Operating with Data - Statistics for the Cardiovascular Surgeon: Part I. Fundamentals of Biostatistics

[...]we could convert the numeric values of age into categorical values, separating them into three groups: children (0-18 years), adult (>18-65 years), and elderly (>65 years); this case represents the conversion of a discrete variable into an ordinal variable. Besides non-normal distributions, non-parametric tests are also indicated for small sample sizes. Examples of non-parametric tests are Wilcoxon tests, the Kruskal-Wallis test, and the Spearman's rank correlation. Since parametric tests are based on normal distributions, they use the mean as a measure of central tendency, while non-parametric tests use the median. [...]parametric tests should be used for normally distributed samples and non-parametric tests for non-normally distributed samples.