Statistics

Multiple testing

Çoklu test
Advanced3 min read
My reading list ↗

Reading progress stays in this browser only and is removed when you clear site data.

What does it mean?

Testing several hypotheses can increase the chance of at least one false positive in the family compared with a single test. Define the family and plan error control.

Think through an example.

An everyday example

If all nulls are true and 20 tests are independent, testing each at alpha=0.05 gives a probability of at least one false positive of 1−0.95²⁰, about 64%.

Test your understanding.

True or false?

With 20 independent tests, the family’s false-alarm risk remains exactly 5%.

Make the next connection

Explore related concepts to deepen your understanding.

Explore further

Sources and further reading

Explore the foundations and methodological details in these resources. Examples on this page were created for illustration.