5
$\begingroup$

Is the delta method valid also for non-normal variables?

Claim:

Let $\sqrt{n}(\hat{X}_n-\theta) \xrightarrow{d} \hat{f} $.

With $\hat f$ having a finite distribution.

Then for every $g$ such that $g'(\theta)$ is nonzero we have:

$\sqrt{n}(g(\hat{X}_n)-g(\theta)) \xrightarrow{d} g'(\theta)\hat{f} $.

It seems to me that the standard proof:

https://en.wikipedia.org/wiki/Delta_method

applies also here, the only point that we need to assess that $\hat{X}_n \xrightarrow{P} \theta$.

I have also the impression that also $\sqrt{n}$ can be substituted by any increasing function of $n$.

Am I thinking wrong? Where can I find such generalizations of the delta method?

$\endgroup$

1 Answer 1

3
$\begingroup$

The generalization is true and the standard reasoning presented in the Wikipedia page can be applied.

Confirmation can be found for example in these notes:

http://sites.stat.psu.edu/~dhunter/asymp/fall2006/lectures/ANGELchpt05.pdf

$\endgroup$

Your Answer

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

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.