R devel - Jan 2002 - na.action=na.omit doesn't omit in glm() (PR#1282)

Full_Name: Ralph Mac Nally
Version: 1.40
OS: MacOS9x/Windows98
Submission from: (NULL) (130.194.13.180)


I'm using a simple glm in which the dependent variable may have missing data
(coded 'NA') as usual. Both the default na.action (na.omit) and explicit
'na.action=na.omit' do not seem to ignore the missing data as I expect,
generating an error to that effect (NA passed to routine). I encounter the same
problem in classic Mac and Windows98 versions, but S works alright.

What is the source/solution to this problem?

Thanks Ralph

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

On Fri, 25 Jan 2002 Ralph.MacNally@sci.monash.edu.au wrote:
> Full_Name: Ralph Mac Nally
> Version: 1.40
> OS: MacOS9x/Windows98
> Submission from: (NULL) (130.194.13.180)
>
>
> I'm using a simple glm in which the dependent variable may have missing
data
> (coded 'NA') as usual. Both the default na.action (na.omit) and
explicit
> 'na.action=na.omit' do not seem to ignore the missing data as I
expect,
> generating an error to that effect (NA passed to routine). I encounter the
same
> problem in classic Mac and Windows98 versions, but S works alright.
>
> What is the source/solution to this problem?
Could we have a reproducible example, please?

Based on example(glm)

counts <- c(18, 17, 15, 20, 10, 20, 25, 13, 12)
outcome <- gl(3, 1, 9)
treatment <- gl(3, 3)
d.AD <- data.frame(treatment, outcome, counts)

glm(counts ~ outcome + treatment, family = poisson, data=d.AD)
Degrees of Freedom: 8 Total (i.e. Null);  4 Residual
Null Deviance:      10.58
Residual Deviance: 5.129        AIC: 56.76

d.AD[1, 1] <- NA
glm(counts ~ outcome + treatment, family = poisson, data=d.AD)
Degrees of Freedom: 7 Total (i.e. Null);  3 Residual
Null Deviance:      10.46
Residual Deviance: 4.011        AIC: 50.91

d.AD[9,3] <- NA
glm(counts ~ outcome + treatment, family = poisson, data=d.AD)
Degrees of Freedom: 6 Total (i.e. Null);  2 Residual
Null Deviance:      8.934
Residual Deviance: 2.725        AIC: 45.28

appears to work exactly as advertised.

Or try not using a data frame:

counts[9] <- NA
glm(counts ~ outcome + treatment, family = poisson)
Degrees of Freedom: 7 Total (i.e. Null);  3 Residual
Null Deviance:      8.971
Residual Deviance: 3.204        AIC: 50.5


-- 
Brian D. Ripley,                  ripley@stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._