Debiasing through raising awareness reduces the anchoring bias
Carolyn Mair, Martin Shepperd · University of the Arts London Research Online (University of the Arts London) · 2014
Poster presented at BPS Annual Conference 2014. Objective: To investigate the impact of the anchoring bias and debiasing interventions with professional, expert software engineers. We hypothesised that debiasing strategies would reduce bias. Method: Ethical approval was given prior to conducting the study. Five experiments were conducted with software engineers (N=118) in two locations. At each location, participants attended a workshop raising awareness of the impact on estimation of a range of cognitive biases, including anchoring. Immediately following the workshop, the participants at each location were divided into two groups (high versus low anchor) and asked to provide an estimate of effort for a familiar task. Data were analysed using Mann-Whitney tests. To test the impact of the workshop on the anchoring bias, effort estimation data collected previously with different samples of software engineers at each location, were compared with effort estimation data following the workshop (ANOVA). Results: The anchor had a large and significant impact (effect size, large, 0.3; Mann-Whitney, p<0.0001) on the effort estimates. The impact of the anchor was significantly reduced following the workshop (effect size, large 0.14; two-way ANOVA, p<0.0001). Conclusion: The impact of the anchoring bias on expert judgement is substantial and difficult to eradicate. Nevertheless, a debiasing activity such as attending a workshop to raise awareness can significantly reduce the effect. The study is limited in that the samples involved in the analysis of the impact of the workshop on effort estimation were independent. This will be addressed in further work.