LPM graphing

[Enkhsaikhan Sukhee]

Hello all,

I need help graphing LPM. The graph must show sth like the below but for men and women separately. I syntaxed sth like this but had no success at all...

regress public i.age_group female i.socclass, base

predict yhat

twoway scatter yhat public age_group if female==1 , connect(1 .) sort (i O) ylabel(0 1)

scatter yhat public age_group if female==0 , connect(l .) symbol(i O) sort ylabel(0 1)

Thank you
Saikhnaa

[Nick Cox]

On the face of it your graph is clear evidence for why many researchers avoid the LPM as you have many fitted values above 1 or below 0.

But your graph is still puzzling, as a fit in terms of

regress public i.age_group female i.socclass

would produce fitted values with more fine structure than your graph shows.