Dear R-listers, I am trying to compute interaction effects in a probit model, and conduct hypothesis tests on these effects correctly.
Specifically, I have a model of the form y = a + b1 m + b2 x + b3 m*x, where both y and m are 0-1 dummies, x is continuous, and I am interested in the sign and statistical significance of the marginal effect for observations for which m=1, over the base (obs for which m=0). Then, I also want to conduct a test of the hypothesis H0: b2 + b3 = 0. As we know, probits are non-linear models, and so coefficients of interaction variables do not carry the usual interpretation of marginal effects over the base. Instead, I think what one needs to do is compute the cross-derivative of y with respect to x and m. Is there an efficient way of doing this in R preferably after having estimated this model through glm(<formula>, family = binomial(link = probit))? Can anyone suggest any code? Thanks! -- Dr. Tobias Mühlhofer Assistant Professor Department of Finance Kelley School of Business Indiana University Tel: +1 812 855 9270 http://tobias.muhlhofer.com ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
