SAS, STATA find centered values for Gravity Model Estimation Code

[Lazaros Antonios Chatzilazarou]

-2 down vote favorite

I am having trouble with something. I have a dataset with years exporters importers and commodity codes. In order to make a panel in STATA and regress I created with the command "egen" a variable i which creates a group of importer exporter and commodity code.
Code: "egen i = group(exporter importer commodity_code)" I successfully created a panel set with i and year, generated the logarithmic forms of my variables and regressed getting great estimations. So, I have my current dataset (with the logs and the i etc) and I need to find the centered (mean) values of each group. I can do it in SAS, as I create a second dataset with the centered values and then merge it with the original to have the centered values inside along with the real ones. How can I do this in STATA?
Here is the code from SAS: (Datafors is the dataset's name I import in SAS)

data Sas1; retain exporter importer COMMODIT year; set Datafors; proc sort; by exporter importer COMMODIT; run; proc means data=Sas1 noprint; var IGDP IGDPCAP ICPI ; OUTPUT OUT=sas3 mean=IGDP IGDPCAP ICPI ; by exporter importer COMMODIT;id I; run; data sas3; set sas3; drop _TYPE_ _FREQ_; run; data sas4; set sas2 sas3; proc sort; by exporter importer commodit year; run; I want the means so when I will re-estimate I will get the predicted ones.
Thanks in advance

[Marcos Almeida]

Welcome to the Stata Forum / Statalist.

I don't work with SAS but, f I understood your query correctly, this information can be found in several sources such as threads from this Forum, the Stata Manual, etc.

Below, a couple of examples:

For "grand mean centering":

summarize myvar
gen myvar_cen1 = myvar – r(mean)

For "group mean centering":

egen mygroupmean = mean(myvar), by(mygroup)
gen myvar_cen2 = myvar - mygroupmean

Hopefully that helps.

[Nick Cox]

The otherwise puzzling start

-2 down vote favorite

can be explained by the fact that this is an unedited copy and paste of a question closed on Stack Overflow.

https://stackoverflow.com/questions/...stimation-code

[Lazaros Antonios Chatzilazarou]

Yes the message from stack overflow was that it was not relevant or something so I posted here where I believe it is. This question was posted from me on Stack Overflow and was not considered proper. However, nobody so far (except of the comment here) has ever solved it. I just want to find an answer for my question nothing more nothing less. Marcos thank you so much I will try everything I can.

[Nick Cox]

My point was just small, to explain what would otherwise have seemed mysterious.

I have never used SAS so don't feel obliged to try to translate. Your question is fine in principle for Statalist but in a discussion forum -- unlike on a helpline -- no one is obliged to answer anything. You might get more attention with formatted SAS code, as here. Sorry if I mess this up.

data Sas1;
retain exporter importer COMMODIT year;
set Datafors;
proc sort;
by exporter importer COMMODIT;
run;  
proc means data=Sas1  noprint;
var  IGDP IGDPCAP ICPI   ;
OUTPUT OUT=sas3 mean=IGDP IGDPCAP ICPI    ;
by exporter importer COMMODIT;id I;
run;  

data sas3; set sas3;
drop _TYPE_ _FREQ_;
run;  

data sas4;
set sas2 sas3;
proc sort;
by exporter importer commodit year;
run;