Quasi-Experimental Design Considerations
A quasi-experimental design looks a lot like an experiment where you are comparing groups, except that you do not randomly assign individuals to the different conditions (e.g., treatment A versus treatment B) so you can compare outcomes. The people in each of the groups are already there. A key risk to validity of results is that there may be other influences on the outcomes (e.g., reasons people chose one treatment over another that might have motivated their outcomes, such as more positive expectations or other biases). Without knowing what these could be, it is difficult to determine if the treatment would work as well for others.
Again, think about the kind of missions and goals of those who wrote your RFP (e.g., to change behaviors). What might be a program evaluation that could be useful to them, at some point, that would employ a quasi-experimental design?
For this Discussion, assume that a quasi-experimental design could be a viable option for the evaluation outlined in your RFP.
- Review Chapter 10 from your course text listed in this weekâ€™s Learning Resources.
- Review U.S. Department of Education Institute of Education Sciences (2016) from this weekâ€™s Learning Resources.
- Consider how you might develop a program evaluation based on quasi-experimental design.
- Consider the variables and hypotheses you might want to test as part of your program evaluation. Also, what are potential threats to internal validity that might be associated with this kind of program evaluation design?
Post your comprehensive response to each of the following:
- Briefly restate the general nature of your teamâ€™s RFPâ€™s program (e.g., school anti-bullying program).
- How might you use a quasi-experimental design to test a question that would be relevant to your program evaluation (e.g., comparing age or gender groups or two preexisting programs)? What is a specific question you could address for your example?
- Describe the specifics of your design for addressing this question. What is your independent variable? What might be a viable dependent variable to answer the question? What hypothesis would you be testing for the proposed evaluation?
- What would be one method you might use to reduce threats to internal validity with this design?