[R] understanding the output

[Peter Dalgaard BSA]

"Ronaldo Reis Jr." <chrysopa at insecta.ufv.br> writes:


>             Estimate Std. Error t value Pr(>|t|)    
> (Intercept)   4.8660     0.9808   4.961 0.000101 ***
> gen2g         3.8504     1.3870   2.776 0.012459 *  
> ---
...
> > tapply(peso,gen,mean)
>     1g     2g 
> 4.8660 8.7164 
> 
> The values is the same in estimate.
> But the SE, when I try using tapply the SE is different os the SE of 
> estimates. look
> 
> > tapply(peso,gen,sd)/sqrt(tapply(peso,gen,length))
>        1g        2g 
> 0.8665567 1.0829789 
> 
> Why this difference? What SE I can use with mean???

(I assume you are only looking at the intercept, i.e. the mean of
group "1g"). glm() assumes a constant SD and uses the pooled SD.

Try for illustration:

sds <- tapply(peso,gen,sd)
lens <- tapply(peso,gen,len)
dfs <- lens - 1
sd.pool <- sqrt(sum(sds^2*dfs)/sum(dfs))
sd.pool/sqrt(lens)

It's a matter of taste which SE you'd put on a plot. Some people feel
that you "shouldn't mix in the other groups" in a mainly descriptive
plot, but as soon as any modelling is involved (e.g., age-adjusted
means) you pretty much have to extract SE's from the model.

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help 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-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._