I was trying to generate samples of 50 observations in a Monte Carlo loop and then just use the first 25 to estimate by OLS. I wrote these instructions: nulldata 50 genr x=100*uniform() loop 1000 genr eps=3*normal() genr y=5+2*x+eps smpl (obs<26) --restrict ols y const x genr bhat=coeff(x) store test.gdt bhat endloop But, when I run the program and open the test.gdt data file, I only get 1 observation of the variable bhat, instead of the 1000 I was supposed to get. I have tried a thousand different things and I do not get right. Any hint?