What does "Non-Inferiority Test" mean?

What is a Non-Inferiority Test?

A non-inferiority test is such an online experiment in which we have reason to implement a tested variant as long as it is not substantially worse than the control. In particular, we would consider it a "winner" so long as it is no worse than a specified non-inferiority margin m.

This might be due to different reasons such as testing speed requirements, side benefits such as lower maintenance costs or improvements in other metrics of interest. Non-inferiority tests are often performed for secondary KPIs such as site speed, email unsubscribe rate, etc.

In contrast, in the more common superiority test we would be happy only if we can reject a null of μ ≤ 0 or even μ ≤ ε where ε is a difference that we must surpass in order to justify implementing the tested variant.

Like this glossary entry? For an in-depth and comprehensive reading on A/B testing stats, check out the book "Statistical Methods in Online A/B Testing" by the author of this glossary, Georgi Georgiev.

Articles on Non-Inferiority Test

The Case for Non-Inferiority A/B Tests
blog.analytics-toolkit.com