[R] How to compute the score-statistics from a logistic model using lrm {Design}

Christian Ametz Christian.Ametz at aon.at
Fri May 22 15:11:13 CEST 2009


is there are way to simply compute the score-statistics for a logistic 
model generated with lrm?

For example, I want to compare the wald-statistics for a given model 
against the score-statistics in order to find the relevant predictors.

 > attach(iris)
 > model = lrm (species~.,iris)
 > anova (model)
                Wald Statistics          Response: Species

 Factor       Chi-Square d.f. P    
 Sepal.Length 1.06       1    0.3032
 Sepal.Width  2.22       1    0.1358
 Petal.Length 3.96       1    0.0465
 Petal.Width  3.52       1    0.0605
 TOTAL        5.56       4    0.2346

So what I now basically want is an output of the score-statistics to 
verify if it leads to the same predictors than the wald-statistics 
(anova) and therefore estimate if the score-statistics would produce a 
better model.

I hope I stated everything clearly.
Thanks for your help!


More information about the R-help mailing list