How to manipulate coefficients in saved .ster model estimates

[Markos Valsamis]

Hi everyone,

I am aware that it is possible to save model estimates using

estimates save my_logistic_regression

to generate for example, a my_logistic_regression.ster file that can then be used in a separate dataset. However, I would like to edit the coefficients of the model by shrinking them (multiplying them by a particular coefficient that I have obtained from internal validation) and also editing the constant term. I am happy to calculate the new values of the coefficients manually (its a simple multiplication), and derive the new constant using some basic code, but I cannot figure out how "edit" the stored coefficients and constant in the .ster file. I would very much value your advice.

Best
Markos

[Andrew Musau]

estimates use my_logistic_regression
manipulation
ereturn post
estimates save my_logistic_regression, replace

See

help ereturn post

or

ssc describe erepost

[Markos Valsamis]

Hi Andrew Musau , I have tried the following

estimates use my_logistic_regression
matrix b = e(b)
matrix b[1,1]=0.08 //this changes just the first coefficient to 0.08
ereturn post b
estimates save my_logistic_model_2

but after the last line I get the error "last estimates not found". It does seem at least that the manipulation is working as if I run

matrix list e(b)

after

ereturn post b

, it gives the correct updated matrix....?

[Markos Valsamis]

I think I worked it out after seeing another Stata post from Joseph Coveney.

cap prog drop my_program
prog my_program, eclass
estimates use my_logistic_regression
matrix A = e(b)
matrix A[1,1] = 0.08
ereturn repost b = A
estimates save logistic_model_2
end

my_program

It then seems to work just like the original stored estimates.