The tolerance of this library is e^-16. Thus, I think it is not possible to get a p-value in this case. In statistics, p-value is the probability of success under conditions of null hypothesis. We require to compare it with 0.01, 0.05, and 0.1 under different alpha-value. Thus, it is useless to use any other value like 0.005. It indicates that assuming null hypothesis is true is misleading. If we use a infinitesimal small probability of making type-I error then the probability of making type-II error will be significantly high.
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…