Abstract: The Jeffreys-Lindley paradox, namely the fact that a point null hypothesis will always be accepted when the variance of a conjugate prior goes to infinity, has often been argued to imply prohibiting the use of improper priors in hypothesis testing. We reevaluate this paradox by considering the role of the prior hypothesis probabilities and obtain a noninformative answer which is equivalent decisionwise to the classical p-value.
Key words and phrases: Hypothesis testing, vague prior, noninformative answer, p-value.