Hi Arash, The p-values from assocTestRegression are indeed two-sided. (The function uses pchisq to calculate them.) Thanks, Adrienne On Jul 13, 2014, at 10:25 AM, Arash Nazeri <arashnazeri at="" gmail.com=""> wrote: > Dear Adrienne, > > We have been using GWASTools for an interaction model with sandwich standard errors (model robust). We wanted to make sure that the p values generated from the Wald tests are two-sided or not. > > My guess is that this is an inherent issue with Wald tests, as there is no directionality left. > > Any help is greatly appreciated. > > Cheers, > Arash