How to Interpret Dummy Variable and Constant Coefficients in Fractional Outcome Probit Model

[Ervin Boyes]

Hi all,

I have a regression model where I am using the percentage of the population that committed suicide as the DV, and as the IV I am using the percentage of the population that is Urban. I also have a dummy variable for males only to see the difference between males and females.

I was wondering how this coefficient should be interpreted? I know that with the other IVs I have to convert the coefficients to margins using 'margins, dyex(_all)', but if I try and convert the dummy variable the number I get is completely wrong; I know by looking at the data theres is a massive gap between males and females but the marginal effect is so tiny that it cannot be correct. I notice that it does not convert the constant term either. Do I therefore just interpret the original coefficient for the dummy variable and constant term?

Thanks for the help!

[Jared Greathouse]

Fractional Outcome Probit Model

?

[Ervin Boyes]

?

What I mean by this is using the fracreg probit command

[Jeff Wooldridge]

How is the outcome variable a proportion (not a percent, I assume) representing a population and one of the explanatory variables measured at the individual level? I can't envision how that comes about. Please follow the FAQ and use -dataex- to show an extract of the data. Also, you're supposed to show what you typed and what Stata gave you, in between code delimiters (# #).

Might you have the proportion of the population reported as male and not a dummy variable?