Instructing participants about the random assignment of patients to treated and non-treated conditions does not diminish causal illusions

Abstract

People sometimes perceive causal relationships between non-contingent events. When having to assess the contingency between a putative cause and an outcome, it is vital to ensure that all other causal forces are held constant whether the studied cause is present or not. Nevertheless, a recent work suggested that, in conventional contingency learning scenarios, people do not necessarily assume that is the case. A possible contributing factor to this asset is that instructions in contingency learning tasks do not typically clarify this point. In two experiments, we manipulated the task instructions so that only half of the participants were explicitly informed that the introduction of the putative cause was randomly decided for each trial. The second experiment further instructed participants in the implications of random assignment regarding the control of alternative causes. Results of both experiments indicated that the manipulation of the instructions had no impact on the strength of causal illusions (minimum BF₀₁ = 5.853). Nevertheless, the susceptibility to develop causal illusions was related to a lack of an appropriate consideration of alternative causal forces and a tendency to overweight the importance of the probability of the outcome in the presence, rather than in the absence, of the putative cause.