Dynamically Integrating Macro and Micro Processes

描述（由申请人提供）：社会世界的许多方面都对重要的健康结果有影响，但对大多数人来说，其影响的确切性质仍不清楚。虽然通过统计方法已经了解了很多，但文献缺乏对宏观和微观过程如何相互关联影响健康的因果理解。本研究采用了一种空间明确的基于主体的建模方法，并借鉴了社会学、地理学和经济学的见解。动态的社会网络是内生性的，这是一个特别创新的元素。通过在一种设置中构建模型，将开发一组建模工具，这些工具将适用于广泛的设置。这些新工具将通过一个公共网站提供。一旦构建了基于主体的模型，它将被用于研究当地环境的社会、空间和生物物理维度之间的相互联系，同时考虑移民、居住选择、土地利用和家庭财富，以及其中的反馈。在这个过程中，人们将更好地理解社区效应标准统计模型中过度简化的后果。该模型还将用于进行有关婴儿死亡率突然变化和经济状况变化的影响的实验。该模型将进行两次现场测试，第一次验证关键模型假设，第二次探索意外结果。事实上，基于代理人的方法的一个优点是在系统或社区一级“出现”的可能性（即，宏观和微观过程的结合产生新的结构，而不是基于个人和家庭行为的简单汇总而预期的）。最后，借用和扩展气象学和控制理论领域的工具，将开发、应用和广泛提供比目前使用的计算效率更高的测试模型灵敏度的新方法。拟议的项目将开发工具，研究涉及个人、家庭、社会网络和社区的与健康有关的社会进程。这些工具的应用将有助于我们更好地理解和解释将社区因素与健康结果联系起来的研究文献，并为标准统计方法提供补充。

DESCRIPTION (provided by applicant): Many aspects of the social world bear on important health outcomes, but for most, the exact nature of their impact is still unclear. Although much has been learned through statistical approaches, the literature falls short of a causal understanding of how macro and micro processes interrelate in affecting health. The proposed research uses a spatially explicit agent-based modeling approach, informed by insights from sociology, geography, and economics. Dynamic social networks are included endogenously, an especially innovative element. Through the construction of a model in one setting, a set of modeling tools will be developed that will be applicable across a wide range of settings. These new tools will be made available through a public website. Once the agent-based model is constructed, it will be used to study interconnections among the social, spatial, and biophysical dimensions of the local context, taking into account migration, residential choice, land use, and household wealth, and feedbacks therein. In the process will come a better understanding of the consequences of oversimplification in standard statistical models of community effects. The model will also be used to conduct experiments about the effects of a sudden shift in infant mortality and shifts in economic conditions. The model will be field tested twice, the first time to validate key model assumptions and the second time to explore unexpected results. Indeed, an advantage of the agent-based approach is the possibility of "emergence" at the system or community level (i.e., the integration of macro and micro processes producing new structures not anticipated based on a simple aggregation of individual and household behaviors). Finally, borrowing and extending tools from the fields of meteorology and control theory, new methods for testing model sensitivity that are more computationally efficient than those now in use will be developed, applied, and made broadly available. The proposed project will develop tools to study social processes involving individuals, households, social networks, and communities in relation to health. The application of these tools will help us better understand and interpret the research literature connecting community factors with health outcomes and provide a complement to the standard statistical approaches.