Dear SAS forum members, I'm a master student in medicine and a newbie regarding statistics and working in SAS (I'm using version 9.4), so please, bear with me. In my cross-sectional study I've investigated the association between intake of coffee and facial pain among pregnant woman. The variable "coffee intake" comes in various forms (divided into number of cups, dichotomized, or cups as divided into 5 dummies with the first dummy being zero intake), and the variable "facial pain" is dichotomized (Yes/No). After adjusting for possible confounders, I've found a significant bimodal dose-response relationship. Even though I naturally strongly suspect that it is in fact coffee intake that is the predictor value and facial pain that is the outcome variable and not vice versa, I would like, if possible, to conduct a causality test in SAS to strengthen this hypothesis. I've read that there is something called Granger causality tests and a PROC VARMAX command respectively, but I have yet too little knowledge to know if such tests are applicable in this case. Would it be possible (and perhaps foremost meaningful) to do a causality test in this case, and how could it then be done in SAS? Sorry in advance if I've explained myself poorly or provided to little information.
... View more