Noah Stovitz, lead author on a paper to appear in AJE Advances: Research in Epidemiology
Communicating research findings: how linking terms and context affect the interpretation of conclusions
When reporting results on the association between two variables, researchers need to effectively communicate the appropriate message about the evidence for a causal relationship. However, readers’ (and authors’) perceptions of causal implication often diverge from academic definitions of linking terms. For example, the linking term “predicts” is defined as a purely associational term (i.e., maybe causal or non-causal) in experimental psychology. Prediction research requires different methods than causal research, and predictors may not be causal. However, participants interpreted “predicts” as having some causal implication. Additionally, interpreting authors’ language does not depend entirely on the linking term used. Our participants’ interpretation of claims using the same linking term varied with contexts that were expected to have different strengths for their causal relationships. Therefore, considering the context of a claim may change how we want to phrase our message. In contexts where there is a consensus that the effects may be causal (or not causal) based on theory and/or prior research, we should be particularly careful with the language used.
