Development and Validation of Agent-Based Models of Alcohol-Related Violence

描述（由申请人提供）：拟议的研究项目将开发和应用基于主体的模型（ABMs）来研究酒精出口密度与暴力犯罪之间的关系。具体来说，该研究将于2009年9月23日在德克萨斯州拉伯克市引入酒类销售网点（或包装好的商店），以验证基于过去五年档案数据开发的基于理论的ABMs。在此日期之前，在拉伯克市范围内禁止出售非销售消费的酒类。在这项禁令被解除的一年内，德州酒精饮料委员会在该市颁发了161个销售许可证。酒精供应的这种变化提供了一个机会，可以对拉伯克的暴力犯罪进行基于计算机的基于代理人的模拟，并评估这些模拟是否能够以足够的准确性再现并确实预测该市犯罪的经验空间模式。拟议研究的具体目标建立在研究小组成员之前对酒精出口密度和暴力犯罪的研究基础上。这些目标解决了一些基本的理论问题，这些问题涉及酒精供应增加导致暴力行为增加的机制，这些问题在以前的统计研究中没有得到解答。具体来说，拟议项目的两个主要目的如下。首先，开发基于主体的模型，生成在引入折扣网点之前的五年间（2004年9月23日至2009年9月22日）拉伯克暴力犯罪的地理空间格局。其次，利用这些基于主体的拉伯克市暴力犯罪模型来预测在引入代销店后的两年内（2009年9月23日至2011年9月22日）暴力犯罪的地理空间分布。该项目将使用四种理论（日常活动理论、社会解体理论、社会影响/选择理论和酒精可得性理论的改进版本）来为基于主体的模型的开发提供信息。模拟模型的质量将根据它们产生的模式与实际暴力犯罪数据中发现的模式相似的程度来评估。仿真模型将在许可变更前和许可变更后的时间段内以每日时间步运行。模拟城市的季度暴力模式将与真实城市的季度暴力犯罪数据进行比较，使用描述性统计和虚拟和真实地图的视觉检查，以及k函数和核密度估计方法。

DESCRIPTION (provided by applicant): The proposed research project will develop and apply agent-based models (ABMs) to the study of the relationship between alcohol outlet densities and violent crime. Specifically, the research will use the introduction of off-sales alcohol outlets (or package good stores) in the City of Lubbock, Texas on September 23, 2009 to validate theoretically-based ABMs developed using archival data from the previous five years. Prior to this date, the sale of alcohol for off-sale consumption was prohibited within Lubbock city limits. Within a year of this prohibition being lifted, the Texas Alcoholic Beverage Commission had issued 161 off-sale licenses in the City. This change in alcohol availability provides an opportunity to develop computer-based agent-based simulations of violent crime in Lubbock and to assess whether these can reproduce, and indeed predict, empirical spatial patterns of crime within the City with sufficient accuracy. The specific aims of the proposed research build on members of the research team's previous studies of alcohol outlet density and violent crime. The aims address some of the fundamental theoretical questions concerning the mechanisms whereby increased alcohol availability leads to increased violence that have been left unanswered by previous statistical studies. Specifically, the two main aims of the proposed project are as follows. First, to develop agent-based models that can generate the geospatial patterns of violent crime in Lubbock for the five- year period (September 23, 2004 to September 22, 2009) before the introduction of off- sale outlets. Second, to use these agent-based models of violent crime in the City of Lubbock to predict what happens to the geospatial distribution of violent crime over the two-year period (September 23, 2009 to September 22, 2011) following the introduction of off-sale outlets. The project will use four theories (routine activities theory, social disorganization theory, social influence/selection theory and a refined version of alcohol availability theory) to inform the development of the agent-based models. The quality of the simulation models will be assessed in terms of the degree to which they generate patterns similar to those found in the actual violent crime data. The simulation model will run on a daily time step for both the pre- and post-licensing change periods. Quarterly violence patterns for the simulated city will be compared to quarterly violent crime data for the real city using descriptive statistics and visual inspection of virtual and real maps, as well as the k-function and kernel density estimation methods.