I am working on a project which includes 5 groups ;6 rats in each, making 30 rats as total and I was asking if It's right to run non-parametric tests on this sample size even if the data showed normal distribution
How can you know if the data are drawn from a normal distribution, particularly with this small a sample? (and as Timothy A Ebert says, a Normal distributional conditional on the model, which is presumably the categorical variable for groups).
Non-parametric tests do not assume a distribution. It could be normal or something else. It doesn't matter. The test for normality is for the residuals of the statistical model. If the data appear normally distributed you will have greater ability to detect treatment differences using a parametric test.
How can you know if the data are drawn from a normal distribution, particularly with this small a sample? (and as Timothy A Ebert says, a Normal distributional conditional on the model, which is presumably the categorical variable for groups).
Non-parametric statistical tests are a better option for very small samples. These statistics do not require data normality and can be applied to any data that involves ordinal, interval, or ratio scores.
However, I would question the validity of results from such small samples, regardless of the statistical test you use.
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running