Please help! We've got lost in the wild jungle of T-TEST!

#1
Hello,

We have been working on an assignment of t-test analysis, but the data we got look very puzzling and perplexing. I really need your help!

The task is to analyze a set of t-tests from a hypothetical study (1 paired t-test and 3 independent t-tests). But, the problem is... the set does not lead us to conclude anything!

Here is the plot: A research group conducted a study on effect of yoga therapy on anxiety. They had two groups, 1 experiment (taking a yoga course) and 1 control. The two groups took pre- and post tests to self-report their levels of anxiety.

As this is a practice of t-test, we got dataset of t-tests, instead of ANOVA, as follows;
1. Pre-test scores between the control & experiment groups;
2. Post-test scores between the two groups;
3. Pre – Post test scores between the two groups
4. Pre – Post test scores of all the participants (paired t-test).

Now, we’ve been struggling with this problem.
1. Pre-test scores between the two groups = No statistical difference
2. Post-test scores between the two groups = Yes, there is a difference.
3. Pre – Post test scores between the two groups = NO DIFFERENCE
4. Pre – Post test scores of all the participants (paired t-test) = There is a difference.

How can we interpret this? Why does the t-test No.3 NOT show difference? In our understanding, if the intervention (yoga) was effective, it should be seen in the results of Test No.2, No, 4, and also No. 3!? :confused:

For the independent tests, No 1 and 3 met the assumption of normality but No.2 did not. Thus, we used Welch’s t-test to analyze the data. The other two independent t-tests met the assumption.

Please help us conclude this result. We’ve been asking people around us but no luck so far…

Thank you for your time in advance.