Dear all,
I quick question relating to multiply imputed survey data, quantile regression and how to take into account sample weights.
STATA 14 does not provide a possibility to deal with multiple imputed data and sample weights simultaneously in the case of estimating quantile regression. I would like to include the final sampling weights (hw0010) as additional covariate in order to reduce any potential selection bias normally corrected for by weighted regressions.
My final dataset is concluded with the following command: mi svyset [pw=hw0010], bsrweight(WR0001-WR1000) /// <--1000 replicate weights
vce(bootstrap)
save hw_mi_wide, replace
I was wondering if i can use something like: mi estimate: qreg x1 x2 [w=hw0010], quantile(.1 .25 .5 .9) reps(500) but it doesnt allow weights.
Alternatively, I tried the following but not sure if they are the correct ones:
svyset [pw=hw0010], vce(bootstrap) bsrweight(WR*) svy: qreg x1 x1 or bs4rw, rw(WR*): qreg x1 x2 [w=finalwgt], q(.1 .25 .5 .9)
Thanks in advance, Ilias
I quick question relating to multiply imputed survey data, quantile regression and how to take into account sample weights.
STATA 14 does not provide a possibility to deal with multiple imputed data and sample weights simultaneously in the case of estimating quantile regression. I would like to include the final sampling weights (hw0010) as additional covariate in order to reduce any potential selection bias normally corrected for by weighted regressions.
My final dataset is concluded with the following command: mi svyset [pw=hw0010], bsrweight(WR0001-WR1000) /// <--1000 replicate weights
vce(bootstrap)
save hw_mi_wide, replace
I was wondering if i can use something like: mi estimate: qreg x1 x2 [w=hw0010], quantile(.1 .25 .5 .9) reps(500) but it doesnt allow weights.
Alternatively, I tried the following but not sure if they are the correct ones:
svyset [pw=hw0010], vce(bootstrap) bsrweight(WR*) svy: qreg x1 x1 or bs4rw, rw(WR*): qreg x1 x2 [w=finalwgt], q(.1 .25 .5 .9)
Thanks in advance, Ilias