regression on a splitted sample

#1
Hello to all.

I'm writing my final dissertation and, after many analysis (OLS, GMM), I need help to do my last one analysis.

I need to split my sample (2700 observation) in 5 sub-groups. I want to make these latter sub-groups from values of 4-5 variables (like cash flow, leverage, etc) and then do a regression for every group. How can I do this with Stata? Does exist another method to do this (also outside Stata)?

Thanks a lot to all
 

Dason

Ambassador to the humans
#2
What exactly is the question? How are you trying to split them up into groups? I splitting them up in STATA the problem or are you having issues because you don't know what the groups should be?
 
#3
What exactly is the question? How are you trying to split them up into groups? I splitting them up in STATA the problem or are you having issues because you don't know what the groups should be?

Hello Dason, thank you for reply and excuse me if my English is not so perfect. How I wrote before, I need to get 5 subgroups on the basis of 5-6 variables. I need something like this:

Group A contains observations with firms who have cash flows>0.8, leverage <0.2, etc

Group B contains observations with firms who have cash flows>0.6, leverage <0.4, etc

....

Then make a regression on all of these subgroups.
 
#4
How about going

preserve

keep if dev>0.5...whatever the criteria for this particular subgroup

then run the regression

store the estimates

then

restore (to get back your full dataset)

then on to the next subgroup

Stephen
 
#5
How about going

preserve

keep if dev>0.5...whatever the criteria for this particular subgroup

then run the regression

store the estimates

then

restore (to get back your full dataset)

then on to the next subgroup

Stephen
Thanks really a lot Stephen.

That solved my problem.

Thanks again and have a nice day