I am testing whether a certain learning intervention affects student achievement. I have two groups of students (control and experimental) with 16 students each. I have their pre-test and post-test scores. I intend to compare the two groups' pre-test scores using independent t-test; the two groups' post-test scores using independent t-test as well, the improvement of scores within groups using paired t-tests, and finally compare the two groups' gain scores (post-test score minus pre-test score) using independent t-test. Are the tests I intend to use correct And should I correct for multiple testing? How? (I'm a research newbie). Thank you!
$\begingroup$
$\endgroup$
2
-
$\begingroup$ It seems to me the only test that really counts with respect to your objective is the last one, comparing the two groups' gain scores. If this is so, then no need to correct for multiple testing (IMO). If the other tests are important in some way, though, you should definitely correct for multiple testing. You might also think about using a nonparametric alternative to the t-tests, e.g., the Wilcoxon test: stats.stackexchange.com/questions/19681/… $\endgroup$– jbowmanCommented Mar 5, 2018 at 17:35
-
$\begingroup$ Since I want to know if the intervention really works, I have to check whether the two groups differed in student achievement (post test scores). I intend to compare their pre-tests as well to check if they have a pretty much "equal" level of understanding of the topic before the intervention was used. Can you please tell me how I can correct for multiple testing? Thank you $\endgroup$– Teacher ResearcherCommented Mar 6, 2018 at 0:36
Add a comment
|