03-08-2012 11:37 AM
Thank you very much.
I did so, but the results are different from the SPSS results, which reported by Mann-Whitney Rank Sum test.
What is different between 'Normal Approximation' and ' t Approximation' for the Statistic?
If I have a small sample size (<10), should I use EXACT test?
Thank you again in advance.
03-08-2012 02:53 PM
The differences in approximations are in the documentation. The approximations are for larger samples.
With a small sample the exact test is better.
Lastly, the differences with SPSS may have to do with the use of a continuity correction in the approximations. Again, see the documentation.