Practical tools for using evidence of mechanisms to establish correlations

Abstract

Abstract. Evidential pluralism is the view according to which we need to think about varieties of evidence. Specifically, to establish causal claims (in medicine and elsewhere), we need evidence of correlation and evidence of mechanisms. The publication of Parkkinen (2018) marked an important moment in the debate, establishing EBM+ as a methodological approach able to expand mainstream methods of evidence-based medicine to seriously consider evidence of mechanism, alongside evidence of correlation. After that, however, some publications strengthen the distinction between evidence of correlation and evidence of mechanism. In this paper, we aim to return to the original formulations of evidential pluralism and to stress how much evidence of correlation and of mechanism are intertwined, rather than separated. Our thesis here is that evidence of mechanism is in fact necessary to establish good quality evidence of correlation, in the sense that considerations about mechanisms inform the choice of variables and their measurement in essential ways, in any correlational study. We also explain why practical tools that were originally developed in Parkkinen et (2018) can prove very useful in assessing evidence, without presupposing strict and artificial separations of kinds of evidence, and even in cases with partial knowledge. We use research on the effects of vitamin D deficiency as an illustration, although our arguments are very general in character.