Posted by Bill on March 30, 2002 at 13:09:25:
I was wondering if somebody can tell me that can one use a binomial logistic regression instead of multinomial logistic regression in the scenario when you have three cases of a dependent variable and instead of using multinomial procedure can one creat three equations using code 1 for the first case and 0 for all other cases. Similarly one can use code 1 for second case and 0 for all others. Thus one can come up with three equations in this way. Is this procedure justified with a nominal dependent variable with three cases or more?