R help - Oct 2006 - VGAM-vglm

Can somebody explain to me why I get different results using multinom(package
nnet) and vglm(package VGAM)?
I think I understand the result I get with multinom, but what has vglm done?
How can I interpret the results?

(paarer: nominal variable, classes for different jobs, K26A:number of
children)

library(nnet)
erg<-multinom(paarer~K26A)
library(VGAM)
erg<-vglm(paarer~K26A,family="multinomial")

erg2
Call:
multinom(formula = paarer2 ~ K26A)

Coefficients:
  (Intercept)       K26A
3    1.036560 0.35340860
4   -0.801480 0.27423903
5    1.191205 0.05197307

Residual Deviance: 3358.351 
AIC: 3370.351 
> erg
Call:
vglm(formula = paarer ~ K26A, family = "multinomial")

Coefficients:
(Intercept):1 (Intercept):2 (Intercept):3        K26A:1        K26A:2 
   -1.1910394    -0.1545859    -1.9922876    -0.0521119     0.3013985 
       K26A:3 
    0.2221076 

Degrees of Freedom: 4524 Total; 4518 Residual
Residual Deviance: 3358.351 
Log-likelihood: -1679.175 


-- 
View this message in context:
http://www.nabble.com/VGAM-vglm-tf2542392.html#a7083843
Sent from the R help mailing list archive at Nabble.com.

> Can somebody explain to me why I get different results using 
multinom(package
 > nnet) and vglm(package VGAM)?

The multinom function in the nnet package
uses the first level of the factor as baseline, whereas the last
level of the factor is used with multinomial().
Consequently the estimated regression coefficients differ.
They can be reconciled very easily with some basic maths.

cheers
Thomas