2

I have done an ordinal regression using lrm from the rms package.

load(url("http://www.produnis.de/R/NEXT-Q0-OR.RData"))
require(rms)
mydata <- NEXT.or
fit <- lrm(QuitJob~QD,data=mydata);fit

I would like to predict the rank in "QuitJob" (1=won't quit ... 5=will quit very soon) by giving a value for "QD" (how much work to do). For example, if "QD" is 78, what rank of "QuitJob" is likely?

If I type in 

predict.lrm(fit, data.frame(QD=78))

I get

        1 
0.3191933

Do I understand it right, that the probability of being in rank 1 is 31,92% ?! If so, how can I get the probabilities for ranks 2-5 ?

Jeromy Anglim
  • 42,044
  • 23
  • 146
  • 250
Produnis
  • 421
  • 1
  • 3
  • 12
  • 4
    `predict(fit, data.frame(QD=78), type="fitted.ind")` gives predicted class probabilities, what you got is the value of the linear predictor, see [this answer](http://stats.stackexchange.com/a/41025/1909). – caracal Jul 24 '13 at 08:39
  • thx, caracal... If you post your commend as an answer, I would mark it as "accepted" – Produnis Jul 24 '13 at 08:45

0 Answers0