Round effects in economic experiments—A novel probabilistic programming approach to round-dependent response behaviour
研究多轮实验中受访者响应行为随轮次变化（轮次效应）的问题，提出一种概率编程方法来捕捉偏好学习、制度学习和疲劳效应三种类型，并展示其适应性和透明度。
Abstract Respondents in multi-round experimental studies can show round-dependent response behaviour, subsumed as round effects, which can distort the analysis of study results. In this paper, we develop a novel probabilistic programming approach to capturing round effects. We apply our model to experiment data, investigating the relevance of three theoretically defined types of round effects: preference learning, institutional learning and fatigue effects. By using a Bayesian modelling workflow featuring the specification of a data generating process and extensive model inspection prior to data analysis, we demonstrate the high adaptability and transparency of our approach, as well as its value for future experimental research.