Parametric vs Nonparametric
Posted on * • 1 minutes • 107 words
Parametric Methods:
- Rely on assumptions about the shape of the distribution.
- Assume that in some way the underlying population and its parameters (means and variances) follow a normal distribution.
Nonparmetric Methods:
- Rely on no or few assumptions about the shape of the underlying distribution.
- Most often used to analyse data which do not meet the distributional requirements of parametric methods - skewed data.
Examples:
| Analysis | Parametric | Nonparametric |
|---|---|---|
| Comparing two independent groups | t-test | Wilcoxon rank-sum test |
| Comparing measurements taken twice (case control) |
Paired t-test | Wilcoxon signed-rank test |
| Comparing three or more independent groups | ANOVA | Kurskal-Wallis test |
| Degree of association between variables | Pearson coefficient of correlation |
Spearman’s rank correlation |