Talk:Durbin–Wu–Hausman test

Latest comment: 2 years ago by Vardhanisation in topic Define the estimators

Text merged from Talk:Hausman specification test

edit

Merge?

edit

I don't speak econese, but is this the same thing as Hausman test? Her Pegship (tis herself) 22:17, 9 June 2008 (UTC)Reply

It really should be merged. Or, even better, deleted, as it barely makes sense. —Preceding unsigned comment added by 158.143.50.170 (talk) 12:31, 11 February 2009 (UTC)Reply

Merge the two (there's nothing in the other one) and I find it funny that it's too technical for a general audience. Well, no duh. The Hausman test is a comparison of a fixed effects vs random effects model. I'm a 3rd year PhD student in economics and I barely understand it. Somethings will never be accessible to a "general audience". 128.195.110.44 (talk) 00:15, 10 June 2008 (UTC)Reply

Merge. But "Somethings will never be accessible to a "general audience". " ... the article needs more context, and an explanation of where the variance matrix is supposed to come from ... and something to distinguish is from other test statistics of a similar form: see for example Mahalanobis distance. Melcombe (talk) 09:07, 10 June 2008 (UTC)Reply

Merge! —Preceding unsigned comment added by 169.237.158.159 (talk) 19:08, 10 June 2008 (UTC)Reply

Wu too?

edit

In virtually every place I've encountered this test, it is ubiquitously known as the Hausman-Wu test. See the reference: Wu, D. Min (1973). Alternative Tests of Independence Between Stochastic Regressors and Disturbances. Econometrica, 41, 733-750. This predates Hausman by about 5 years, and in fact I have seen this called the Wu-Hausman test in many places. — Preceding unsigned comment added by P4wnc6 (talkcontribs) 02:40, 8 February 2012 (UTC)Reply

It would be much more meaningful if you gave sources that actually use the term "Wu-Hausman test". Melcombe (talk) 08:30, 8 February 2012 (UTC)Reply
Just do a Google search for Wu-Hausman test or Hausman-Wu test. Basically every serious academic source includes Wu somewhere in the name. It's so ubiquitous that I wouldn't even know where to start with a "source" that calls it that ... try any econometrics article in the last 10 years that calls for use of the test. For example, just look at the hits for Wu-Hausman test when searching Google books: [1] Most papers just refer to it this way since it is stated this way in most books. — Preceding unsigned comment added by P4wnc6 (talkcontribs) 04:01, 10 February 2012 (UTC)Reply
The point is that there are no immediate citation in the article to any such usage, or to meet the requiremnents of WP:N. Melcombe (talk) 09:03, 10 February 2012 (UTC)Reply
This has been fixed now. The page title is now Durbin-Wu-Hausman test. Vardhanisation (talk) 16:19, 2 May 2022 (UTC)Reply

Define the estimators

edit

In the current article, b0 and b1 aren't even defined! Btyner (talk) 17:01, 26 April 2015 (UTC)Reply

The exact definitions of b0 and b1 are not needed as long as one of them is decidedly more efficient than the other — correct me if I'm wrong. They both estimate b in y = Xb + e. This comment was made in 2015 so I believe it has been fixed sometime in the last seven years. Vardhanisation (talk) 16:24, 2 May 2022 (UTC)Reply

Derivation is not correct

edit

Sign of variance changes after applying "covariance is 0". Couldn't find a reference myself to fix it at the generality level of the article though. — Preceding unsigned comment added by Victordamatta (talkcontribs) 14:07, 28 July 2020 (UTC)Reply

it's correct, just confusing the way it's worded. the covariance it's referencing is cov(q,b1), i.e. "covariance of an efficient estimator with its difference from an inefficient estimator is zero" ... b1 is the efficient estimator, b0 is the inefficient estimator, and q=b0-b1. the proof is in the hausman reference — Preceding unsigned comment added by 2601:401:180:6FE0:79D9:3CB2:C8E2:EE6E (talk) 15:18, 22 February 2022 (UTC)Reply