Displaying 1 result from an estimated 1 matches for "multi2a".
Did you mean:
multi29
2011 Jun 13
0
How to formulate an (effect-modifying) interaction with matching variable in a conditional logistic regression?
...ova(mult2.cl,multi1.cl)
# then comes the interaction with smoke
# no good! Here's the problem
multi2.cl<-update(mult2.cl,.~.+age:smoke)
summary(multi2.cl)
anova(mult2.cl,multi2.cl)
# has to define my own variable?
dataset$ageNsmoke<-as.numeric(dataset$smoke=="yes")*dataset$age
multi2a.cl<-update(mult2.cl,.~.+ageNsmoke)
summary(multi2a.cl)
anova(mult2.cl,multi2a.cl)
# but suppose I had a model with race, then I'd have to
# define two variables
dataset$ageNblack<-as.numeric(dataset$race=="black")*dataset$age
dataset$ageNother<-as.numeric(dataset$race=="...