I encountered an integration problem when I read a statistics textbook, which involve the following relationship.

$\int_{-\infty}^{0}\phi(z;\beta,1) dz = 1-\phi(\beta)$

Here $\phi$ is a normal cdf. $\phi(z;\beta,1)$ represents a normal distribution of z with mean $\beta$ and standard deviation 1 and beta is also a random variable.

I wonder how to prove this relationship?

  • $\begingroup$ This should be $\Phi(x\beta)$ in the right member of the first inequality. $\endgroup$ – Stéphane Laurent Jun 30 '17 at 13:06
  • $\begingroup$ I had thought he means both sides are $\Phi$ which is quite hard. $\endgroup$ – Deep North Jun 30 '17 at 13:16

Let $X \sim {\cal N}(\mu, 1)$. Your integral is equal to $\Pr(X \leq 0)$ when $\mu = x\beta$. One has $X = Z + \mu$ where $Z \sim {\cal N}(0,1)$. Thus your integral is equal to $\Pr(Z \leq -\mu) = \Phi(-\mu) = 1 - \Phi(\mu)$.


Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.