I followed
https://www.stata.com/meeting/germany14/abstracts/materials/de14_jann.pdf
to plot coefficients of age.
I did
logit Y i.age
eststo plotplot
coefplot plotplot
But the plot has age on vertical axis, and coefficient on horizontal axis. It should be the opposite.
Also, I only want the coefficients of age, not constant or anything else. How do I do that?
And does "at" command work for discretized explanatory variable like i.age ? Or is it the case that it just takes so much time (I had to stop it because it was taking so much time)?
The logit coefficients are not particularly interpretable, so it is not clear why you want to plot them. There is also no at
command, though the margins
command has an at()
option. If you do want the index function coefficients, just omit the margins
step from below. coefplot
and marginsplot
do very similar things, though coefplot
is considerably more flexible.
I am guessing you have something in mind like this:
sysuse auto, clear
xtile mpg_tercile = mpg, nq(3)
logit foreign i.mpg_tercile weight
margins, dydx(mpg_tercile) at(weight = 3000) post
coefplot, vertical
marginsplot, recast(scatter)
This calculates the average marginal effect of falling in the middle or top third of the mile per gallon distribution on the probability of the car being foreign with weight of 3000.