Right way to extract Confidence Intervals of coefficients of interaction term

[Lars Pete]

Dear all,

I was wondering about what would the right way to form a list or matrix or a variable (either of them would work for me) of confidence intervals of the coefficients of the interaction term treat#ib2005.year from the regression:

regress index treat i.id ib2005.year treat#ib2005.year, vce(cluster id) (2005 is the base year)

I am aware of the command - _b[foreign] - invttail(e(df_r),0.025)*_se[foreign] to generate the 95% lower bound and using + to generate upper bound. However, how to specify the interaction term in place of "foreign"?

I want to form a list or matrix of confidence intervals for all 10 coefficients (corresponding to 10 years) of treat#ib2005.year.

What is the right way to do that?

Thanks in advance!

[William Lisowski]

After running your regress command to fit your model, run the command

regress coeflegend

to learn the names of the coefficients, including those of the interaction terms. Then notice that in the output of

help regress postestimation

the lincom command seems to do the calculations you want, so perhaps the output of

help lincom

will start you in a useful direction to producing your desired output.

[Wouter Wakker]

In addition, have a look at r(table) after your estimation command.

mat list r(table)

This already contains confidence intervals for each parameter so you would only need to extract them.

[Lars Pete]

Thank you very much William. I used to think that lincom was supposed to be used only after svy setting the data. I had no idea it could be used without it too. Thanks!

[Lars Pete]

Thank you so much Wouter. I was aware of r(table) but I didn't know what to do with that list . So I did matrix c = r(table) and then extracted what I wanted from here.