[R] about systemfit

John Fox jfox at mcmaster.ca
Thu Apr 5 19:45:05 CEST 2007


Dear Martin,

Note that the F-test provided by linear.hypothesis() in the car package
appears to agree with the Wald test provided by both linear.hypothesis() and
waldtest.systemfit(), while ftest.systemfit() gives a much smaller p-value
and an F that's even larger than the Wald chisquare. This suggests to me
that the first place to look for a problem is in ftest.systemfit().

Regards,
 John

--------------------------------
John Fox
Department of Sociology
McMaster University
Hamilton, Ontario
Canada L8S 4M4
905-525-9140x23604
http://socserv.mcmaster.ca/jfox 
-------------------------------- 

> -----Original Message-----
> From: r-help-bounces at stat.math.ethz.ch 
> [mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of Martin Ivanov
> Sent: Thursday, April 05, 2007 1:02 PM
> To: r-help at stat.math.ethz.ch
> Subject: [R] about systemfit
> 
> Thank you very much for your responsiveness. Here are the 
> tests that show the same results, as they must:
> 
> linear.hypothesis(model=adfResulm,hypothesis.matrix=Rrestr10,t
> est="Chisq"):
>  Res.Df     RSS  Df Sum of Sq  Chisq Pr(>Chisq)
> 1    127  7.3782                                
> 2    137  7.6848 -10   -0.3066 5.2769      0.872
> 
> waldtest.systemfit(object=adfResu,R.restr=Rrestr10):
> 
> Wald-test for linear parameter restrictions in equation systems
> Wald-statistic: 5.277
> degrees of freedom: 10
> p-value: 0.8719 
> 
> So the Chisq test with linear.hypothesis from car and the 
> waldtest.systemfit give the same output. But the  F test with 
> linear.hypothesis from car and ftest.systemfit give 
> absolutely different results, as I demonstrated in my first 
> post. I have no idea why.
> 
> Regards,
> Martin
> 
> ______________________________________________
> R-help at stat.math.ethz.ch 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.
>



More information about the R-help mailing list