Greetings,
I'm Alessandro and I'm new at the community and at Stata, thus I apologize for any mistake I possibly make in this post.
I have an activity at college where I should try to replicate the regression from the Mincer-based model below. I gathered all the information needed, and now I'm trying to figure which is the best way to make the regression:
So, firstly I tryed to make a simple .regress using all the variables as factor, like this:
To do that I had to reset the matsize. And I don't think this is the right way to make it work.
So, I tryed doing the .xtreg, based on the panel options, with fixed effects (also random, to make sure) both didn't work: Not enough observations.
Then, I found the .areg, which I think may be the right way to do it. I tryed like this:
But this one probably isn't considering the interaction between "micr" and "year".
Based on this and my limited skills I have the following questions: i) There is a way to consider the interaction between the absorbed indicator and other variable? ii) Could you indicate if .areg is actually a good way to regress this model or there is another way to do it more efficiently?
I would aprecciate any help or comment.
Thank you and have a good day!!!
I'm Alessandro and I'm new at the community and at Stata, thus I apologize for any mistake I possibly make in this post.
I have an activity at college where I should try to replicate the regression from the Mincer-based model below. I gathered all the information needed, and now I'm trying to figure which is the best way to make the regression:
So, firstly I tryed to make a simple .regress using all the variables as factor, like this:
Code:
regress logie i.year##(i.age3##i.edattain i.micr)
Source | SS df MS Number of obs = 19368
-------------+------------------------------ F(1616, 17751) = 247.41
Model | 12091.0481 1616 7.4820842 Prob > F = 0.0000
Residual | 536.822207 17751 .030241801 R-squared = 0.9575
-------------+------------------------------ Adj R-squared = 0.9536
Total | 12627.8703 19367 .652030272 Root MSE = .1739
-----------------------------------------------------------------------------------------------
logie | Coef. Std. Err. t P>|t| [95% Conf. Interval]
------------------------------+----------------------------------------------------------------
year |
2000 | .7381118 .1104081 6.69 0.000 .5217013 .9545224
2010 | 1.071136 .0997068 10.74 0.000 .8757009 1.266571
|
age3 |
20 | .3379559 .0107893 31.32 0.000 .3168079 .3591039
30 | .491482 .0108114 45.46 0.000 .4702906 .5126734
40 | .379497 .0108421 35.00 0.000 .3582454 .4007486
|
edattain |
Primary completed | .3042252 .0109946 27.67 0.000 .2826747 .3257756
Secondary completed | .8034804 .0117411 68.43 0.000 .7804668 .8264941
--more--
So, I tryed doing the .xtreg, based on the panel options, with fixed effects (also random, to make sure) both didn't work: Not enough observations.
Then, I found the .areg, which I think may be the right way to do it. I tryed like this:
Code:
areg logie i.year##(i.age3##i.edattain), abs(micr)
Linear regression, absorbing indicators Number of obs = 19368
F( 47, 18797) = 6241.16
Prob > F = 0.0000
R-squared = 0.9503
Adj R-squared = 0.9488
Root MSE = 0.1828
-----------------------------------------------------------------------------------------------
logie | Coef. Std. Err. t P>|t| [95% Conf. Interval]
------------------------------+----------------------------------------------------------------
year |
2000 | .6260014 .0113266 55.27 0.000 .6038002 .6482026
2010 | .9427891 .0113664 82.95 0.000 .92051 .9650682
|
age3 |
20 |--Break--
r(1);
.
Based on this and my limited skills I have the following questions: i) There is a way to consider the interaction between the absorbed indicator and other variable? ii) Could you indicate if .areg is actually a good way to regress this model or there is another way to do it more efficiently?
I would aprecciate any help or comment.
Thank you and have a good day!!!
Comment