Forecasting the outcome of a time-varying Bernoulli process: Data from a laboratory experiment

• Main Authors: Mel W. Khaw, Luminita Stevens, Michael Woodford
• Format: Article
• Language: English
• Published: Elsevier 2017-12-01
• Series: Data in Brief
• Online Access: http://www.sciencedirect.com/science/article/pii/S2352340917305243

The data presented in this article are related to the research article entitled âDiscrete Adjustment to a Changing Environment: Experimental Evidenceâ (Khaw et al., 2017) [1]. We present data from a laboratory experiment that asks subjects to forecast the outcome of a time-varying Bernoulli process. On a computer program, subjects draw rings with replacement from a virtual box containing green and red rings in an unknown proportion. Subjects provide their estimates of the probability of drawing a green ring. They are rewarded for their participation and for the accuracy of their estimates. The actual probability of drawing a green ring is initially drawn from a uniform distribution. It then changes intermittently throughout the session, and each subsequent probability is an independent draw from the uniform distribution. Each session involves 1000 ring draws. The dataset contains the values of the underlying probability, the sequence of ring draws that are realized, and the subjectsâ estimates and response times. The dataset contains the performance of 11 subjects who each completed 10 sessions over the course of several days.