Good Morning -
I am trying to model high school coursetaking in specific career and technical education CTE) school subjects. I have student-level coursetaking data for 4 years of high school (i.e., 9th, 10th, 11th, and 12th grade) for each student in the dataset. So, four observations per student in the panel.
My primary predictor is the fraction of individuals employed in the CTE subject per 1000 jobs in an observed participant's area. I have a number of additional model covariates as well. Screen shot of the data:
So far, I have been modeling CTE subject coursetaking (measured as the count of courses) using xtpoisson. However, since the majority of participants in the study did not take any courses in a given CTE subject area, there are many observed zeros. Consequently, I am considering a hurdle model, using Stata's churdle command. This theoretically checks out in my mind, with CTE coursetaking in high school being the function of two decisions: should I take CTE, and how many should I take?
My question is the following:
Is there a way to use Stata's churdle command in order to account for the nested nature of these data? I have (successfully) run the following command:
churdle linear ACcreds AC_jobs1000 AC_wtmean2 $xlist , select($xlist) ll(0) vce(cluster STU_ID)
But I am not sure that this model properly takes autocorrelation among the observations into account like, for instance, xtpoisson might.
Thank you in advance
I am trying to model high school coursetaking in specific career and technical education CTE) school subjects. I have student-level coursetaking data for 4 years of high school (i.e., 9th, 10th, 11th, and 12th grade) for each student in the dataset. So, four observations per student in the panel.
My primary predictor is the fraction of individuals employed in the CTE subject per 1000 jobs in an observed participant's area. I have a number of additional model covariates as well. Screen shot of the data:
So far, I have been modeling CTE subject coursetaking (measured as the count of courses) using xtpoisson. However, since the majority of participants in the study did not take any courses in a given CTE subject area, there are many observed zeros. Consequently, I am considering a hurdle model, using Stata's churdle command. This theoretically checks out in my mind, with CTE coursetaking in high school being the function of two decisions: should I take CTE, and how many should I take?
My question is the following:
Is there a way to use Stata's churdle command in order to account for the nested nature of these data? I have (successfully) run the following command:
churdle linear ACcreds AC_jobs1000 AC_wtmean2 $xlist , select($xlist) ll(0) vce(cluster STU_ID)
But I am not sure that this model properly takes autocorrelation among the observations into account like, for instance, xtpoisson might.
Thank you in advance
Comment