Hi,
I am running an analysis for a experimental study and would like to get some opinions about the most appropriate method.
My research question goes: whether there is a significant difference between the control group and test group's post-test measures (significant change over time) on physical scores, emotional scores and motivational scores.
Probably no need to mention, but the pre-test and post-test values between subjects are not directly comparable, because of differences between groups' starting levels. I am only interested if the intervention has significant effect on the mean post-test difference between groups, and if so, what measures are affected.
I was quite confident to use repeated measures MANCOVA, according to similar study designs, but the more I research the more I am doubting whether ANOVAs or paired samples t-tests were more appropriate. I would love to receive some feedback and detailed advice about the assumption tests, analysis and what to input as dependent variables, fixed variables, covariates etc. I read that if the data is normally distributed I should use paired samples t-test, but if the normality is violated Wilkinson signed rank test and ANCOVA (in case the pre-test scores significantly differ across groups).
Should I run paired samples t-tests for normally distributed data and something else for the not normally distributed data? Is it not possible run one analysis only or do I need to repeat manually the tests for different components?
Thank you so much in advance, I am a bit confused because of the numerous measure pairs and assumptions for different tests. Would really appreciate some help!
-Emilia
I am running an analysis for a experimental study and would like to get some opinions about the most appropriate method.
- I have two groups: control (no treatment) and experimental groups (intervention)
- I test all individuals before the experiment and after (pretest-posttest)
- I collect 10 different measures from each participant before and after; 5 physical condition measures, 4 emotional state measures and 1 motivation/productivity measure, so the total of 20 measures for each participant.
My research question goes: whether there is a significant difference between the control group and test group's post-test measures (significant change over time) on physical scores, emotional scores and motivational scores.
Probably no need to mention, but the pre-test and post-test values between subjects are not directly comparable, because of differences between groups' starting levels. I am only interested if the intervention has significant effect on the mean post-test difference between groups, and if so, what measures are affected.
I was quite confident to use repeated measures MANCOVA, according to similar study designs, but the more I research the more I am doubting whether ANOVAs or paired samples t-tests were more appropriate. I would love to receive some feedback and detailed advice about the assumption tests, analysis and what to input as dependent variables, fixed variables, covariates etc. I read that if the data is normally distributed I should use paired samples t-test, but if the normality is violated Wilkinson signed rank test and ANCOVA (in case the pre-test scores significantly differ across groups).
Should I run paired samples t-tests for normally distributed data and something else for the not normally distributed data? Is it not possible run one analysis only or do I need to repeat manually the tests for different components?
Thank you so much in advance, I am a bit confused because of the numerous measure pairs and assumptions for different tests. Would really appreciate some help!
-Emilia