The impact of spin on the interpretations of abstracts of randomized controlled trials in the field of clinical psychology: an online experiment

Objectives: to assess the impact of spin on the interpretation of the results and conclusions reported in abstracts describing randomized controlled trials investigating behavioural treatments for depression. Methods: we selected 10 abstracts from published randomized controlled trials investigating behavioural treatments for depression that claimed a beneficial treatment effect despite one or more nonsignificant primary outcomes. We rewrote the abstracts to create a spinless version of each of the 10 abstracts. We recruited clinicians and clinical researchers via email invitations. Participants were randomly assigned to read either the ‘with spin’ (n = 91) or ‘without spin’ (n = 99) version of an abstract and answered questions about the study described in the abstract. The primary outcome was the participants’ perception of the effectiveness of the treatment proposed in the abstract. Results: the Bayesian analysis showed no convincing support for an effect of spin on the participants’ perception of the treatment effectiveness (posterior mean = 0.53, s.e. = 0.27, 95% CI [−0.01, 1.07]; posterior mode = 0.54, 95% highest density interval [−0.02, 1.06]). However, following the Bayes factor of 1.8, no convincing support for a null effect was found either. Conclusion: presence of spin in abstracts did not convincingly impact clinicians’ and clinical researchers’ perception of the effectiveness of the proposed treatment.