# [R] Advice on an appropriate condition for warning in a function

Mike Rennie mrennie at utm.utoronto.ca
Mon Jan 22 06:05:18 CET 2007

Hi, there

I have written a function that will allow me to calculate the variance
components for a model II (random effects) single factor ANOVA (perhaps this is
me re-inventing the wheel, but I handn't yet come across anything in R that
does this for me, but I'm sure people will let me know from this posting if
there are). The function takes as it's argument an object which holds the
results of an lm call. e.g.,

object1<_lm(y~x)

The function is as follows (some comments included):

comp.var.estimates<-function(object.lm)
{
anovmod<-anova(object.lm) #get the anova table for the lm
MStreat<-anovmod[1,3]; MSErr<-anovmod[2,3] #extract Mean Squares
dataframe<- as.data.frame(object.lm[13])
ni <- tapply(dataframe[,1], dataframe[,2], length)
nisq<-ni^2
no<-(1/(length(ni)-1))*(sum(ni)-(sum(nisq)/sum(ni)))
s2a<-((MSTreat-MSErr)/no)
stot<-s2a + MSErr
treatvar<-s2a/stot*100
errorvar<-MSErr/stot*100
list(treat.var.comp=s2a, err.var.comp=MSErr,
p.var.treat=treatvar, p.var.err=errorvar)
}
comp.var.estimates(object1)

I'd like to include a "warning" statement in the function that
returns 'function requires arguments that are objects of the form obj<-lm
(y~x)', but I need a case to evaluate the object against in order to throw the
warning.

I feel like my best opportunity is after the first line into the function,
where I ask for the ANOVA table. Since this is a 2 X 5 table, presumably I
should be able to evaluate it against the size of that table? Any thoughts on
how to check that? I welcome any suggestions.

Cheers,

Mike

--
Michael Rennie
Ph.D. Candidate, University of Toronto at Mississauga
3359 Mississauga Rd. N.
Mississauga, ON  L5L 1C6
Ph: 905-828-5452  Fax: 905-828-3792
www.utm.utoronto.ca/~w3rennie