[R] ordered logistic regression

bunny , lautloscrew.com bunny at lautloscrew.com
Thu Jun 7 16:47:15 CEST 2007


Hi there,


i tried to run an ordered logistic regression with polr. so far it  
worked after i turned my data into factors.
but here´s my problem:
my output is like this:

Call:
polr(formula = factor(fulltest[, 1]) ~ factor(fulltest[, 2]) +
     factor(fulltest[, 10]), method = "logistic")

Coefficients:
factor(fulltest[, 2])0  factor(fulltest[, 2])1  factor(fulltest[, 2]) 
2  factor(fulltest[, 2])3 factor(fulltest[, 10])0 factor(fulltest[,  
10])1 factor(fulltest[, 10])2
               1.0850358               0.8269005                
0.8850035               1.0263442                
1.3724189               1.8258853               2.2263393
factor(fulltest[, 10])3
               2.5234381

Intercepts:
        -1|0         0|1         1|2         2|3
-2.42111430 -2.13077351  0.07966516  2.85951997

Residual Deviance: 1219.493
AIC: 1243.493


as far is i understand, there´s a dummy var introduced for every  
possible outcome ( 0 , 1, 2 ,3) . this is nice because it contains a  
lot of information, but far more than i need.
and i have to many variables to research to use dummies for every  
single one of them. Can i get one coefficient per variable ?

Is there another package for logistic regression ? did is use the  
factor thing the wrong way ?

thx in advance

mattthias



More information about the R-help mailing list