Run Bivariate regressions for different groups quickly

[Sean O'Connor]

Hi folks,

I'm looking to see how employment specialisation effects national GVA. I've 35 different groups and rather than running 35 different regressions I was hoping to use a continuous method.

I have my data set up as panel data, with id 1=Aerospace Industry 2=Pharma Industry and so on.

Is anyone aware of a command in STATA which would allow me to run a bivariate regression for each of these panels?

Originally I tried,

forvalues I=1(1)35{
regress GVA Employment Spec if id=`i'
{

However, this code doesn't seem to work. Does anyone have any alternatives, please?

Kind regards,
Sean

[Svend Juul]

You should tell in which way the code "doesn't seem to work". But perhaps is it because you used an uppercase I in the first line and a lowercase `i' in the second line.

[Sean O'Connor]

Apologises, I actually used a lower case i. Running it this way provides the following error.

forvalues id=1(1)66{regress rgdpl ki if id =`i'}
program error:  code follows on the same line as open brace
r(198);

Running it this way provides no estimation.

. forvalues i=1(1)66{
 
regress rgdpl ki if id`i'
 {

Does anyone know what the issue might be, or can they offer an alternative solution?

[Svend Juul]

In your second attempt, you used { as the closing brace. It should be }.

[Nick Cox]

  
bysort id: regress rgdpl ki

Also check out statsby