Hi all,
I am trying to figure out how to do a DDD with the new xtdidregress Stata 17 package for panel data (the tedidregress documentation only shows DDD with didregress).
I have a panel dataset with 191 school districts (panel ID = leaid) and years 2011-2022. I am looking at changes in enrollment for treated districts that switched to fully distanced instruction for the 2020-2021 school year (year 2021 in the data). The treatment indicator (distanced) = 1 for years 2021 and 2022 for those districts that implemented distanced instruction. Tech is the level of technology provision, on a scale from 1-4 (1-3 are considered low tech provision districts).
My DD model is as follows:
xtset leaid
xtdidregress (enroll_grades_pk_12) (distanced), group(leaid) time(year)
I would like to run a DDD model to look at whether the impact on enrollment was greater for distanced districts that had low provision of technology (d_lowtech = 1). These can be thought of as "high-treatment" districts. I could use d_lowtech as my treatment indicator, but this model would move all of the low-treatment districts into the control group and would just be another version of the DD.
Any suggestions?
amy
I am trying to figure out how to do a DDD with the new xtdidregress Stata 17 package for panel data (the tedidregress documentation only shows DDD with didregress).
I have a panel dataset with 191 school districts (panel ID = leaid) and years 2011-2022. I am looking at changes in enrollment for treated districts that switched to fully distanced instruction for the 2020-2021 school year (year 2021 in the data). The treatment indicator (distanced) = 1 for years 2021 and 2022 for those districts that implemented distanced instruction. Tech is the level of technology provision, on a scale from 1-4 (1-3 are considered low tech provision districts).
My DD model is as follows:
xtset leaid
xtdidregress (enroll_grades_pk_12) (distanced), group(leaid) time(year)
I would like to run a DDD model to look at whether the impact on enrollment was greater for distanced districts that had low provision of technology (d_lowtech = 1). These can be thought of as "high-treatment" districts. I could use d_lowtech as my treatment indicator, but this model would move all of the low-treatment districts into the control group and would just be another version of the DD.
Any suggestions?
amy
Attached Files
