An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20090210/a0da5507/attachment-0002.pl>
Help regarding White's Heteroscedasticity Correction
4 messages · Kishore, John C Frain, John Fox +1 more
Generally in the presence of heteroskedasticity of unknown form OLS
produces consistent estimates of your regression coefficients. The
estimates of standard errors are biased in the presence of
heteroskedasticity, White's procedure is a way of producing
consistent estimates of the standard errors. It does not change the
estimates of the coefficients. It does not change the residuals.
Patterns in your residuals may show up as heteroskedasticity when
tested but they may be an indication of wrong functional form or of
missing variables or of some other form of misspecification.
Best Regards
John
2009/2/10 Kishore <gladikishore at gmail.com>:
Hi I am actually running the White test for correcting Heteroscedasticity. I used sandwich() & car(), however the output shows the updated t test of coefficients, with revised Standard Errors, however the estimates remained same. My problem is that the residuals formed a pattern in the original regression equation. After running the White's test, I got some new standard errors - but from here I didn't understand how to plot the residuals (or) how to correct the estimates?? Can some one direct me in this regard.. Best, Kishore/.. http://kaykayatisb.blogspot.com [[alternative HTML version deleted]]
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
John C Frain Trinity College Dublin Dublin 2 Ireland www.tcd.ie/Economics/staff/frainj/home.html mailto:frainj at tcd.ie mailto:frainj at gmail.com
Dear Kishore, Yes, White's heteroscedasticity-consistent standard errors are just that -- standard errors for the OLS coefficients that are consistent in the presence of heteroscedasticity. The coefficients themselves don't change. There is an issue here: although the standard errors and OLS coefficients are consistent, the OLS estimates lose efficiency. If you know that pattern of heteroscedasticity, then you might get more efficient estimates by taking it into account, e.g., in weighted-least-squares regression, or, if the residual spread increases with the fitted values, by transforming the response. I hope this helps, John ------------------------------ John Fox, Professor Department of Sociology McMaster University Hamilton, Ontario, Canada web: socserv.mcmaster.ca/jfox
-----Original Message----- From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]
On
Behalf Of Kishore Sent: February-10-09 9:19 AM To: r-help at r-project.org; r-help at stat.math.ethz.ch Subject: [R] Help regarding White's Heteroscedasticity Correction Hi I am actually running the White test for correcting Heteroscedasticity. I used sandwich() & car(), however the output shows the updated t test of coefficients, with revised Standard Errors, however the estimates remained same. My problem is that the residuals formed a pattern in the original regression equation. After running the White's test, I got some new standard errors - but from here I didn't understand how to plot the residuals (or) how to correct the estimates?? Can some one direct me in this regard.. Best, Kishore/.. http://kaykayatisb.blogspot.com [[alternative HTML version deleted]]
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
or simultaneously estimate the coefficients and variance structure via nlme::gls and its 'weights' argument...
On Tue, Feb 10, 2009 at 7:57 AM, John Fox <jfox at mcmaster.ca> wrote:
Dear Kishore, Yes, White's heteroscedasticity-consistent standard errors are just that -- standard errors for the OLS coefficients that are consistent in the presence of heteroscedasticity. The coefficients themselves don't change. There is an issue here: although the standard errors and OLS coefficients are consistent, the OLS estimates lose efficiency. If you know that pattern of heteroscedasticity, then you might get more efficient estimates by taking it into account, e.g., in weighted-least-squares regression, or, if the residual spread increases with the fitted values, by transforming the response. I hope this helps, John ------------------------------ John Fox, Professor Department of Sociology McMaster University Hamilton, Ontario, Canada web: socserv.mcmaster.ca/jfox
-----Original Message----- From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]
On
Behalf Of Kishore Sent: February-10-09 9:19 AM To: r-help at r-project.org; r-help at stat.math.ethz.ch Subject: [R] Help regarding White's Heteroscedasticity Correction Hi I am actually running the White test for correcting Heteroscedasticity. I used sandwich() & car(), however the output shows the updated t test of coefficients, with revised Standard Errors, however the estimates remained same. My problem is that the residuals formed a pattern in the original regression equation. After running the White's test, I got some new standard errors - but from here I didn't understand how to plot the residuals (or) how to correct the estimates?? Can some one direct me in this regard.. Best, Kishore/.. http://kaykayatisb.blogspot.com [[alternative HTML version deleted]]
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. ______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.