Significance level
Significance level is the maximum probability of wrongly rejecting a true null hypothesis that a researcher will tolerate. Denoted alpha, it is set before the test and conventionally fixed at 5% or 1%.
Also known asalpha
FrameworkHypothesis testing
See it move
A normal distribution curve plots test statistic on the horizontal axis against probability density on the vertical axis, with the right-hand tail beyond the critical value shaded. That shaded area equals α = 5%, the pre-committed significance level and the maximum tolerated probability of rejecting a true null hypothesis — a Type I error. The significance level is fixed before any data are collected, ensuring the decision rule cannot be revised to fit a convenient result.
Check yourself
A quality analyst sets α = 0.05 before testing whether a production line is out of specification. The test returns a p-value of 0.03. What does α = 0.05 mean, and what is the correct conclusion?