PECASE: Taking Attitudes Seriously: A Multi-Contextual Approach to Behavioral-Modeling

交通和土地使用的行为模型被用来为规划和政策制定提供信息，并预测未来的情景。离散选择模型是一个关键组成部分，解释了利益的选择（人和公司的空间位置，模式，活动等）。作为备选方案的可观察属性（如时间和成本）以及决策者的可观察特征（如家庭规模和收入）的函数。长期以来，行为科学家一直强调态度作为推动许多决策过程的“情感核心”的重要性。尽管他们的重要性，这样的行为结构并没有很好地代表在离散选择范式。本研究的目的是通过建立在行为科学的研究成果，并结合结构方程模型和心理测量数据的方法，有效地整合到离散选择分析的态度结构。该模型系统使用多个方程和最大似然估计，同时估计离散选择和潜变量模型。该框架提供了优于现有方法的几个优点。首先，它可以结合离散选择分析（例如，灵活的错误结构和复杂的选择集）以及线性和非线性结构方程模型的最新发展。其次，它产生一致的和有效的估计的影响，态度的选择。第三，它提供了一种同时推断每个人的一组内在态度以及这些态度如何影响一些相互关联的决策的方法。这种方法已在有限的情况下，并将进一步发展和扩展在本研究计划。扩展包括：（i）对一个人的固有态度及其对一些相互关联的旅行，活动和土地使用决策的影响进行建模，以及（ii）对这些态度随时间的趋势性质进行建模。研究目标是证明该方法的可行性，展示其附加价值，并使其易于为研究人员所用。教育计划使用态度和行为领域作为实践的重点，小组研究活动旨在教授科学方法并加强课堂上教授的基本原则。将为研究生、本科生和中学班级开发教育模块，重点是探索学生自己的行为以及他们的家人和同事的行为。研究和教育计划通过态度和行为研究领域和一个共同的数据集进行整合，从而允许通过研究和教育的各个层面的演示和指导进行互动。该教育计划有助于培养对科学探索感到兴奋的学生，并掌握科学方法的技能，从而增加他们从事科学和工程职业的可能性。该研究计划有助于更好地了解交通基础设施和城市规划的发展，为城市扩张，交通拥堵，环境质量和紧急疏散等问题提供见解。所提出的方法承诺丰富的行为定量模型，从而改善预测，更好地为规划，管理和政策制定提供信息。利用这种方法的估计软件将被开发并免费提供，为各种领域的研究人员打开大门，以丰富具有态度结构的行为模型。

Behavioral models of transport and land use are used to inform planning and policy making and forecast future scenarios. Discrete choice models are a key component and explain the choice of interest (spatial location of people and firms, mode, activities, etc.) as a function of observable attributes of the alternatives such as time and cost and observable characteristics of the decision maker such as household size and income. Behavioral scientists have long emphasized the importance of attitudes as the "emotional core" driving many decision-making processes. Despite their importance, such behavioral constructs have not been well represented in the discrete choice paradigm. The objective of this research is to effectively integrate attitudinal constructs into discrete choice analysis by building on the findings of behavioral science and incorporating the methods of structural equation modeling and psychometric data. The model system uses multiple equations and maximum likelihood estimation to simultaneously estimate discrete choice and latent variable models. This framework provides several advantages over existing approaches. First, it can incorporate the latest developments in discrete choice analysis (for example, flexible error structures and complex choice sets) as well as linear and non-linear structural equation models. Second, it produces consistent and efficient estimators of the influences of attitudes on choice. Third, it provides a means of simultaneously inferring both a set of intrinsic attitudes for each individual and how these attitudes affect a number of interrelated decisions. This methodology has been employed in limited cases and will be further developed and extended in this research program. Extensions include (i) modeling the set of intrinsic attitudes of a person and their influence on a number of interrelated travel, activity, and land use decisions, and (ii) modeling the trending nature of these attitudes over time. The research goals are to prove the feasibility of the approach, demonstrate its added value, and make it accessible to researchers.The educational program uses the domain of attitudes and behavior as the focus for hands-on, group research activities aimed at teaching the scientific method and re-enforcing basic principles taught in class. Educational modules will be developed for graduate, undergraduate, and secondary school classes that focus on exploration of the students own behavior and that of their family and colleagues. The research and educational programs are integrated through the domain of attitude and behavior research and a common dataset, thereby allowing for interaction through presentations and mentoring across all levels of the research and education. The education program contributes to the development of students excited about scientific exploration and with skills in the scientific method, thereby increasing their probability of careers in science and engineering. The research program contributes to better understanding of the development of transportation infrastructure and urban planning, providing insight into issues of urban sprawl, traffic congestion, environmental quality, and emergency evacuation. The proposed methods promise enriched quantitative models of behavior, thereby improving forecasts and better informing planning, management, and policy making. Estimation software that makes use of this approach will be developed and made freely accessible, opening the door for researchers in a wide variety of fields to enrich behavioral models with attitudinal constructs.