Statistical significance describes an observed pattern that would be relatively unlikely under a specified null model. It can make a claim worth investigating, but it does not automatically show that one factor caused the result. section 1.4
A successful pricing system—or a departing competitor?
A director claims that a new pricing system boosted revenue by $70 million because an analysis found statistical significance. That result may be real, yet a competitor’s retreat from several markets could have delivered the extra customers instead. Statistical significance cannot replace checking alternative explanations. section 1.4
Read the number alongside sampling bias, process assumptions, ignored variables, aggressive outlier removal, logical fallacies, and the incentives behind the study. Statistical results can be shaped toward a desired finding or pressured into revision. section 1.4
