Hello, I am trying to manually implement the survey design to compute standard errors using the arhomme command since svyset is not compatible with it. I am using bsample with the options strata(raestrat) cluster(raehsamp) weight(new_weight), but when I include the weight() option, I receive the error: "estimates post: matrix has missing values." Any advice on how to properly implement the weights would be greatly appreciated. My code is listed below.
Code:
****bootstrap survey design SE*****
*Step 1, save observed coefficients****
preserve
quietly xi: arhomme log_avrg_cost i.inc_d i.endentulism i.race i.age_cat i.male i.education i.veteran i.mothered i.wealth i.smoke_now chronicdisease[pw=new_weight], select(r11dentst = dentalinsurance_w1 endentulism inc_d race age_cat male education veteran mothered wealth smoke_now chronicdisease) quantiles(0.10, 0.25, 0.50, 0.75, 0.90) taupoints(29) rhopoints(35) meshsize(0.5) frank nostderrors centergrid(-0.20)
matrix beta = e(b)
global N "`e(N)'"
global Ns "`e(sN)'"
restore
** step 2 generate program for bootstrap****
capture program drop arhomme_bootstrap
program arhomme_bootstrap, eclass
preserve
bsample, strata(raestrat) cluster(raehsamp) weight(new_weight)
quietly xi: arhomme log_avrg_cost i.inc_d i.endentulism i.race i.age_cat i.male i.education i.veteran i.mothered i.wealth i.smoke_now chronicdisease[pw=new_weight], select(r11dentst = dentalinsurance_w1 endentulism inc_d race age_cat male education veteran mothered wealth smoke_now chronicdisease) quantiles(0.10, 0.25, 0.50, 0.75, 0.90) taupoints(29) rhopoints(35) meshsize(0.5) frank nostderrors centergrid(-0.20)
matrix beta_boot = e(b)
forvalue i = 1/119 {
ereturn scalar beta_boot_`i' =beta_boot[1, `i']
}
restore
end
estimates post: matrix has missing values
r(504);
