I am trying to intuitively understand a theorem on $p$-values in Wasserman's book.

Let $X$ be a random variable. The theorem says that if $T(X)$ is a statistic such that the null hypothesis is rejected for $T(X)$ large, i.e. $T(X) > c$ for some constant c, then for an observation $x$, the p-value is the supremum of $P(T(X) \ge T(x))$ among all parameter values in the null hypothesis.

So if the p-value is small, it means that the observed $T(x)$ is pretty large, i.e. it lies along the "extreme" end of the range of values of $T(X)$. Since the observed $T(x)$ is large, according to the rejection rule in the theorem above, we reject the null hypothesis.

Is my understanding correct as to why a small p-value based on the above definition implies rejecting the null hypothesis?

Edit: What I'm really confused about is : what is the importance of the rejection rule being $T(X)>c$? Why can't the rejection rule be $T(X)<c$ with this definition?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.