Hypothesis Test
Statistical decision procedure — checks whether an observed deviation is a real effect or chance.
Also known as: hypothesis test, significance test, statistical test
A hypothesis test compares two mutually exclusive statements: the null hypothesis H₀ ("no effect") against the alternative hypothesis H₁ ("effect present"). The test computes a test statistic from the data and derives a p-value; this together with the significance level α decides whether H₀ is rejected.
- One-sample questions — Does a mean differ from a target? → one-sample t-test
- Two-sample comparison — Are two processes/machines equal? → two-sample t or Welch t
- Several groups — Does at least one differ? → ANOVA (parametric) or Kruskal-Wallis (nonparametric)
- Compare variances → Bartlett / Levene test
- Check distribution shape → normality tests (Shapiro-Wilk, Anderson-Darling, D'Agostino-Pearson)
See also
Used in
In the Algorithm Lab
Sources
- Fisher, R. A. — Statistical Methods for Research Workers, Oliver and Boyd, 1925
- Neyman, J. & Pearson, E. S. — On the Problem of the Most Efficient Tests of Statistical Hypotheses, Philosophical Transactions A, 231, 1933
- Wasserstein, R. L. & Lazar, N. A. — The ASA Statement on p-Values, The American Statistician, 70(2), 2016