Graphing Logistic Regressions in Stata

#1
Hi all,

I'm testing for slope/threshold (intercept bias). My Y variable is dichotomous (past suicide attempt, yes/no) and my predictor variable, X1, is dichotomous (gender) and the other, X2, is continuous (scores on a test). Intercept bias is occurring with my X1 predictor variable, and I need to determine if it is over or under predicting y. In other words, are we over- or under-predicting if men have engaged in past suicide attempts.

Essentially, I am trying to graph two logit curves within one graph to examine intercept differences - one logit curve for men and one for women. The slopes should look the same, but there should be a difference in intercepts.

Anybody willing to offer some advice on doing this in STATA? :)
 

maartenbuis

TS Contributor
#2
Here is a (silly) example:

Code:
sysuse auto, clear
gen byte good = rep78 > 3 if !missing(rep78)
logit foreign i.good weight

margins, over(good) at(weight=(1800(200)4800))
marginsplot