conditonal regression

[Raphael George]

How do I run a regression conditional on dropout? dropout is an indicator equal to 1 or is just missing. distid is the unique district identifier.

	
	

	
	
* Example generated by -dataex-. To install: ssc install dataex
clear
input float distid byte y float x byte child_sch_drop
516 .  6.018593 .
516 .  6.018593 .
277 .  6.016157 .
 44 .    3.7612 .
516 .  6.018593 .
516 .  6.018593 .
279 .   5.31812 .
307 .  5.765191 .
 91 .  4.060443 .
516 .  6.018593 .
444 .  5.252274 .
358 .  5.631212 .
336 .  6.196444 .
126 .  5.916202 .
395 .  5.934894 .
 24 .   4.65396 .
 20 . 4.3694477 .
559 .  5.398163 .
516 .  6.018593 .
482 .  6.475433 .
515 .  5.497168 .
516 .  6.018593 .
 67 .  5.117994 .
235 .  3.135494 .
562 .  5.129899 .
310 .  6.727432 .
501 .  6.194406 .
 39 . 4.5108595 .
515 .  5.497168 .
113 .  5.720312 .
510 .  6.011267 .
 20 . 4.3694477 .
515 .  5.497168 .
515 .  5.497168 .
515 .  5.497168 .
518 .  7.557473 .
515 .  5.497168 .
515 .  5.497168 .
528 .  6.448889 .
 20 . 4.3694477 .
517 .  6.400258 .
515 .  5.497168 .
515 .  5.497168 .
515 .  5.497168 .
515 .  5.497168 .
532 .  6.563856 .
 22 . 4.3307333 .
 22 . 4.3307333 .
515 .  5.497168 .
551 . 3.9512436 .
 22 . 4.3307333 .
377 .  6.238325 .
481 .  6.851185 .
 22 . 4.3307333 .
498 .   6.23637 .
479 .  6.523562 .
515 .  5.497168 .
 86 .  5.303305 .
189 .  6.109248 .
 22 . 4.3307333 .
515 .  5.497168 .
515 .  5.497168 .
415 .  5.181784 .
558 .  6.186209 .
415 .  5.181784 .
514 .  5.351858 .
 66 .  4.744932 .
514 .  5.351858 .
514 .  5.351858 .
320 .  6.985642 .
502 .  6.793466 .
514 .  5.351858 .
179 .  5.187386 .
 22 . 4.3307333 .
531 .   7.04316 .
514 .  5.351858 .
 28 .  6.144186 .
514 .  5.351858 .
531 .   7.04316 .
391 .  6.364751 .
155 .    5.7301 .
514 .  5.351858 .
 10 .  4.787492 .
513 .  5.869297 .
426 . 3.6888795 .
513 .  5.869297 .
513 .  5.869297 .
348 .     5.826 .
567 .  5.267858 .
 22 . 4.3307333 .
367 . 4.5849676 .
 22 . 4.3307333 .
 20 . 4.3694477 .
558 .  6.186209 .
521 .  5.638355 .
101 .  6.688354 .
513 .  5.869297 .
386 .  5.648974 .
514 .  5.351858 .
551 . 3.9512436 .
end
------------------ copy up to and including the previous line ------------------

Listed 100 out of 5716961 observations
Use the count() option to list more
------------------ copy

[Clyde Schechter]

It is not a good idea in Stata to code yes/no variables with 1/missing value. It is much better to code them as 1 = yes, 0 = no. So:

mvencode child_sch_drop, mv(0)

regress y x if child_sch_drop

will do it.

Note: In your example data, none of the observations have child_sch_drop == 1, so as applied to the example data, this code just produces a "no observations" error at the -regress- command. Presumably this is not the case in your real data.