I want to compare 2 independent variables, but the problem is that the first variable is normally distributed, while the second variable is not. What is the appropriate test in this case? Can I use an independent-samples t-test or the nonparametric Mann-Whitney test?
Asked
Active
Viewed 1,923 times
2
kjetil b halvorsen
- 63,378
- 26
- 142
- 467
noor
- 21
- 2
-
What about the two variables do you want to compare exactly? The mean? – Patrick Coulombe Aug 30 '14 at 15:10
-
1How do you know that one is normally distributed? What do you want to compare, exactly? – Glen_b Aug 30 '14 at 16:32
-
I do shapero test for both variables to test normality. – noor Aug 31 '14 at 14:33
-
I want to compare clinical parameter(lipoprotein level) in 2 different group of patients, but the results of the test ,when tested by shapero test and Q-Q plots in SPSS, was normally disterbuted in first group but not normally disterbuted in the second group.this confusing for me and so I need your advice in selecting the right tst. – noor Aug 31 '14 at 14:41
1 Answers
1
It is normally not advised to choose hypothesis test after looking at data, for instance, by a test of normality. If you don't know at the outset that the data will be approximately normally distributed, choose some appropriate test for that situation (before looking at data) and use it. Of course you can do normality testing or other goodness-of-fit testing just to learn for the future.
A discussion about how to choose between t-test or nonparametric is here.
kjetil b halvorsen
- 63,378
- 26
- 142
- 467