Practical statistics in R
A working reference for statistical tests in R: normality checks, one- and two-sample tests, ANOVA, and their non-parametric counterparts. Written after finally understanding why statistics works in practice.
A practical reference for choosing and running statistical tests in R. The core decision is always the same: if data is normally distributed, use parametric tests (more powerful). If not, use non-parametric tests (fewer assumptions). Start by checking normality with a histogram, QQ-plot, and Shapiro-Wilk test, then pick the right test for your design.
The article covers one-sample tests, paired two-sample tests, independent two-sample tests, one-way ANOVA and its non-parametric counterpart (Kruskal-Wallis), and two-way ANOVA for multiple factors and interactions. Each section pairs the parametric and non-parametric alternatives with the R function and the condition that makes each appropriate. Written after finally understanding why statistics works in practice. Published on Dev Genius, August 2022.