Reply With Quote
There are things like ordered logistic regression or multinomial logistic regression (assuming no ordering between groups) which you might want to investigate
Hi,
I want to run a logistic regression analysis, but my dependent variable is not bivariate but three-variable. What I mean is I have two different patient groups and one control group, coded with 2,1 and 0 as the dependent group.
I can recode the 2 into 1 and consider the two patient groups as one single patient group and run the test, however I would like to learn whether it is possible to run the test with 3 different dependent groups.
There are things like ordered logistic regression or multinomial logistic regression (assuming no ordering between groups) which you might want to investigate
However - why are you treating which group a patient was in as the response?
"His programming is malfunctioning. It begins! Get your weapons, he's going to become a killbot!!!" - bryangoodrich
@duskstar, thank you for your answer, I will tried the multinomial but it didnot work
@Dason, because I want to see which of my parameters are significantly associated with the disease I'm working with. In recent years I have worked with another disease and I had one patient and one control group (only 1s and zeros). And I was thaught by a statician to use the patient-control column as the dependent variable.
however in this project, the disease has two similar forms, therefore I have two different patient groups which I coded with 2 and 1. additionally I alsohave a control group coded with zero. At the end of regression I want to see which parameters are associated with which disease. should I do it separately instead?
Posting Permissions
|
|
