Marginal effect of a continuous variable in probit model

[Sayoree Gooptu]

I have conducted a probit regression of school choice (1 if government school and 0 private) on the share of private tuition expenditure in MPCE. Using margins command in stata, I have calculated the marginal effect at the median MPCE. which comes out to be 0.1. Do I interpret it as "the marginal effects as 'change in probability of attending government school due to one unit of change in the share is 0.1 percentage points"?

I am not sure whether I should interpret the change in X which is the share of private tuition expenditure in MPCE as a change of 1 unit, since it is a share. Any help would be really beneficial.

[Jeff Wooldridge]

Sayoree: You're on the right track. If x is the variable you're interested in, and it's restricted to be in the interval [0,1], then increasing it by one is too large a change if you're using the derivative and the calculus approximation. A 0.1 increase in x seems reasonably. So take the marginal effect and multiply it ny 0.1. Then you get the change in the probability given a 0.1 increase in MPCE (which is a 10 percentage point increase in MPCE).