Is it possible to conduct Correlation analysis between variables that has pretest and posttest measures? If yes, how? And if no, what kind of analysis should I be conducting?
I would like to see if mindfulness correlates well with psychological wellbeing and perceived stress. However, I collected the data in a pre and post intervention manner (for mindfulness, psychological wellbeing and perceived stress) whereby mindfulness teaching was the intervention. Seeing that I have both pre and post scores, I was confused as to how should I conduct Pearson's correlation to see the relationship between mindfulness and perceived stress as well as mindfulness and psychological wellbeing.
***Some researchers suggested averaging the Pretest & Posttest scores while some have suggested using repeated measures correlation (rmcorr).
Would it work to take the difference between pre-test and post-test scores and then test if the mean equals zero? How you set this up depends partly on your hypothesis.
Ha: There is a difference between pre-test and post-test. In this case you use a two-tailed test.
Ha: After 20 beers the hypothesis is that pre-test > post-test. In this case you use a one-tailed test.
The question seems straight forward, but I should consider that some details are missing. With the right data one could ask if the difference in pre-test and post-test score is related to the pre-test score. If I score a 5 on the pre-test (out of 100), then I have a great deal of room for improvement. If I scored 97 on the pre-test, then I might get a lower score on the post-test simply because I have lost interest/concentration and miss one additional question (maybe I guessed on one question and I guess wrong on the post-test).
Maybe you could try a discriminant analysis? It will show which factors that
has the greatest impact for the parameter you want to test.
Mindfulness teaching was the intervention, i.e. the independent variable (between the pre/before and post/after measures). The logical next step, in my opinion, is to run an ANCOVA (analysis of covariance) with the post measure(s) as dependent variable(s) and the pre measure(s) as covariate(s) for the mindfulness teaching factor. This is a regression, an additive linear model:
post = pre + teaching group (in that order). (You may want to add the interaction factor: 'pre * teaching group').
Most statisticians would agree that change scores are NOT the way to analyze the effect of the mindfulness teaching factor, in this case.
An ANCOVA is essentially an ANOVA in which differences between levels of a factor (i.c., mindfulness teaching) are tested after controlling for (an) other variable(s), termed covariates (i.c., pre measure(s)).
- Badges
- Science topic
- Similar topics
- Biological Science
- Physiology
- Physiological Processes
- Stress