I learn that fuzzy regression discontinuity (FRD) is like IV estimation. STATA provide the command rdrobust for non-parametric estimation. I want to know how to test weak instrument when I use rdrobust with p(2).
Below is an example of the STATA code of some previous papers. For linear polynomial regression (i.e., p(1)), they use three tests (i.e., test Conventional, test Bias, test Robust). It seems that these tests are to calculate F-value for testing weak instrument? If yes, can I use them for the case of quadratic polynomial regression, namely, p(2)? If not, how can I test weak instrument for the case of p(2) after using rdrobust?
. rdrobust y x, fuzzy (z) all p(1)
. test Conventional
. test Bias
. test Robust
Many thanks
Below is an example of the STATA code of some previous papers. For linear polynomial regression (i.e., p(1)), they use three tests (i.e., test Conventional, test Bias, test Robust). It seems that these tests are to calculate F-value for testing weak instrument? If yes, can I use them for the case of quadratic polynomial regression, namely, p(2)? If not, how can I test weak instrument for the case of p(2) after using rdrobust?
. rdrobust y x, fuzzy (z) all p(1)
. test Conventional
. test Bias
. test Robust
Many thanks
Comment