I have the negative binomial distribution and want to find the fisher information: $I(\theta) = V[\ell'(\theta)]$

How do i calculate this?

I know that the derivative of the log-likelihood is: $\ell'(\theta) = \frac{r}{\theta} - \frac{X - r}{1 - \theta}$

Also i know that $V[\ell'(\theta)] = \frac{V[X]}{(1-\theta)^2}$, but I don't understand how you get to that calculation.

I need to find the Fisher information to calculate Jeffreys prior, but now I am stuck.


1 Answer 1


Ok, now I got it.

$$p(y) = {y-1 \choose r-1} \theta^r(1-\theta)^{y-r}$$

Let's denote $X$ as the random variable, then:

$$\ell(\theta) = C + r \ln(\theta) + (X-r) \ln(1-\theta)$$

$$\ell'(\theta) = \frac{r}{\theta} - \frac{X-r}{1-\theta}$$

$$\ell''(\theta) = -\frac{r}{\theta^2} - \frac{X-r}{(1-\theta)^2}$$

Now, we calculate the Fisher information, remember that $E[X] = \frac{r}{\theta}$

$I(\theta) = V[\ell'(\theta)] = -E[\ell''(\theta)]= -E\left[-\frac{r}{\theta^2} - \frac{X-r}{(1-\theta)^2}\right]$

Everything in the expression is constants except for $X$.

$-E\left[-\frac{r}{\theta^2} - \frac{(X-r)}{(1-\theta)^2}\right] = \frac{r}{\theta^2} + \frac{E[X]-r}{(1-\theta)^2} = \frac{r}{\theta^2} + \frac{\frac{r}{\theta}-r}{(1-\theta)^2} = r\left(\frac{1}{\theta^2} + \frac{\frac{1 - \theta}{\theta}}{(1-\theta)^2}\right)$

Now this gets simplified to:

$r\left(\frac{1}{\theta^2} + \frac{\frac{1 - \theta}{\theta}}{(1-\theta)^2}\right) = r\left(\frac{1}{\theta^2} + \frac{(1 - \theta)}{\theta(1-\theta)^2}\right) = r\left(\frac{1-\theta}{(1-\theta)\theta^2} + \frac{\theta}{\theta^2(1-\theta)}\right) = r\left(\frac{1}{\theta^2(1-\theta)}\right)$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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