I performed a rank-based inverse normal (RIN) transformation so that I can run parametric tests and it normalized my data, so I can use Pearson product-moment correlation coefficient, but are there any traps when using RIN?
There are some problems for small sample sizes and complex models.
http://link.springer.com/article/10.1007/s10519-009-9281-0
The authors suggest to consider "monotonic non-rank-based transformations, classic rank-based non-parametric tests, and computer intensive resampling methods".
There are some problems, especially for small sample sizes and complex models.
http://link.springer.com/article/10.1007/s10519-009-9281-0
The authors suggest to consider "monotonic non-rank-based transformations, classic rank-based non-parametric tests, and computer intensive resampling methods".
- Badges
- Science topic