I want to calculate the effect size for a pretest-posttest-control group design and the variance of the effect size.
The equations are d_ppc2 and the variance of d_ppc2 from Morris (2008):
https://doi.org/10.1177/1094428106291059
In studies the sample size between pretest and posttest varies because there are some missings (withdrawal, dropouts …). So I am unsure which is the correct sample size to choose. I have not found any answers to my question in the literature so far.
n_c: sample size for control group; n_t: sample size for treatment group
Questions
Hello Kelvin,
For your specific questions:
1. The paper appears to presume that the pretest n and posttest n are the same for a given group. As the ES is conditioned on pretest variation, I'd use the pretest n if it were the case that posttest n was different (you may wish to evaluate the ES using only the matchable cases, which would generally be a subset of the pretest n for a group; if it's the case that your ES estimates are notably different, you'd want to make note of that).
2. I believe the same reasoning (as #1) would apply here.
Good luck with your work.
If cases can be matched from pre-test to post-test, then I would analyze only cases with both pre and post measures. The n is then the number of cases with complete data.
Thanks David Morse and Paul E. Spector, I really appreciate your help.
Unfortunately, I do not have access to the data sets, I only have the descriptive statistics.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size