mcrobust

To apply this routine, please copy the relevant ado files into your personal ado folder, typically, c:\ado\personal\ . Please cite the following paper as the source: Chunmian Ge, Ke-wei Huang, and I.P.L. Png, "Engineer/Scientist Careers: Patents, Online Profiles, and Misclassification Bias", NUS Business School, June 2014.

mcrobust: Check of robustness to misclassification. This routine applies to a binary dependent variable, which is subject to misclassification. It computes the rates of false positives and false negatives of the dependent variable that result in the estimated coefficient of the explanatory variable(s) of interest taking the opposite sign of that in the estimate ignoring misclassification. The routine comprises the following ado files:

• mcrobust.ado

• stip_estimate.ado

• stipulate_fnfp.ado

• bisearch_zero.ado

This routine is based on bootstrap regressions. It can take a long time to run, so, you might first explore it with low granularity (say fpinc(2) and fninc(2) and relatively few simulations, say num_sim(20). Please also refer to the help file.