exploring a quadatic variable post logit estimation

[Doug Hess]

I am using age and age squared in a voter turnout model using survey data. (Stata 15.1). For a simple example:

logit voted i.year c.age##c.age i.race i.gender i.income

Age has a positive coefficient and age squared a negative coefficient: voting increases with age, but voting declines for people over a certain age.
Questions: 1) what's the best way to calculate the point at which age-squared takes over age, so to speak, from the logit estimate? 2) What's the best way to graph this?
Thank you.

[Clyde Schechter]

the point at which age-squared takes over age

Taken at face value, and given that age is necessarily non-negative, age squared becomes greater than age when age > 1. I suspect that's not what you meant, though. Are you looking for the value of age associated with the maximum probability of voted, i.e. the turning point? If so.

nlcom -_b[age]/(2*_b[c.age#c.age])

As for your second question, see -help marginsplot-.