Talk:Specification (regression)

From Wikipedia, the free encyclopedia

[edit] Is this encyclopedic?

I'm making a few small corrections to this article, but I have a larger question. Do we really need a separate article about "specifying" a regression model? I think this subject must already be adequately covered in another article (about regression analysis, or regression models, or econometrics, or something like that).

I intend to hunt around a little bit and see if this article can be merged into a truly encyclopedic existing article. DavidCBryant 13:37, 17 May 2007 (UTC)

I think this subject is encyclopedic. Right now the article is a little on the sparse side, but I'm planning to add to it. The specification itself probably doesn't warrant an article all on its own, but there are several related concepts that can be lumped in here: functional form, specification bias/error, and specification tests all need a place in Wikipedia. I think they can be combined here, and it works better than having separate articles for each. There is far too much information between these subjects to merge into the econometrics or regression analysis articles. Heavy textbooks and hundreds of journal articles have been written on regression. Information on it has to be broken out some how, and I think the this set up (specification including functional form, bias, and tests) makes sense.--Bkwillwm 04:35, 18 May 2007 (UTC)