ivprobit result coefficient interpretaion and comparison

[Stanly Lin]

Dear Listers and Statistic expert,

I have ivprobit with two step estimator, my dependent variable is failure or not (1,0) , Endogenous variable is watch_percentage (continous) ,instrument variable weather(1,0) and the other variable(such as new_pattern(binary variable))

My variable named new_pattern which have seven type (0,1,2,3,4,5,6) ,I also set "1" on based

and I want to compare my exogenous variable(new_pattern) to know which pattern will lead to failure relative on pattern "1".

I have refered to other posts or stata manual and found the "margins" instruction.

The result on new_pattern margin (dy/dx) is below：



How can I interpret the result : such as when new_pattern = 3 the failure probability will increase 37% based on pattern 1

If my saying is incorrect, how can I compare it which pattern is higher or lower than pattern 1

Thanks for help! Really appreciate!

[Stanly Lin]

attach is the coefficient on ivprobit result!