I am conducting a quasi experiment with two classes: a treatment group and a control group. Both groups will take a pre-test and a post-test. The classes are already in place so they are not randomly assignment.
*Should I test the data for normal distribution and homogeneity or I can assume lack of both?
*Should i use non-parametric test Mann-Whitney U test because the groups are not randomly assigned and I am assuming that they are not normally distributed and homogeneity of variance is violated?
*Should I use ANCOVA to control for covariance?
*Or should I use another design?
Hello Mona,
If you intend to use parametric test(s), you should check the assumptions. So, yes to your first question.
Existing/intact groups may still be compared with parametric tests. Switching to nonparametric doesn't cause the comparison to be more/less equitable.
If you do collect pretest scores, these might be useful as a covariate, to adjust (somewhat) for pre-existing differences in groups.
Of course, a randomized study would be ideal, but it's not always logistically feasible to do so.
Good luck with your work.
Dear Mona,
You have a classic schema of experiment in psychology or pedagogics. If Your measurments are scale You should compare the treatment and control groups twice - before and after training program - by means of Mann-Whitney test. Then You need comparing 2 testing only for treatment group using Wilcoxon signed rank test, then make the same one only for control group.
If You can read russian, I would like to propose my book
http://psychlib.ru/mgppu/SMm-2014/MMa-405.htm#$p1
Best regards, Marina Sorokova
Thank you so much. I appreciate your suggestion. So actually the Wilcoxon singed rank test would allow me to compare the pretest and posttest of one group to show the growth in grades then I can compare the growth of each group to each other. Did I get this right?
There is a very good method of comparing the effect in the intervention group compared to the comparison group: this is the effect size and the improvement index. The method is very well described in WWC Version 4.1 Procedures Handbook on the portal IES What Works Clearinghouse
IES What Works Clearinghouse. Handbooks and other resources. Available at: https://ies.ed.gov/ncee/wwc/Handbooks (Accessed 05.06.2020).
- Badges
- Science topic
- Similar topics
- Correction