Talk:Errors-in-variables models

Latest comment: 8 years ago by ExpertIdeasBot in topic Dr. Gottschalk's comment on this article


Page moved from the “Errors-in-variables model”.

References

edit

There is currently the {{refimprove}} tag on the page. Could somebody elaborate — which exactly statements seems controversial and require additional referencing, or maybe whether the quality of existing references is unsatisfactory? ... stpasha » talk » 02:31, 12 September 2009 (UTC)Reply

Measurement error, attenuation

edit

Somehow Measurement error should be related to this subject and thus to regression dilution and disattenuation, preferably by something more than a "See also" or three.

The proliferation of articles on regression dilution, disattenuation, and correction for attenuation has been mentioned to the statistics project WP:WPSTAT. --P64 (talk) 17:16, 29 March 2010 (UTC)Reply

Regression dilution and disattenuation are just two very narrow topics referring to the old-school methods of correction for measurement error in a simple linear regression. These methods are probably still used in practice, at least judging from the fact that such procedures still exist in modern statistical packages such as Stata. However the generic term “errors-in-variables models” or “measurement error models” refer to the whole variety of methods for dealing with this problem of regression dilution. The difference is the same as between “regression models” and “linear regression”.
As for measurement error itself — this should probably be converted into a disambiguation article. It could refer to either the observational error, or the theory of propagation of uncertainty, or this topic.  // stpasha »  07:17, 30 March 2010 (UTC)Reply

coding issues

edit

The "Motivational example" still needs concise explanation of variance star and what happened to variance epsilon (with TeX). --P64 (talk) 15:15, 30 March 2010 (UTC)Reply

User:stpasha has provided the latter and has also revised some of the comma, single-quote, and double-quote characters inside and outside the |math| code.
I had revised some double-qmarks and added some thin spaces earlier (or tried to do so). Using two different computer stations early and late this afternoon (American EST) I have noticed some differences between them in the displays of both the code and the current version of the article (which I must examine later, no time now). Other editors, too, may find it instructive to compare the last several versions, using different browser settings and wikipedia settings. --P64 (talk) 21:46, 30 March 2010 (UTC)Reply

Second Paragraph needs clarification

edit

It is not clear what is going to 0 in the second paragraph, the bias of what? MATThematical (talk) 18:14, 23 April 2010 (UTC)Reply

It was a typo. Nothing goes to zero, the estimator is simply biased towards zero — meaning that its magnitude is smaller than that of the true parameter.  // stpasha »  18:43, 23 April 2010 (UTC)Reply

Is each of the proposed remedies unbiased and consistent?

edit

The lede and the motivational example identify the problem with errors in right-side variables as one of bias and inconsistency. It's a long article, so maybe I've missed it, but I don't think the article states anywhere whether any of the proposed approaches is unbiased and/or consistent. Duoduoduo (talk) 20:58, 23 February 2011 (UTC)Reply

Every proposed estimator is consistent (otherwise it would never be published), however unbiasedness is probably the luxury none of them can afford.  // stpasha »  02:12, 24 February 2011 (UTC)Reply
If you have references for that, it would be good to put it into the article. Duoduoduo (talk) 20:23, 24 February 2011 (UTC)Reply

direction of bias ambiguous

edit

Is there any evidence for this statement

"(although in multivariate regression the direction of bias is ambiguous)"?

If not, it should be removed. Muhali (talk) 02:04, 18 April 2012 (UTC)Reply

Dr. Gottschalk's comment on this article

edit

Dr. Gottschalk has reviewed this Wikipedia page, and provided us with the following comments to improve its quality:


This is a good review of the literature on measurement error that includes both the statistical and econometric literature The econometric literature is sometimes misinterpret but these problems can easily be addressed. https://drive.google.com/file/d/0B8-wWhGFpGYCTnVjdkdXWGhlLU0/view?usp=sharing


We hope Wikipedians on this talk page can take advantage of these comments and improve the quality of the article accordingly.

We believe Dr. Gottschalk has expertise on the topic of this article, since he has published relevant scholarly research:


  • Reference : Alessandro Barattieri & Susanto Basu & Peter Gottschalk, 2010. "Some Evidence on the Importance of Sticky Wages," Boston College Working Papers in Economics 740, Boston College Department of Economics.

ExpertIdeasBot (talk) 20:58, 23 September 2016 (UTC)Reply