I'm attempting to test for a trend in age category and other ordinal variables on my dependent variable after running a logistic regression with multiply imputed data. I am able to test whether a collection of coefficients are zero with the
Code:
mi test
Here is a sample of my code, if that's helpful:
Code:
mi estimate, dots saving(depmicov, replace): meqrlogit suithink i.visit##i.controlgroup i.sex i.age_4cat ||phc_name: || pid: if cohort==1, or mi test 2.visit 3.visit mi test 1.controlgroup#2.visit 1.controlgroup#3.visit mi test 2.age_4cat 3.age_4cat 4.age_4cat
Much appreciated,
Luke
0 Response to Test for trend using mi postestimation
Post a Comment