Type I vs Type II Errors

A Type I error is a false positive: concluding there is an effect when there is none. A Type II error is a false negative: failing to detect a real effect. The probabilities are alpha (α) and beta (β), and power is 1 − β.

Definitions

• α is the significance level you choose, often 0.05.
• β depends on sample size, effect size, variability and α. Studies often aim for 80% to 90% power, so β of 10% to 20%.

Trade-offs

Lowering α reduces false positives but increases false negatives unless sample size rises. In screening or safety questions, the cost of each error type differs and may justify different choices.

Multiple testing

Each additional test increases the overall chance of at least one Type I error. Approaches such as pre-specifying a primary outcome or adjusting for multiple comparisons (for example Bonferroni or other methods) help; seek statistical advice for the right method.

Illustrative example

Invented for teaching. If you test 20 independent outcomes at α = 0.05 when no true effects exist, the chance of at least one "significant" result is 1 − 0.95²⁰ ≈ 64%. Reporting that single result as a discovery would be a Type I error. Conversely, a trial of 30 patients may miss a real, moderate effect (Type II error) because it lacks power.

Common pitfalls

• Treating "not significant" as proof of no effect.
• Many unplanned comparisons.

Key takeaways

• Type I = false alarm; Type II = missed signal.
• Control them by pre-specification, adequate power and appropriate adjustment.

This page is general educational information about research methods. It is not medical, legal or regulatory advice; follow your institution's ethics board and your target journal's instructions.