Find corresponding X for a threshold of Y in estimated model

[Leon Nguyen]

Dear Statalist,

I have estimated a model like this:

elec= -12.83927*ldisp + .7345735*ldisp² + (other variables)

What I wanna do is to find the corresponding value of ldisp at a threshold level of elec, say elec=0.7 (this is not turning point). I could solve equation -12.83927*ldisp + .7345735*ldisp² + average of all other variables=0.7 to find ldisp, but I would like to know whether any command in Stata can help.

Thank you in advance!

[Clyde Schechter]

Use the quadratic formula. If ax² + bx + c = 0, then x =[ - b +/- sqrt(b² - 4ac)]/2a.

You can do the calculation with locals or scalars.

[Leon Nguyen]

Use the quadratic formula. If ax² + bx + c = 0, then x =[ - b +/- sqrt(b² - 4ac)]/2a.

You can do the calculation with locals or scalars.

Thanks Clyde, that's what I have done. Just curious whether any other way to do it quicker :D
Clyde Schechter about parameter "c' in that formula, I simulate effects of all other variables at average level by parameter "c", do you think that is a appropriate way?

[Clyde Schechter]

I simulate effects of all other variables at average level by parameter "c", do you think that is a appropriate way?

I'm not sure exactly what you mean by that. The value of c in the equation should be the term representing the average contributions of all the other variables to the regression, and then subtract 0.7. I don't know in what sense you say you "simulate" that. I think the simplest way to calculate that is

margins, at(ldisp = 0) atmeans

Note: The above code will only work correctly if you used factor variable notation in your original regression.