Effect of the directionality of verbal probabilities in a shoot/don’t shoot task

Juanchich, Marie, Gourdon-Kanhukamwe, Amélie, Riege, Anine and Sirota, Miroslav (2019) Effect of the directionality of verbal probabilities in a shoot/don’t shoot task. In: '27th Subjective Probability, Utility and Decision Making (SPUDM) Conference; 18-22 Aug 2019, Amsterdam, Netherlands. (Unpublished)

Abstract

Past research has shown that framing uncertainty using the directionality of verbal probabilities affect decision making in vignette studies. We tested the effect of directionality in a “decision based on experience” methodology, where participants could experience the decision situation and the consequence of their decision. In three experiments (N = 87, 75 and 55) participants made a series of decisions on whether to shoot a suspect or not based on a prediction. In all experiments the procedure of each decision trial was the same: participants read a prediction that quantified the risk of the suspect having a gun, then, participants were shown the suspect (with a gun or not) and had a second to decide whether to shoot or not, followed with the provision of feedback. The probability of the suspect having a gun was calibrated onto the probability conveyed in the prediction. We manipulated the probability conveyed by the prediction and its directionality in a 4 (probability level: very low, low, high, very high) × 2 (directionality: positive vs. negative) within-subjects design in which participants made 8 decisions per prediction, totalling 64 decision trial organised in 8 blocks of the 8 experimental conditions. In Experiment 2, we also manipulated the degree to which participants could see the item that the suspect hold and in Experiment 3 we withdrew feedback. Directionality and probability conveyed had an effect on decision in the three experiments. The effect of directionality was the same whether the image of the suspect was net or blur and across blocks of decisions. However, the effect of directionality was larger when we withdrew performance feedback in Experiment 3 compared to Experiments 1 and 2 (+4%, +3% and +13%) indirectly pointing out the role of experience.

Actions (Repository Editors)

Item Control Page Item Control Page