Talk:Latent variable

From Wikipedia, the free encyclopedia

This could do with some links to related topics

I don't think that Latent variable model and Latent variable should be merged - they're different concepts, and neither is obviously a subcategory of the other. The articles currently contain quite different information, so this isn't a case where there is duplication of effort happening. I believe (thought it is not my field of mathematics) that the use of the term Latent variable is much wider than its article implies, and includes situations quite unrelated to Latent variable models. Chrisjohnson 01:38, 8 April 2007 (UTC)

I agree. A Latent variable model is a particular statistical model, while the concept of latent variable is a broad concept of a variable with unobserved outcomes. They aren't the same thing at all. A latent variable model does use the concept in the framework of a statistical model so might be considered a subset of the concept of latent variables, but it's a very specific subset that veers of in it's own direction. The should stay seperate. (Gary Carson)

Niether article is very helpful since they lack references. By comparison, I would say you dont really need one page for "factor" and one for "factor analysis" if you are only going to discuss the factors employed in factor analysis. Both articles are short so they could both just appear on the model page.Phalpin 02:48, 12 November 2007 (UTC)ph

Just a brief note that I entirely agree with the decision not to merge latent variable model with latent variable. It is certainly reasonable to discuss the concept of a latent trait or variable in a discussion of latent variable models. However, the concept of a latent trait is important in its own right. In addition, the concept of a latent trait or variable arguably differs in item response theory compared with factor analysis and other models. As stated above, the concept of a latent trait is general, and it is debatable whether it can be tied to any specific model given any reasonable analysis of literature. I think it is reasonable to expect people to seek answers to what a latent variable or trait is. I agree they should remain separate. I also agree they article needs work, and will try to get to this myself, for one, as time permits. Also, please be aware it is a psychometric topic as well as econometric. Holon (talk) 03:08, 15 January 2008 (UTC)