t1=p1*c1 p1 is the mean of a variable, c1 is the coefficient of a regression for this variable
t2=p2*c2 the same as t1 with a different variable
tot=t1+t2
I calculated the value of tot by the linear combinations of the numbers. My coauthor asked me to get the variance for the value of tot.
I created a bootstrap program to calculate the confidence intervals of tot, but there are many error messages. I am wondering whether I am at the right direction. Here are the codes
Code:
capture program drop bootc program define bootc,rclass drop quant p1 p2 c1 c2 t1 t2 tot gen quant=40000 mean cond1 gen p1=r(mean) mean cond2 gen p2=r(mean) glm cost i.cond $char, family(gamma) link(log) le(95) gen c1=_b[1.cond] gen c2=_b[2.cond] gen t1=quant*p1*c1 gen t2=quant*p2*c2 gen tot=t1+t2 end simulate tot=r(tot), reps(4) seed(12345): bootc estat bootstrap, all
0 Response to bootstrap to get the confidence intervals for a linear cobmination of regression coefficients
Post a Comment