Towards a Resolution of Statistical Dogmas: Beyond p-values, Normality, and Big Data

Statistics has been dominated by dogmatic practices: the fetishiza- tion of p-values, the overextension of the normal distribution, and the blind faith in big data. This paper argues that these paradigms must be discarded in favor of a pragmatic, structural, and temporal view of inference. We demonstrate, through mathematical formulations, why these dogmas misrepresent reality and how a principle of contextual sufficiency can replace them.