Do I use the correct method to perform hypothesis testing?
Message posted by Kathy on August 1, 2000 at 12:00 AM (ET)

My question is: I have 2 groups: experimental group (sample size =18) and control group (sample
size =29). After teaching in using different methods, then, i give "test 1" to experimental
group and control group respectively immediately. Next, I give "test 2" to experimental group
and control group respectively after 5 min. Again, i give "test 3" and "test 4" to experimental
group and control group respectively after 2 weeks. Finally,I give "test 5" and "test 6" to
experimental group and control group respectively after 4 weeks.

The content of "test 1", "test 2", "test 3" and "test 5" are the same, except the order of the
question is different. Moreover, "test 4" and "test 6" are similar themselves, but in different
testing format as compared with "test 1", "test 2", "test 3a" and "test 4a". But these 6 tests
also have the same objective.

Now, i want to perform 6 hypothesis tests.
1) compare whether there is population mean difference between experimental group and control
group in test 1.
2) Compare whether there is population mean difference between experimental group and control
group in test 2.
3) Compare whether there is population mean difference between experimental group and control
group in test 3.
4) Compare whether there is population mean difference between experimental group and control
group in test 4.
5) Compare whether there is population mean difference between experimental group and control
group in test 5.
6) Compare whether there is population mean difference between experimental group and control
group in test 6.

My soloution for these hypothesis testing is:
First, I want to use two independent-samples t test, moreover, the sample size for experimental
group (n=18) and control group (n=29) is small,and the distribution for each group in each test
is not normal. Therefore, the assumption doesn't meet. For this reason, I choose to use
Mann-Whitney Test (non-parametric).

Now, my question is:
1) Do the hypothesis that I state above are same when I use Mann-Whitney Test? That is, can I
still test about the differences of population means?
2)Do I encounter the problem of "multiple comparison"? As i know, multiple comparison problem
arise when there is more than 2 levels. moreover, there are only 2 levels for each hypothesis
testing in my case.
3) If multiple comparison problem doesn't exist, is there any other problem regards this type of

data exists? How to solve it and what is the name of this problem?

READERS RESPOND:
(In chronological order. Most recent at the bottom.)

Re: Do I use the correct method to perform hypothesis testing?
Message posted by nancy diehl on August 10, 2000 at 12:00 AM (ET)

It would appear that an ANOVA is the way to analyze this data