How to ask twenty questions and win: Machine learning tools for assessing preferences from small samples of willingness-to-pay prices
How to ask twenty questions and win: Machine learning tools for assessing preferences from small samples of willingness-to-pay prices
复制标题
DOI：
10.1016/j.jocm.2023.100418
复制
发表时间：
2023-05-27
影响因子：
2.4
通讯作者：
Kvam，Peter D.
中科院分区：
文献类型：
--
作者：
Sokratous，Konstantina;Fitch，Anderson K.;Kvam，Peter D.
主观价值长期以来一直使用二元选择实验来衡量，但支付意愿等反应可以成为评估个体差异风险偏好和价值的有效且高效的方法。托尼·马利 (Tony Marley) 的工作表明，动态随机模型允许根据二元选择之外的范式上的过程级数据对认知进行有意义的推断，但其中许多模型仍然难以使用，因为它们的可能性必须通过模拟来近似。在本文中，我们开发并测试了一种使用深度神经网络来估计难以处理的行为模型参数的方法。经过训练后，这些网络可以进行准确且即时的参数估计。我们比较了不同的网络架构，并表明它们准确地恢复了与效用、响应谨慎、锚定和非决策过程相关的真实风险偏好。为了说明该方法的有用性，随后将其应用于估计大量具有人口代表性的美国参与者样本的模型参数，这些参与者完成了包含 20 个问题的定价任务——这是一项用以前的方法无法完成的估计任务。结果说明了机器学习方法在拟合认知和经济模型方面的实用性，为量化稀疏数据中风险偏好的有意义差异提供了有效的方法。
Subjective value has long been measured using binary choice experiments, yet responses like willingness-to-pay prices can be an effective and efficient way to assess individual differences risk preferences and value. Tony Marley’s work illustrated that dynamic, stochastic models permit meaningful inferences about cognition from process-level data on paradigms beyond binary choice, yet many of these models remain difficult to use because their likelihoods must be approximated from simulation. In this paper, we develop and test an approach that uses deep neural networks to estimate the parameters of otherwise-intractable behavioral models. Once trained, these networks allow for accurate and instantaneous parameter estimation. We compare different network architectures and show that they accurately recover true risk preferences related to utility, response caution, anchoring, and non-decision processes. To illustrate the usefulness of the approach, it was then applied to estimate model parameters for a large, demographically representative sample of U.S. participants who completed a 20-question pricing task — an estimation task that is not feasible with previous methods. The results illustrate the utility of machine-learning approaches for fitting cognitive and economic models, providing efficient methods for quantifying meaningful differences in risk preferences from sparse data.