Save regression coefficients and number of obs in new variable

[Milan Quentel]

Dear Statalist users,

I would like to run a regression by different groups (roughly 1000), saving regression coefficients and number of observations in a new variable, such that for each individual observation in the group I have a variable showing the respective regression coefficient that the regression for that group yielded.

I have had a look at combining by, sort and regress:

sort group
by group: regress y x

and, alternatively, statsby

statsby _b, by(group): regress y x

but was not able to find out how to store the coefficient values in a new variable.

How can I do it? And, what is the most convenient way of doing this? Do you have any other suggestions?

Many thanks,
Milan

[Nick Cox]

Since this is precisely what statsby does for you, it is not clear what you are unclear about. But consider this example

sysuse auto, clear
statsby n=e(N) _b , by(rep78) : regress mpg weight

* optional!
merge 1:m rep78 using auto
egen tag = tag(rep78) 

The next to last line shows that you can have the results as new variables in your original dataset should you wish. But watch out for multiple counting; the tag variable you can create with egen can be useful there. See the help for egen.

[Milan Quentel]

Thank you very much. Apparently, I had misunderstood how the command works. I should have read the description more carefully. Now that I have gone through your example, it seems very straightforward. Apologies for the question and thank you for the excellent clarification.