Hi ,
in my study I will compare the effectiveness of Al generated feedback and traditional feedback in writing lesson.It will be a quasi-experimental design,but l have 3 classes not two.What should I do? Is it better to work with just 2 classes? I just worry about the sampling size one class consist of 25 students. As there is no random choice in quasi experimental if I add the third class, then if l am right I can not divide the students into a control and experimental group. While I was designing the study first I decided to apply a true experimental design,but l thought that it could be unethical or not generalizable so I changed my mind. I hope I am on the right way. Can you share your thoughts with me?
You can use all three. This would allow you to have two experimental groups (AI-generated feedback vs. traditional feedback) and one control group (possibly receiving no additional feedback change).
I am not sure why randomizing who receives who receives AI versus personal feedback is an ethical issue. If giving different kinds of feedback raises an ethical issue, then that would also apply to differentiating on a class by class basis.
So, I would recommend randomizing all three classes.
Having groups of equal sizes is the most efficient from a statistical perspective, but is not necessary. Sometimes one chooses unequal sizes for pragmatic reasons, such as to help encourage cooperation from people who might feel they are missing out on something.
However, like David L Morgan I don't see what the problem is with an experimental design. If it's unethical to impose the treatment in an experimental design, then it is unethical to impose it in a quasi-experimental design (unless you are giving people a choice, and that would be undesirable because it would bias the samples).
I am curious how you plan to administer the study. For some types of studies, it make sense to divide the students by classrooms (it is hard to teach half the students one way and half the other way). However, when providing feedback, that seems doable on a student-by-student level, in which case you can randomize students within classrooms, allowing equal distributions across both groups.
- Badges
- Science method
- Similar topics
- Methodology
- Research Methodology
- Qualitative Research