Re: [Rd] Standardized Pearson residuals (and score tests)

From: Gordon K Smyth <>
Date: Thu, 17 Mar 2011 09:29:30 +1100 (AUS Eastern Daylight Time)

Hi Peter and others,

If it helps, I wrote a small function glm.scoretest() for the statmod package on CRAN to compute score tests from glm fits. The score test for adding a covariate, or any set of covariates, can be extracted very neatly from the standard glm output, although you probably already know that.


Professor Gordon K Smyth,
NHMRC Senior Research Fellow,
Bioinformatics Division,
Walter and Eliza Hall Institute of Medical Research, 1G Royal Parade, Parkville, Vic 3052, Australia.

> Date: Tue, 15 Mar 2011 12:17:46 +0100
> From: peter dalgaard <>
> To: Brett Presnell <>
> Cc:
> Subject: Re: [Rd] Standardized Pearson residuals
> On Mar 15, 2011, at 04:40 , Brett Presnell wrote:
>>>> Background: I'm currently teaching an undergrad/grad-service course
>>>> from Agresti's "Introduction to Categorical Data Analysis (2nd edn)"
>>>> and deviance residuals are not used in the text. For now I'll just
>>>> provide the students with a simple function to use, but I prefer to
>>>> use R's native capabilities whenever possible.
>>> Incidentally, chisq.test will have a stdres component in 2.13.0 for
>>> much the same reason.
>> Thank you. That's one more thing I won't have to provide code for
>> anymore. Coincidentally, Agresti mentioned this to me a week or two
>> ago as something that he felt was missing, so that's at least two
>> people who will be happy to see this added.
> And of course, I was teaching a course based on Agresti & Franklin:
> "Statistics, The Art and Science of Learning from Data", when I realized
> that R was missing standardized residuals.
>> It would also be nice for teaching purposes if glm or summary.glm had a
>> "pearsonchisq" component and a corresponding extractor function, but I
>> can imagine that there might be arguments against it that haven't
>> occured to me. Plus, I doubt that anyone wants to touch glm unless
>> it's to repair a bug. If I'm wrong about all that though, ...
> Hmm, how would that work? If there was one, I'd worry that people would
> start subtracting them which is usually not the right thing to do. I do
> miss having a test on the residual deviance occasionally (even though it
> is only sometimes meaningful), having to fit a saturated model
> explicitly can be a bit silly. E.g. in this case (homogeneity of birth
> rates):
>> anova(glm(births~month,poisson,data=bb), test="Chisq")
> ...
> Df Deviance Resid. Df Resid. Dev P(>|Chi|)
> NULL 11 225.98
> month 11 225.98 0 0.00 < 2.2e-16 ***
>> anova(glm(births~1,poisson,data=bb), test="Chisq")
> ...
> Df Deviance Resid. Df Resid. Dev P(>|Chi|)
> NULL 11 225.98
> Notice that the latter version gives me the correct deviance but no
> p-value.
> A better support for generic score tests could be desirable too. I
> suspect that this would actually be the Pearson Chi-square in the
> interesting cases.
> --
> Peter Dalgaard
> Center for Statistics, Copenhagen Business School
> Solbjerg Plads 3, 2000 Frederiksberg, Denmark
> Phone: (+45)38153501
> Email: Priv:

The information in this email is confidential and intend...{{dropped:4}} mailing list Received on Wed 16 Mar 2011 - 22:32:38 GMT

Archive maintained by Robert King, hosted by the discipline of statistics at the University of Newcastle, Australia.
Archive generated by hypermail 2.2.0, at Thu 17 Mar 2011 - 14:40:35 GMT.

Mailing list information is available at Please read the posting guide before posting to the list.

list of date sections of archive