Hi everyone,
I am having trouble in estimating marginal effects after a user-written ordinal probit. The ordinal probit runs well but I get "variable __000001 not found" after the code for marginal effects. I have adapted the following code of a multinomial logic to my ordinal probit http://www.stata.com/statalist/archi.../msg00113.html
Can anyone tell me what I am doing wrong ? I am using an updated version of Stata 11.2 in a Mac. I have replicated this problem in a public data:
I am having trouble in estimating marginal effects after a user-written ordinal probit. The ordinal probit runs well but I get "variable __000001 not found" after the code for marginal effects. I have adapted the following code of a multinomial logic to my ordinal probit http://www.stata.com/statalist/archi.../msg00113.html
Can anyone tell me what I am doing wrong ? I am using an updated version of Stata 11.2 in a Mac. I have replicated this problem in a public data:
Code:
cap program drop lfoprobit1
program lfoprobit1
args lnf xb a1 a2
local y "$ML_y1"
quietly replace `lnf' = ln( normal(`a1'-`xb')) if `y'==1
quietly replace `lnf' = ln( normal(`a2'-`xb') - normal(`a1'-`xb')) if `y'==2
quietly replace `lnf' = ln(1-normal(`a2'-`xb')) if `y'==3
end
cap program drop lfoprobit6
program lfoprobit6, eclass
ml model lf lfoprobit1 (rep78 = foreign length mpg, nocons) /cut1 /cut2
ml search, repeat(100)
ml maximize, difficult
eret local cmd="lfoprobit6"
eret local predict="lfoprobit6_p"
end
cap program drop lfoprobit6_p
program define lfoprobit6_p
syntax newvarlist [if] [in], [ p1 p2 p3]
tempvar xb2 xb3 xb4 touse
marksample touse
qui _predict double `xb2' if `touse', xb equation(#1)
qui _predict double `xb3' if `touse', xb equation(#2)
qui _predict double `xb4' if `touse', xb equation(#3)
if "`p1'"!=""{
gen double `varlist'=normprob(_b[/cut1]-`xb2')
}
else if "`p2'"!=""{
gen double `varlist'=normprob(`xb3' - _b[/cut1]) - normprob(`xb3' - _b[/cut2])
}
else if "`p3'"!=""{
gen double `varlist'=normprob(`xb4' - _b[/cut2])
}
label var `varlist'
end
clear
sysuse auto
replace rep=1 if rep==2 | rep==5| rep==.
recode rep (3=2)
recode rep (4=3)
lfoprobit6
I get :
rep78 Coef. Std. Err. z P>z [95% Conf. Interval]
eq1
foreign .2548369 .3573541 0.71 0.476 -.4455643 .9552381
length -.0035718 .0111222 -0.32 0.748 -.025371 .0182274
mpg -.0478152 .0398704 -1.20 0.230 -.1259598 .0303293
cut1
_cons -1.998984 2.835411 -0.71 0.481 -7.556287 3.55832
cut2
_cons -.9009799 2.832889 -0.32 0.750 -6.45334 4.65138
mfx, predict(p1) nonlinear
I get :
variable __000001 not found
mfx, predict(p2) nonlinear
mfx, predict(p3) nonlinear
I get the same than above.
margins, dydx(*) predict(p1)
I get :
variable __marg_pvar_1 not found