This analysis critiques traditional statistical significance testing for covariables, suggesting alternative approaches for better understanding confounding effects.
How badly a covariable confounds the association between a disease and a primary trait depends on the strength of two other associations: between covariable and disease, and between covariable and primary trait. To assess the confounding potential by testing one of these associations for statistical significance, using a 'traditional' critical level, is inadequate. Such a preliminary test places the burden of proof in the wrong direction, ignores a major aspect of confounding potential, and may be inappropriately influenced by other factors. If it must be used at all, the significance level should be 0.25 or 0.50, possibly even higher.
No takes yet. Share an insight, caveat, or question.
Dales et al. (1978) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: