Yujie Wang
2011-Oct-10 19:35 UTC
[R] How to test if two C statistics are significantly different?
Hey all, In order to test if a marker is a risk factor, I built two models (using cox proportional hazard model). One model included this marker, and the other is not. Then, I use R package risksetROC to test how much predictive value did the marker add to this model. I get two C statistics by analyzing the linear predictors of the two models into this package. The qustion is How to test if two C statistics are significantly different? Your help will be greatly appreciated! Yujie [[alternative HTML version deleted]]
Eik Vettorazzi
2011-Oct-11 09:25 UTC
[R] How to test if two C statistics are significantly different?
Hi Yujie, there is still a lot of work in progress, I think. As http://faculty.washington.edu/heagerty/Software/SurvROC/RisksetROC/risksetROCdiscuss.pdf states: "[...] for inference and variance estimation, we now suggest bootstrapping [...]". Recently I catched a glimpse on roc.test from the pROC package, they implemented, amongst others, a bootstrap algorithm - maybe this is a start for your own work? Hth. Am 10.10.2011 21:35, schrieb Yujie Wang:> Hey all, > > In order to test if a marker is a risk factor, I built two models (using cox > proportional hazard model). One model included this marker, and the other is > not. > > Then, I use R package risksetROC to test how much predictive value did the > marker add to this model. I get two C statistics by analyzing the linear > predictors of the two models into this package. > > The qustion is How to test if two C statistics are significantly different? > > Your help will be greatly appreciated! > > Yujie > > [[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.-- Eik Vettorazzi Institut f?r Medizinische Biometrie und Epidemiologie Universit?tsklinikum Hamburg-Eppendorf Martinistr. 52 20246 Hamburg T ++49/40/7410-58243 F ++49/40/7410-57790 -- Pflichtangaben gem?? Gesetz ?ber elektronische Handelsregister und Genossenschaftsregister sowie das Unternehmensregister (EHUG): Universit?tsklinikum Hamburg-Eppendorf; K?rperschaft des ?ffentlichen Rechts; Gerichtsstand: Hamburg Vorstandsmitglieder: Prof. Dr. Guido Sauter (Vertreter des Vorsitzenden), Dr. Alexander Kirstein, Joachim Pr?l?, Prof. Dr. Dr. Uwe Koch-Gromus
alanm (Alan Mitchell)
2011-Oct-11 13:30 UTC
[R] How to test if two C statistics are significantly different?
?Hmisc::rcorrp.cens -Alan -----Original Message----- From: Eik Vettorazzi [mailto:E.Vettorazzi at uke.de] Sent: Tue 10/11/2011 2:25 AM To: Yujie Wang Cc: r-help at r-project.org Subject: Re: [R] How to test if two C statistics are significantly different? Hi Yujie, there is still a lot of work in progress, I think. As http://faculty.washington.edu/heagerty/Software/SurvROC/RisksetROC/risksetROCdiscuss.pdf states: "[...] for inference and variance estimation, we now suggest bootstrapping [...]". Recently I catched a glimpse on roc.test from the pROC package, they implemented, amongst others, a bootstrap algorithm - maybe this is a start for your own work? Hth. Am 10.10.2011 21:35, schrieb Yujie Wang:> Hey all, > > In order to test if a marker is a risk factor, I built two models (using cox > proportional hazard model). One model included this marker, and the other is > not. > > Then, I use R package risksetROC to test how much predictive value did the > marker add to this model. I get two C statistics by analyzing the linear > predictors of the two models into this package. > > The qustion is How to test if two C statistics are significantly different? > > Your help will be greatly appreciated! > > Yujie > > [[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.-- Eik Vettorazzi Institut f?r Medizinische Biometrie und Epidemiologie Universit?tsklinikum Hamburg-Eppendorf Martinistr. 52 20246 Hamburg T ++49/40/7410-58243 F ++49/40/7410-57790 -- Pflichtangaben gem?? Gesetz ?ber elektronische Handelsregister und Genossenschaftsregister sowie das Unternehmensregister (EHUG): Universit?tsklinikum Hamburg-Eppendorf; K?rperschaft des ?ffentlichen Rechts; Gerichtsstand: Hamburg Vorstandsmitglieder: Prof. Dr. Guido Sauter (Vertreter des Vorsitzenden), Dr. Alexander Kirstein, Joachim Pr?l?, Prof. Dr. Dr. Uwe Koch-Gromus
Frank Harrell
2011-Oct-11 13:39 UTC
[R] How to test if two C statistics are significantly different?
Thanks for mentioning rcorrp.cens which is much more powerful than testing for differences in C. Likelihood ratio tests would be even more powerful. Ordinary differences in C index yields a test with power that is too low. Frank alanm (Alan Mitchell) wrote:> > ?Hmisc::rcorrp.cens > > -Alan > > > > -----Original Message----- > From: Eik Vettorazzi [mailto:E.Vettorazzi@] > Sent: Tue 10/11/2011 2:25 AM > To: Yujie Wang > Cc: r-help@ > Subject: Re: [R] How to test if two C statistics are significantly > different? > > Hi Yujie, > there is still a lot of work in progress, I think. As > http://faculty.washington.edu/heagerty/Software/SurvROC/RisksetROC/risksetROCdiscuss.pdf > > states: "[...] for inference and variance estimation, we now suggest > bootstrapping [...]". > Recently I catched a glimpse on roc.test from the pROC package, they > implemented, amongst others, a bootstrap algorithm - maybe this is a > start for your own work? > > Hth. > > Am 10.10.2011 21:35, schrieb Yujie Wang: >> Hey all, >> >> In order to test if a marker is a risk factor, I built two models (using >> cox >> proportional hazard model). One model included this marker, and the other >> is >> not. >> >> Then, I use R package risksetROC to test how much predictive value did >> the >> marker add to this model. I get two C statistics by analyzing the linear >> predictors of the two models into this package. >> >> The qustion is How to test if two C statistics are significantly >> different? >> >> Your help will be greatly appreciated! >> >> Yujie >> >> [[alternative HTML version deleted]] >> >> ______________________________________________ >> R-help@ 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. > > > -- > Eik Vettorazzi > Institut f?r Medizinische Biometrie und Epidemiologie > Universit?tsklinikum Hamburg-Eppendorf > > Martinistr. 52 > 20246 Hamburg > > T ++49/40/7410-58243 > F ++49/40/7410-57790 > > -- > Pflichtangaben gem?? Gesetz ?ber elektronische Handelsregister und > Genossenschaftsregister sowie das Unternehmensregister (EHUG): > > Universit?tsklinikum Hamburg-Eppendorf; K?rperschaft des ?ffentlichen > Rechts; Gerichtsstand: Hamburg > > Vorstandsmitglieder: Prof. Dr. Guido Sauter (Vertreter des Vorsitzenden), > Dr. Alexander Kirstein, Joachim Pr?l?, Prof. Dr. Dr. Uwe Koch-Gromus > > ______________________________________________ > R-help@ 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. >----- Frank Harrell Department of Biostatistics, Vanderbilt University -- View this message in context: http://r.789695.n4.nabble.com/How-to-test-if-two-C-statistics-are-significantly-different-tp3891857p3894430.html Sent from the R help mailing list archive at Nabble.com.