Hello all,
I'm working with a dataset of about 1,400 elementary schools, I'm looking to explore the trajectories of how percentage of white students changes over time within schools. My cases are schools and I have variables for: number of white students, number of total students, and percentage of white students over about two decades.
I'm using the traj user-written command found here: https://www.andrew.cmu.edu/user/bjones/
I've gotten the traj code to work ok with using a cnorm probability distribution, but as there is a large number of schools and years with 0% white, I'm also looking into using the zero-inflated poisson distribution. I want to use that with the count of white students and exposure as total number of students. However, when I run the code listed below (and I've tried several combinations for number of groups and orders), the model always either fails to converge (variance matrix is non-symmetric error) or if I don't get that error message, I still don't get standard errors. Any insights as to why this might be happening or suggestions to troubleshoot the issue? I tried using start values close to zero, but that did not help.
traj , var(WhiteStudents*) indep(yearn*) model(zip) expos(TotalStudents*) order(1 1 1)
Thank you for any and all responses!
I'm working with a dataset of about 1,400 elementary schools, I'm looking to explore the trajectories of how percentage of white students changes over time within schools. My cases are schools and I have variables for: number of white students, number of total students, and percentage of white students over about two decades.
I'm using the traj user-written command found here: https://www.andrew.cmu.edu/user/bjones/
I've gotten the traj code to work ok with using a cnorm probability distribution, but as there is a large number of schools and years with 0% white, I'm also looking into using the zero-inflated poisson distribution. I want to use that with the count of white students and exposure as total number of students. However, when I run the code listed below (and I've tried several combinations for number of groups and orders), the model always either fails to converge (variance matrix is non-symmetric error) or if I don't get that error message, I still don't get standard errors. Any insights as to why this might be happening or suggestions to troubleshoot the issue? I tried using start values close to zero, but that did not help.
traj , var(WhiteStudents*) indep(yearn*) model(zip) expos(TotalStudents*) order(1 1 1)
Thank you for any and all responses!
Attached Files
Comment