MerQuR+Statistics Glossary

Statistics Glossary → Type I and Type II errors

Type I and Type II errorsconcept

In a hypothesis test, rejecting a true null hypothesis is a Type I error, and failing to reject a false null hypothesis is a Type II error.

A hypothesis test can go wrong in two ways. Declaring an effect when there is none is a Type I error (a false positive); its probability is capped by the significance level α. Missing an effect that really exists is a Type II error (a false negative); its probability is written β, and 1 - β is the power of the test.

For a fixed sample size, lowering α raises β: a more cautious test misses real effects more often. The main way to reduce both errors at once is a larger sample. When many tests are run, the chance of at least one false positive grows quickly even if each test uses level α, which is why the familywise error rate or the false discovery rate has to be controlled.

A p value is not the probability of a Type I error; it is the probability of a result at least this extreme if the null hypothesis is true.

An everyday example

Think of a fire alarm. An alarm that goes off when there is no fire is a Type I error; one that stays silent during a real fire is a Type II error. Making the alarm less sensitive cuts down false alarms but raises the risk of missing a real fire.

In MerQur

In Moran's I and Getis-Ord Gi* Hotspot each location is tested separately, so the risk of clusters or hot spots that look significant purely by chance (Type I errors) matters especially; the 'Multiple-comparison correction' option (None, FDR (Benjamini-Hochberg), Bonferroni) classifies with corrected p values to limit false positives.

Analyses where this concept appears

Moran's I (Spatial Autocorrelation)Getis-Ord Gi* Hotspot

Related concepts

Significance level (α)Multiple comparisons and the familywise error rate (FWER)Observed (post hoc) power