Mann-Whitney is a good choice for small samples, and the normal approximation is considered reasonable for sample sizes > 20 or so.
Similarly, the exact distribution for Kruskal-Wallis is used for small samples and a normal approximation beyond that.
Of course, with very small samples, the test power is limited.
Note, though, that these statistics are testing null hypotheses that are a bit different from a t test, and you should still think about whether the assumptions they require are reasonable for your situation.
--
------Original Message------
Hello, Bonjour, Aloha ;
On our latest research, as Master's Students, we could not be sure Kruskal Wallis or Whitney Mann would be suitable for 30 sample populations.
Do you have any suggestions?
Thnx
B
------------------------------
Baris Orak
------------------------------
#SPSSStatistics