Null hypothesis

Also known as: H0, H_0

The null hypothesis $H_0$ is the default claim under test — often “no effect” ($\beta=0$, $\mu=\mu_0$). The alternative $H_1$ is what you entertain when rejecting $H_0$.

Tests control Type I error under $H_0$. Failure to reject is not proof that $H_0$ is true. Report the estimate, SE / CI, and $p$-value, not only a reject/accept bit.

← Back to CS Glossary