I am running a probit model of math proficiency (proficient or not) on education expenditure (rupees), child gender (binary), student attendance ratio (proportion of school days attended) and other controls. I am interested in the marginal effects of the three variables mentioned.

Code:
probit math edu_exp i.child_gender attend
margins, dydx(edu_exp) post //1
margins, dyex(edu_exp) post //2
margins, dydx(attend) post //3
margins, dyex(attend) post //4
margins, dydx(child_gender) post //5
My questions are:

1. What's the difference in interpretation between lines commented as //1 and //2? Which one should be used for a sensible interpretation?
2. Is line //4 "valid"? In other words, does it make sense to ask "if attendance rises by 1%, what's the change in probability of being proficient in math"?

Finally, a question about exporting the results. Stata doesn't allow using dydx and dyex options in the same line. I am using user contributed -esttab- to export results to LaTeX. Is there a way to combine dydx and dyex with minimal manual work? Right now I am using:

Code:
margins dyex(edu_exp) post
est store results1

esttab results1  using "mypath", ///
    tex label replace title("my title")
How to include the results from -margins, dydx(child_gender) post- in this?