Margins with Fractional Logit

[Shazmeen Maroof]

Dear All,

I am stuck with interpreting and correct usage of margins command even after reading the previous related threads and Stata manual. I couldn’t find an exact answer of my below query:

I am estimating a fractional logit model where my dependent variable is in proportions (i.e., share of FDI received by a country). The explanatory variables comprise of:

a) continuous variables i.e., GDP
b) discrete variables i.e., score of countries on human rights indicator running from 1 to 5.

I am using the following Stata command:

glm FDIShare GDP HR, family(binomial) link(logit)

For easier interpretation I am using the following margins commands.

estpost margins, dyex(_all)


My interpretation of coefficients after using the above margins command.

1. 1% increase in GDP will result in b1% reduction in FDI share, holding other variables constant
2. 1 point score increase in HR score will result in b2% reduction in FDI share, holding other variables constant

Where b1 is the coefficient on GDP after using margins and b2 is the coefficient on HR.

I suspect there is something wrong with either my interpretation or the command I am using. Any help will be appreciated.

Best regards,
Shazmeen Maroof.