Interpreting marginal effects in probit

[qocande]

Hi all,

I need to provide proper explanation/interpretation of marginal effects in probit model. If anybody knows about this, please advise or share link(s) to relevant resources.

Thank you

Le

[Maarten Buis]

This is potentially a very large subject. Could you narrow it down by providing some context on why you want to know this?

[qocande]

Hi Marteen and all, Sorry for the general question, I want to know this as I am using probit for modeling my data and reporting marginal effects at means. I need to make sure I explain/interprete marginal effects correctly and simply (i.e., easy to understand for those who are not familiar with probit). More specifically: - I understand that marginal index effect is actually another way to say coeficient assuming zero conditional mean error. Is this a correct statement? What if the zero conditional mean assumption is not met? - Margin command after probit command produces marginal probability effects not marginal index effects? Is this correct statement? - Marginal probability effects are the partial effects or percentages of changes each explanatory variable has on the probability that the observed dependent variable Y equals 1? Is this a correct statement ? If any statements above are not correct, please help correct it. If you do not mind I can copy an example of my stata probit outputs re marginal effects and my corresponding explanations for your review. Thanks Le

[Richard Williams]

Here are some introductory materials:

http://www3.nd.edu/~rwilliam/xsoc73994/Margins01.pdf

http://www3.nd.edu/~rwilliam/xsoc73994/Margins02.pdf

http://www3.nd.edu/~rwilliam/xsoc73994/Margins03.pdf

[qocande]

Thanks Richard for the useful links. Le