Hello eveyone. I would like to know your oppinion on the following. Do you think that it is adequate to compare two groups with very different sample size and non normal sample distribution? for example. n1=57; n2=7. Many thanks in advance for your answers
José Luis Palazón F , you should not compare them using parametric tests. Non-parametric tests are more robust and a non-normal distribution will not be as problematic. Mann-Whitney U test can be a solution if you have two levels in the IV, or Kruskal Wallis if you have more. That'd be a simple way to compare these tests with t-tests o ANOVAs, respectively.
José Luis Palazón F It is not recommended to compare the mentioned samples with sample sizes using parametric tests. Chi square or Mann-Whitney U test can be a solution. If the data is itself a non parametric data then you can also use Wilcoxon rank sum test to compare means.
In this instance a wilcoxon rank sum test is what I have used. Here is a link that explains the test in more detail:
https://sphweb.bumc.bu.edu/otlt/mph-modules/bs/bs704_nonparametric/bs704_nonparametric4.html#:~:text=The%20Mann%20Whitney%20U%20test%2C%20sometimes%20called%20the%20Mann%20Whitney,populations%20have%20the%20same%20shape).
Given non-normal data, two independent sample t is not suitable. Therefore, Mann-Whitney U test (also called Wilcoxon rank-sum test), the non-parametric version of the two-sample t test is the most appropriate.
- Badges
- Science topic
- Similar topics
- Cognitive Science
- Thinking