Hi all, I am currently using stcox to analysis the timing of turning points of US stock market cycles relative to the presidential elections. However, the relatively small number of stock market cycles makes inference based on standard asymptotic suspect. Therefore, I want to use bootstrap techniques. Because of the interested variable is a categorical variable, the standard method, case resampling, may not work, I would like to know how to do perform residuals resampling instead. Is there any related example available in Stata? Many Thanks in advance.