Causal implicatures from correlational statements

Abstract

Correlation does not imply causation, but this does not necessarily stop people from drawing causal inferences from correlational statements. We show that people do in fact infer causality from statements of association, under minimal conditions. In Study 1, participants interpreted statements of the form "X is associated with Y" to imply that Y causes X. In Studies 2 and 3, participants interpreted statements of the form "X is associated with an increased risk of Y" to imply that X causes Y. Thus, even the most orthodox correlational language can give rise to causal inferences.

Fig 1. Summary of results for Study 1 (left) and Study 2 (right).

Participants were given statements such as ‘Themaglin is associated with Pneuben’ (Study 1) or ‘Denoden is associated with an increased probability of Flembers’ (Study 2). A response of 1 indicates participants interpreted the statement to mean the first variable causes the second, 0 that they took it to mean the second variable causes the first, and 0.5 that they answered at random. The figure shows that without context, simple association is taken to imply the second variable causes the first. With minimal context, the association statement is taken to strongly imply the first variable caused the second. Error bars show standard error estimates for a proportion, $\sqrt{p(1-p)/N}$, where p = M/N, M is the number ‘Yes’ responses, and N is the total number of responses.

Fig 2. Summary of results for Study 3.

Participants were given the same statements as in Study 2A, with the addition of a ‘Neither’ option to report that no causal direction was preferred. The y-axis now indicates the proportion of participants who chose each response. Error bars show standard error estimates for a proportion. The dotted line indicated expected random response levels.