Dear Emmanuel,
thank you for this. Unfortunately we get the following very cryptic error
message when running "predict(mod,df)": "Error in match.arg(type) : 'arg' must
be NULL or a character vector". Please see code and output below.
We can't work out what this error message is supposed to mean. We tried
redefining both variables as numbers (sex = 0/1, poll = 0/1) and redefining
them as characters using as.character, but to no avail.
We'd be grateful for suggestions.
Many thanks,
Nina Hobbhahn and Megan Welsford
> DF.MegsSort<-DF.Megs[phy17[[1]]$tip.label, ]
> mod<-compar.gee(FHsize~poll*sex, phy=phy17[[1]], data=DF.MegsSort,
> family=gaussian)
Beginning Cgee S-function, @(#) geeformula.q 4.13 98/01/27
running glm to get initial regression estimate
(Intercept) pollW sexM pollW:sexM
265.31404 -30.81771 -29.03028 -134.19117
Beginning Cgee S-function, @(#) geeformula.q 4.13 98/01/27
running glm to get initial regression estimate
(Intercept) pollW sexM pollW:sexM
265.31404 -30.81771 -29.03028 -134.19117
> mod
Call: compar.gee(formula = FHsize ~ poll * sex, data = DF.MegsSort,
family = gaussian, phy = phy17[[1]])
Number of observations: 34
Model:
Link: identity
Variance to Mean Relation: gaussian
QIC: 1920817
Summary of Residuals:
Min 1Q Median 3Q Max
-210.193547 -138.663156 -119.285868 4.672248 736.012732
Coefficients:
Estimate S.E. t Pr(T > |t|)
(Intercept) 244.51760 1.256536e+02 1.945966e+00 0.1285677
pollW 54.77936 5.384660e+01 1.017323e+00 0.3703143
sexM -29.03269 8.148837e-05 -3.562802e+05 0.0000000
pollW:sexM -134.18408 1.502553e-04 -8.930409e+05 0.0000000
Estimated Scale Parameter: 64026.63
"Phylogenetic" df (dfP): 7.734134
> df<-expand.grid(poll=levels(DF.MegsSort$poll), sex=levels(DF.MegsSort$sex))
> df
poll sex
1 I F
2 W F
3 I M
4 W M
> predict(mod,df)
Error in match.arg(type) : 'arg' must be NULL or a character vector
On 2013-12-03, at 1:05 PM, Emmanuel Paradis wrote:
> Dear Nina,
>
> You may try:
>
> df <- expand.grid(poll = levels(data$poll), sex = levels(data$sex))
> predict(mod, df)
>
> mod is the output from compar.gee. This will give you the predicted means for
> each combination of the factors (which are linear combinations of the
> coefficient estimates).
>
> HTH
>
> Emmanuel
_______________________________________________
R-sig-phylo mailing list - R-sig-phylo@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Searchable archive at http://www.mail-archive.com/r-sig-phylo@r-project.org/
