Expectancy Effects Threaten the Inferential Validity of Synchrony-Prosociality Research

Abstract

Many studies argue that synchronized movement increases prosocial attitudes and behavior. We reviewed meta-analytic evidence that reported effects of synchrony may be driven by experimenter expectancy, leading to experimenter bias; and participant expectancy, otherwise known as placebo effects. We found that a majority of published studies do not adequately control for experimenter bias and that multiple independent replication attempts with added controls have failed to find the original effects. In a preregistered experiment, we measured participant expectancy directly, asking whether participants have a priori expectations about synchrony and prosociality that match the findings in published literature. Expectations about the effects of synchrony on prosocial attitudes directly mirrored previous experimental findings (including both positive and null effects)-despite the participants not actually engaging in synchrony. On the basis of this evidence, we propose an alternative account of the reported bottom-up effects of synchrony on prosociality: the effects of synchrony on prosociality may be explicable as the result of top-down expectations invoked by placebo and experimenter effects.

Keywords: expectancy; experimenter bias; placebo effects; prosociality; synchrony.

Figure 1.

Participants’ expectations about synchrony match reported effects of the experience of synchrony, from a previous study. Each set of bars represents participants’ ratings of emotions and attitudes toward other people in one of five different categories (x-axis) after actually walking with them in synchrony (Experienced Synchrony; data from Wiltermuth & Heath, ; Exp. 1); actually walking normally/out of synchrony (Experienced Control; data from Wiltermuth & Heath, ; Exp. 1); imagining how other people would expect to feel after walking in synchrony (Imagined Synchrony); or imagining how others would feel after walking normally (Imagined Control). Error bars represent 95% confidence intervals. **p < 0.00625, ˆp < 0.05.