[R-sig-ME] gee p values

[Kevin E. Thorpe]

On 09/10/2010 05:22 PM, John Sorkin wrote:
> I am sending this Email to r-sig-mixed-models at r-project.org having previously sent it to r-help and it was suggested that I follow up on this listserver.
> windows Vista
> R 2.10.1
>
> Is it possible to get p values from gee? Summary(geemodel) does not appear to produce p values (see code at end of this Email message)
>
> I received the following response from Peng:
> There are two z-scores reported in the summary: Naive z and Robust z.
>
> pvalue=2*min(pnorm(z-score), 1-pnorm(z-score))   # two-sided test
>
> I replied to Peng as follows:
>> Peng,
>> If the answer were as simple as you suggest, I would expect that gee would
> automatically produce the p
>> values. Since gee does not produce the values, I fear that the computation may
> be more complex, or perhaps
>> computing p values from gee may be controversial. Do you know which, if either
> of my speculations is true?
>> Thank you,
>> John

FWIW, I usually use the geepack package for GEE.  It does provide p-values.

>
> Ben Bolker responded:
>    May be worth following up on r-sig-mixed-models .  My guess would be
> that if you're willing to treat your data set as 'large' (e.g. your
> guess is that the 'residual degrees of freedom', whatever that may
> mean, are>  40 ), then you could go ahead and use the naive translation
> from Z-score to p-value; otherwise it probably devolves to the
> usual 'effective residual degrees of freedom for complex multilevel/
> smoothing models' can of worms.
>
>
>
>
>
>
>
>
>
>
>
>
>
>> fit4<- gee(y~time, id=Subject, data=data.frame(data))
> Beginning Cgee S-function, @(#) geeformula.q 4.13 98/01/27
> running glm to get initial regression estimate
> (Intercept)        time
>    1.1215614   0.8504413
>> summary(fit4)
>
>   GEE:  GENERALIZED LINEAR MODELS FOR DEPENDENT DATA
>   gee S-function, version 4.13 modified 98/01/27 (1998)
>
> Model:
>   Link:                      Identity
>   Variance to Mean Relation: Gaussian
>   Correlation Structure:     Independent
>
> Call:
> gee(formula = y ~ time, id = Subject, data = data.frame(data))
>
> Summary of Residuals:
>            Min            1Q        Median            3Q           Max
> -2.5224390768 -1.4384989365 -0.0006304408  1.4385426203  2.5229173416
>
>
> Coefficients:
>               Estimate Naive S.E.  Naive z Robust S.E.  Robust z
> (Intercept) 1.1215614  0.8023886 1.397778  0.31918831  3.513792
> time        0.8504413  0.0993967 8.556031  0.03851821 22.078938
>
> Estimated Scale Parameter:  2.642821
> Number of Iterations:  1
>
> Working Correlation
>       [,1] [,2] [,3]
> [1,]    1    0    0
> [2,]    0    1    0
> [3,]    0    0    1
>
>
>
> John David Sorkin M.D., Ph.D.
> Chief, Biostatistics and Informatics
> University of Maryland School of Medicine Division of Gerontology
> Baltimore VA Medical Center
> 10 North Greene Street
> GRECC (BT/18/GR)
> Baltimore, MD 21201-1524
> (Phone) 410-605-7119
> (Fax) 410-605-7913 (Please call phone number above prior to faxing)
>
> Confidentiality Statement:
> This email message, including any attachments, is for ...{{dropped:15}}