in the male group the number of participants is 22 while the female group is 119. with unequal sample size can I use the mann-whitney test, or are there other alternative analysis techniques that are more appropriate for me to use?
Hume F. Winzar , your statement is misleading. The power is lower compared to a test with the same total sample size but equal group sizes.
Comparing 22 vs. 119 still has a larger power than comparing 22 vs. 22, but a lower power than 70 vs. 70 (and obviousely also a lower power than 119 vs. 119, but that's trivial).
Yes, you can use unequal sample sizes, but statistical power is reduced as the difference in sample sizes increases.
See this discussion in stackexchange:
https://stats.stackexchange.com/questions/40342/mann-whitney-u-test-with-unequal-sample-sizes
Hume F. Winzar , your statement is misleading. The power is lower compared to a test with the same total sample size but equal group sizes.
Comparing 22 vs. 119 still has a larger power than comparing 22 vs. 22, but a lower power than 70 vs. 70 (and obviousely also a lower power than 119 vs. 119, but that's trivial).
Thanks for pointing that out, Jochen Wilhelm . I agree, my answer was poorly phrased.
The test apparently assigns a rank regardless of the group to which the value belongs. Unequal sample sizes isn't an issue.
Man-Whitney U test used when the parametric assumptions are violated and then u can use above test for comparison of independent sample of unequal size.
A. P. Chaudhary it's important to note that the Wilcoxon Mann-Whitney does not test the same hypothesis as a t-test! It tests the hypothesis that a person at random from one group will have a higher score than a person at random from the other group.
It is also a parametric test, by the way. The once-indispensable textbook by Sidney Siegel entitled “Nonparametric Statistics for the Behavioural Sciences” must be credited with popularising many of the methods we use today, but also popularising the term “nonparametric” for a large body of tests, some of which actually estimate parameters.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size