to randomly assign inmates to participate in this program – so conducting an experimental design is out of question. Design a quasi-experimental study in which you control for as many threats to internal validity as possible. Explain how you are controlling for these threats.
A multiple time-series design study would study the experimental group (prisoners going to the residential program upon release) and a comparison group (prisoners not going to the residential treatment upon release).
Both groups will be administered multi pretests at given time points before the experimental group receives the intervention. As Rubin & Babbie (2017, p. 275) discuss, multiple pretests will strengthen internal validity by helping to determine that the groups are comparable, that one is not already engaged in a change process. After the experimental group receives the intervention (residential program), both groups will be given posttests over given points in time to determine if the intervention was effective.
Rubin & Babbie (2017, p. 279) also point out that while the multiple time-series design study is not a good as a study where participants can be randomly assigned, it is an improvement to the simple time-series design. By using the multiple pretests and posttests application, the researcher accounts for history as an internal threat by measuring reactions over time. If a participant experiences an external event that affects the responses to either test, it will be evident in the series when that event likely occurred. Under this same premise, the multiple time-series design also controls for maturation as a threat to internal validity.
3) Select some aspect of your own behavior that you would like to improve and develop a plan to improve it. Design a single-case experiment. Discuss the potential measurement problems of reactivity and bias.
The behavior being modified in this single case experiment is nail-biting. I will use a single case withdrawal/reverse design (ABAB) to lessen the frequency of the behavior. As Rubin & Babbie (2017, p. 307), this design allows for a second baseline before the intervention is reintroduced and gives more plausibility to the fact that an extraneous event did not cause the change.
The baseline will be determined by a 3-day period where self-monitoring and obtrusive observation (husband noting that I am biting my nails as I am not always aware I am doing it) are used to determine the frequency of incidents.
The intervention will be to place bitter-tasting nail polish on my fingernails for one week.
After removing the nail polish, we will use the same methods used in collecting the previous baseline data to compile a second baseline over another three days.
Finally, the intervention (nail polish application) will be administered again for one week.
As Rubin & Babbie (2017, p. 263) note, the use of obtrusive observation lends to research reactivity in that the participant is keenly aware of being observed and may inadvertently act in ways to meet experimenter expectation. In the case of this experiment, it is highly likely that I will want to avoid any potential conflict with my husband over his willingness to repeatedly point out my bad habit ( )
Rubin & Babbie (2017, p. 302) also note that self-monitoring is subject to reactivity in that the self-measurement process can bring about change in the targeted behavior. In consciously working on recording how many times I’m biting my fingernails, I may start to do it less often as I am more aware of the act.
4) Think of a particular case or intervention that has piqued your curiosity about practice effectiveness. Design a single-case experiment that is relevant to that case or intervention. Try to design it in a way that would be feasible to implement.
As a hospice volunteer manager, one of the services provided to hospice patients through my department was pet therapy visits. This single case experiment will be designed to determine if the pet therapy visits are effective in decreasing depression in the patient receiving the visit
A basic single-case design (AB) study will be used to determine the effectiveness of the intervention.
A baseline will be obtained through the use of a scale that measures feelings of depression. After completing the assessment, the patient will receive a pet therapy visit within three days of completing the initial baseline assessment. The shorter the amount of time between the two is better to avoid the internal validity threat of history and an extraneous event that could be the cause of less depression Immediately after the visit, the depression scale will be administered again to determine if the pet therapy visit was effective in lessening the feelings of depression in the patient.
Rubin, A., & Babbie, E. R. (2017). Research methods for social work (9th ed.). Boston: Cengage Learning.
Create a nonequivalent comparison groups design for evaluating the effectiveness of a parent education program for parents at high risk for child abuse. What would you do to assure the readers of your study that the threat of selection bias seems remote? Provide a description of the dependent variable, and when and how it would be measured.
In designing the non-experiment comparison group experiment, the researcher would select two groups of parents from two schools. Parents from school A and parents from school B. They are parents of children at risk of child abuse. The researcher would take baseline data to ensure that the characteristics of the parents, including demographic characteristics and risk of child abuse, are similar in both groups. It is possible to locate similar individuals within the community with similar characteristics and select them to act as the comparison group. Parents of children in School A would receive the education program, while those with children in School B would not receive the intervention. The researcher would then collect baseline data on knowledge about child abuse from both groups, then implement the education program for parents of children in School A, and compare outcomes in both groups to determine the effectiveness of the intervention.
Since there is no randomization, there is a high risk of selection bias. The addition of a control group strengthens the causal inference, and it is possible to overcome selection bias. One can assure readers of the study that there is no threat of selection bias by showing that the effect in the intervention group was in the intervention group, rather than the control (Salkind, 2010). Besides, one can report the approaches used to select participants to show that the researcher carefully selected participants to ensure proper matching. The dependent variable would be parental knowledge and attitude towards child abuse. When experimenting, a questionnaire would be developed to capture parental knowledge and attitudes at baseline. During the implementation of the program, both the control and the group receives education intervention.
Imagine you have been hired by the Florida Department of Corrections as a consultant to evaluate one of its prisoner reentry programs in which inmates are sent to this residential program before being released to the community. You are not able to randomly assign inmates to participate in this program – so conducting an experimental design is out of question. Design a quasi-experimental study in which you control for as many threats to internal validity as possible. Explain how you are controlling for these threats.
The study would utilize a non-equivalent control/comparison design to determine the effectiveness of the program. The first phase of the study would involve collecting data on the baseline characteristics of the inmates, including demographics and recidivism rates in two groups of inmates, those selected to undergo the program, and those who do not go through the program. Those who do not undergo the program would be the control, while those undergoing the program would be the experimental group. Data on recidivism rates among the two groups would be compared over time, for example, one year, to determine the impact of the residential program.
Several strategies would be implemented to control bias. First, the study would incorporate a control group or those who do not undergo the program to reduce bias associated with historical or seasonal bias (Schweizer, Braun, & Milstone, 2016). Furthermore, a comparison of recidivism rates among inmates who undergo the program, and those who do not would strengthen the causal inference. Second, the researcher would use matching to ensure proper selection of the control group, through the use of demographic data, and prior recidivism rates. The third approach would be to conduct repeated data over a given period to determine outcomes or utilize an interrupted time-series design. Such a design would allow for the control of trends in maturation and seasonality (Schweizer, Braun, & Milstone, 2016). For example, one would collect data at baseline, sixth months after the interventions, and also collect data from other groups that undergo the program, after the initial study.
Select some aspect of your own behavior that you would like to improve and develop a plan to improve it. Design a single-case experiment. Discuss the potential measurement problems of reactivity and bias.
I believe that one of my weaknesses is poor social communication skills. The proposed intervention would be a ten-hour interview session with a social communication skills expert. Each session would last two hours and involve activities such as mentoring, direct instructions, and role-modeling. Before participating in the interview session, I would assess my interpersonal communication skills using the Conversational Skills Rating Scale developed by Spitzeberg (1995). The scale has 25 items that capture four skills clusters, coordination, expressiveness, composure, and attentiveness. To determine the effectiveness of the intervention, I would compare the baseline data from the scale with my score after participating in the interview. An independent observer would be selected to evaluate my skills using the scale prior and after the intervention to determine the effectiveness.
Since the study would not involve randomization, there is the risk of response-guidance bias and regression artifacts (Krasny-Pacini & Evans, 2018). Furthermore, since I would be aware that I will be undergoing evaluation, there is a high likelihood that I would alter my performance, resulting in reactivity bias. Such bias may affect the ability of the independent observer to assess the effectiveness of the intervention.
Think of a particular case or intervention that has piqued your curiosity about practice effectiveness. Design a single-case experiment that is relevant to that case or intervention. Try to design it in a way that would be feasible to implement.
Child abuse and neglect is a significant problem in the U.S.A and have negative social-emotional, physical, and cognitive consequences of children (Prevent Child Abuse America, 2020). Multidisciplinary resolution approach interventions are effective in reducing child abuse. The proposed study is a single-case-experimental design with two phases, Phase A, which involves the collection of data about specific signs of child abuse, based on the input of informants, and Phase B, which will include the implementation of the intervention. The resolution approach intervention will consist of the creation of a safe environment for the children, treatment of psychological trauma, if identified, and monitoring of the children’s safety. Participants will be families of children aged between 6 and 12 years who have show specific signs of abuse. Informants from a community setting will be selected to identify families with children who have experienced neglect or abuse. The primary outcomes will be incidents of child abuse and neglect, while secondary outcomes will be parental-relationship, parental stress, and a child’s behavioral and emotional problems. A pre-test-post-test design will be utilized, where baseline data of both primary and secondary outcomes will be compared to data after the intervention at 3rd, 6th, and 12th month.
Krasny-Pacini, A., & Evans, J. (2018). Single-case experimental designs to assess intervention effectiveness in rehabilitation: A practical guide. Annals of Physical and Rehabilitation Medicine, 61(3), 164-179.
Prevent Child Abuse America. (2020, July 4). The continuum of therapeutic approaches towards prevention of abuse and neglect- resolution. Retrieved from Prevent Child Abuse America: https://preventchildabuse.org/resource/the-continu…
Salkind, N. (2010). Encyclopedia of Research Design. Thousand Oaks, CA: Sage.
Schweizer, M., Braun, B., & Milstone, A. (2016). Research methods in healthcare epidemiology and antimicrobial stewardship- Quasi-experimental designs. Infect Control Hosp Epidemiology, 37(10), 1135-1140.
Spitzberg, B. (1995). CSRS, The conversational skills rating scale: An instructional assessment of interpersonal competence. National Association of Communication.