Cinco preguntas para precautelar la honestidad estadística en la sociedad actual

Abstract

This study has the purpose of proposing a general methodological strategy to avoid dishonesty derived from deception or statistical ignorance. To do this, we present a tour of some historical events that highlight the importance of applied statistics. It is often the case that non-experts tend to confuse correlation with causality. Any statistical relationship, strong and suggestive as it may appear, cannot establish a causal connection. Our ideas about causality must come from outside of statistics, ultimately from some theory that refers to the area of knowledge from which the data comes. To identify these problems, based on experience, we propose a simple methodological alternative, which consists of answering five key questions about the origin, quantity, variability, context and meaning of the data. The reflections extracted have important epistemological implications for applied statistics.