, 柳林, 周素红Spatial heterogeneity of micro-spatial factors' effects on street robberies: A case study of DP peninsula
XUChong, LIULin, ZHOUSuhong通讯作者:
收稿日期:2017-06-13
修回日期:2017-09-16
网络出版日期:2017-12-15
版权声明:2017《地理研究》编辑部《地理研究》编辑部
基金资助:
作者简介:
-->
展开
摘要
关键词：
Abstract
Keywords：
-->0
PDF (6534KB)元数据多维度评价相关文章收藏文章
本文引用格式导出EndNoteRisBibtex收藏本文-->
1 引言
犯罪问题尤其是城市犯罪问题是每个人都必须重视的问题。随着中国城镇化水平的不断提高,越来越多的人口集中在城市生活,城市的犯罪问题关乎到每个城市公民的切身利益,并严重影响城市的正常发展,对经济发展和社会和谐均带来巨大危害[1,2]。在犯罪学研究中,“理性选择理论”和“日常活动理论”是作为解释犯罪行为发生的重要理论。“理性选择理论”出自“情境犯罪预防”的概念,并综合了四种学术理论：社会学、环境犯罪学、经济学、认知心理学[3]。理性选择理论认为,犯罪人的所有决定都是根据期待要花费的精力和可以从犯罪中获得的回报,与被抓、被判刑的轻重所比较和平衡后而做出的[4]。“日常活动理论”由Cohn和Felson于1979年提出[5],有时也被称为“机会理论”。该理论认为犯罪的成功必须同时满足三个条件：潜在的犯罪者,合适的目标,犯罪防范的缺失。由于人们日常活动的变化使得这三种条件时空汇合,即同时被满足的机会增加,造成了犯罪行为的发生和犯罪率的提高。结合两个理论可以得出：通常潜在的犯罪者是在对受害者以及犯罪场所进行理性选择之后来确定是否实施犯罪行为的,由此可见犯罪场所作为犯罪发生的承载空间的重要性。是否理性的选择犯罪发生场所将很大程度决定犯罪是否发生。而近年来蓬勃发展的犯罪地理学恰恰是以其独特的地理学空间视角,对犯罪发生场所的空间位置进行了大量的研究工作,进一步发展了传统的“理性选择理论”与“日常活动理论”中关于犯罪产生的空间影响因素的研究与讨论。现有研究表明,犯罪行为的发生总在某些特定的空间环境下表现出集聚的态势[6-9]。在确定了犯罪发生空间上存在集聚特征现象后,部分****开始尝试探索犯罪的发生究竟和哪些特定的空间要素有关,也开始关注微观空间要素对于犯罪发生的影响,例如重要的空间节点（交通枢纽、车站）对于抢劫犯罪的影响[10,11],不同土地利用类型地块上的空间设施对于犯罪发生的影响等[12]。此外,对于空间要素对犯罪发生的影响程度在空间上存在异质性现象的研究成果也不断丰富[13-21]。在取得大量研究成果的同时还存在以下几点不足：首先,未能与相应的理论相结合。部分研究存在未对犯罪类型进行细化的问题,不同类型的犯罪在发生机理上存在区别,因此对空间环境的依赖程度大相径庭,作为一个整体进行研究难免会互相影响。其次,现有研究的尺度往往是由行政边界界定的,因此选取的因素多为行政单元的社会经普数据,如人口密度、居民平均收入等,微观的环境因素的差异性往往难以纳入考虑。其三,由于研究中考虑的因素多为宏观社会因素,研究结果未能对日常的警务活动给予指导。鉴于以上三点,选择街头抢劫犯这一犯罪类型,结合“理性选择理论”与“日常活动理论”来甄选对该类犯罪发生有强烈影响的微观空间要素,探讨局部微观环境因素对此类犯罪发生影响程度的差异性。街头抢劫犯罪是严重危害城市居民日常生活的一种恶性犯罪类型,由于犯罪实施者与犯罪受害者通常都是处在相对的运动过程,所以该类案件的发生更加依赖于犯罪场所也就是微观空间环境的选择。通过研究微观空间因素对案件发生影响程度的空间异质性将更有助于日常警务政策制定,尤其在警务日常巡逻线路的制定与警力的部署方面。2 研究方法与数据来源
2.1 研究区概况
H市位于中国东南沿海某省南部的特大城市,近年来经济发展水平稳步提高。城市常住人口不断增加。研究区DP半岛（图1）位于H市的中心城区,岛上除了几座与外部联通的桥梁之外仅有两条与周边相连的小路,四面环水。半岛上拥有新老各色居住建筑形式,更能反映岛上的居民在经济能力能和社会背景等因素的差异,是中国近年来快速城市化发展的一个缩影。半岛内集居住、商业、生产、交通多种城市功能于一身,用来分析空间环境对街头抢劫可以尽量剔除研究区以外的环境因素影响,保证研究区域的相对完整独立,可近似地将它视为一个相对独立完整的“缩微城市”。
显示原图|下载原图ZIP|生成PPT图1研究区DP半岛示意图
-->Fig. 1The study area-DP peninsula
-->
2.2 数据来源
街头抢劫案件数据和研究区基本现状数据分别来源于H市公安局2006-2011年110接警数据和P-GIS（警务地理信息系统）数据库。研究区DP半岛隶属于三个派出所管区（其中一个于2010年左右合并）,所以以此为滤镜进行筛选共得到496个案件点。经过自动匹配和人工校正过程后,剔除掉不在岛上的案件点,共得到有效定位案件点数373个。基本现状数据主要包括研究区的土地利用,建筑物的地理编码和交通路网信息等基础信息,并在此基础上通过百度地图对部分地物信息（如公交站点位置）进行修正。2.3 概念模型
2.3.1 微观空间因素对街头抢劫的影响 本文主要研究微观空间环境因素对街头抢劫犯罪发生影响的异质性,原因有以下两点：其一,人的日常活动总是无时无刻不处在相应的空间环境之内,空间环境往往能反映身处其中的人的一些社会属性信息,例如具有不同消费能力的人群总是固定的出现在某些消费场所中,不同年龄的人群总是会出现在特定的空间场所周边（如学校、酒吧、网吧等）;其二,空间环境所包含的微观空间因素能够反映该空间本身是否处于犯罪防范缺失和是否有利于犯罪后逃逸等信息,例如是否具备监控摄像装置,是否具有良好的通达性。从“日常活动理论”中得出,要成功的实施犯罪必须同时满足犯罪者、合适的目标与犯罪防范的缺失三个必要条件,而“理性选择理论”则侧重于从犯罪者的角度去考虑以最小的风险代价获取最大的犯罪收益。街头抢劫犯罪行为结合了上述两个理论的特点,其主要表现为犯罪者通过理性的抉择,在空间上考虑某些微观空间因素对于有经济价值的受害目标产生集聚影响进而锁定某些空间环境,并同时考虑在犯罪实施与逃逸过程中哪些微观空间因素抑制犯罪实施和利于逃逸,综合权衡下实施抢劫犯罪（图2）。由此可见,空间环境能影响犯罪目标和防范缺失以及作案后逃逸多个犯罪实施环节,会对犯罪行为的发生带来正效应或负效应的影响。而空间环境本身是由更加具体的微观空间环境因素组合形成的,不同的组合形式带来不同的具体空间形态。因此,关于某类微观空间因素对街头犯罪的影响的直接定论,无论是有无影响抑或是影响程度的大小,都是武断的。由于犯罪者本身基于“理性”总是在综合衡量各种微观空间因素影响的基础上决定是否实施犯罪的。因此,必须结合具体的空间环境来进一步分析微观环境因素对街头抢劫犯罪发生的影响方式和程度,也就是探讨微观空间因素对街头抢劫犯罪的发生影响是否存在空间上的异质性现象。
显示原图|下载原图ZIP|生成PPT图2街头抢劫犯罪者理性选择过程的示意图
-->Fig. 2The process of rational choice of offenders in street robberies
-->
2.3.2 微观空间因素对街头抢劫影响的空间异质性 传统的全局模型将影响犯罪发生的因素纳入到统一的模型中进行分析,因此得到的结果是该因素对犯罪发生的影响程度是唯一的定值,这样一刀切的做法往往会造成生态谬论的结果。例如许多犯罪分析结果总是发现社区犯罪发生率与社区人口密度之间存在正相关,即人口密度越高的社区会有较高的犯罪发生情况,但这样的结果并不一定是真正成立的。由于从全局整体的趋势考虑,导致局部区域呈现负相关的结果被忽视,得出的自变量系数结果会由于全局模型本身的特性而被均质化[15]。因此,整体层面对所有变量数据的全局分析处理必将掩盖局部层面一些真实相关性的情况,因此在进行微观分析的时候要尽量避免该现象。地理加权回归模型（GWR）扩展了传统回归方式的框架,以局部回归的方式改变了原有全局的参数估计,通过在线性回归模型中假定回归系数是观测点地理位置的空间函数,将自变量的空间特性纳入到模型的分析中,能有效的分析微观空间因素对街头抢劫犯罪影响是否存在空间异质性。
2.4 变量选取与处理
2.4.1 因变量的确定 为了确定案件发生与微观空间因素之间的关系,整个过程要求分析单元的空间尺度尽可能的小。而本文研究区位于H市中心城区,面积约6 km2,区域本身在行政单元上并不独立,难以以行政单元边界做出有效的研究单元的划分。若以城市街道为尺度进行划分得到的研究单元不仅数量少,而且不同街道的面积大小相差太大、空间形态也各异,因此仍然很难满足分析使用的要求。鉴于此,使用格网化研究区域的方法,依据Griffith等的网格化处理研究区域的计算公式[22],首先将整个研究区域用150 m×150 m的网格进行切割划分（图3）。经过初步统计,共得到233个有效的样本单元。然后,将2006-2011年的街头抢劫案件数据进行空间地址匹配后,其中有案件落入的网格共有103个,不到网格单元总数的二分之一,用实际案件点个数作为因变量无法有效运用于最小二乘法多元回归模型（OLS）与地理加权回归模型（GWR）的分析要求。因此,在因变量上面进行核密度处理,在通过对搜索半径以及输出栅格的一系列的敏感性分析后,最终确定以150 m为栅格图形输出值的边长（此网格设定可保证核密度输出单元与网格化研究单元统一）,500 m为搜索半径参数下的密度值结果作为分析研究的因变量。在此参数下的因变量值可以消除大量网格的0值情况,又可以尽量保留研究区原有的街头抢劫案件强度的空间格局。
显示原图|下载原图ZIP|生成PPT图3格网化研究区街头抢劫犯罪发生强度（密度）的空间分布示意图
-->Fig. 3The kernel density of street robberies in grid DP peninsula
-->
2.4.2 自变量的选取与处理 本文模型中的变量因子选取以“理性选择理论”和“日常活动理论”两个解释犯罪行为发生的重要理论为基础,并结合已有实证研究中关于影响犯罪行为活动发生的微观空间因素[10-15,23,24],共选取6个空间因子：每个格网中的土地利用混合程度（通过统计每个网格中POI设施的数量与种类的综合指标）;每个格网中路网的长度总合（包含城市主干道、城市次干道、城市支路三类综合）;每个格网与周围相邻的八个格网内（后文简称“八格网邻域”）所包含的公交站点总个数①(① 参考《城市道路交通规划设计规范（gb50220-95）》中公交站点车站服务半径300 m标准,选择八网格邻域较为适当。);每个格网重心点距离研究区内最近的派出所的距离;八格网邻域内所包含的交叉路口的总个数②(② 参考《城市道路交通规划设计规范（gb50220-95）》中城市主、次干道交叉口250~300m间距标准,选择八网格邻域较为适当。);每个格网重心点距离几个出岛重要路口中的最短取值。该6个自变量因子皆以假定犯罪者为理性人为前提,前3个自变量因子分别是考虑哪些空间环境能提供潜在的易受犯罪的目标和场所,后3个自变量因子更侧重犯罪者在犯案过程中对犯案风险的考虑和案件实施后如何有效实施逃逸的选择。通过相关性分析,6个自变量因素之间相关性均低于0.5,符合自变量相互间不共线的基本要求,因此在最初阶段可以同时放入回归模型中进行分析。
2.5 空间异质性存在的检验方法
将选择的微观空间因素变量首先进行OLS模型进行分析。采用Queen邻接方式创建空间关系矩阵,分别计算街头抢劫发生强度（密度）的全局自相关Moran's I指数和OLS模型结果的残差值的Moran's I指数并绘制散点图,探测因变量及残差的空间自相关性。如果检验出强烈的空间自相关现象,则说明影响街头抢劫犯罪发生的强度（密度）的空间分布存在较强的空间自相关性,则需要用GWR模型进行分析,并讨论该模型能否很好的展现出空间因素变量的空间异质性现象。2.5.1 OLS模型 OLS模型为全局线性回归模型,广泛运用于多个学科,在各区域都用全部自变量估计因变量的值,其模型公式为[25]：
式中：
2.5.2 GWR模型 在OLS的模型分析的基础上,如果计算结果中的
式中：
函数,GWR使用每个研究单元质心
参数
或
式中：d为位置i和j之间的距离;h为带宽,是决定权重计算方案的决定因子,可采用交叉验证最小化法,也可人为设定,或者使用Akaike信息准则（Akaike Information Criterion,简称AIC）最小化来确定。本文选取GWR 4.0中的AIC信息准则最小化来确定最佳带宽。
权重
3 结果分析
3.1 变量描述与统计分析
通过表1中对因变量和各个自变量的统计分析描述可以发现,2006-2011年DP半岛网格化划分之后的233个单元的抢劫犯罪密度值y基本符合正态分布的要求。在空间分布上（图3）,可以看出DP半岛上的街头抢劫案件在空间上的分布呈现出极强的集聚特性,部分区域街头抢劫案件的强度非常明显。相较于因变量,各自变量取值的分布则不存在很强的集聚特性,因此满足回归分析的基本要求。Tab. 1
表1
表1DP半岛个研究单元的变量描述性统计指标
Tab. 1Descriptive statistics of variables
| 变量 | 最小值 | 最大值 | 平均值 | 中位数 | 标准差 |
|---|---|---|---|---|---|
| Y（件/km2） | 0 | 361.21 | 69.50 | 45.65 | 75.58 |
| X1公交站点个数 | 0 | 5 | 1.49 | 1 | 1.13 |
| X2交叉口点个数 | 0 | 24 | 6.85 | 6 | 4.77 |
| X3土地利用混合程度 | 1 | 4 | 2 | 1 | 0.94 |
| X4单位网格路网长度（m） | 0 | 644 | 201 | 195 | 152.49 |
| X5最近派出所距离（km） | 0.03 | 1.29 | 0.57 | 0.56 | 0.25 |
| X6最近出岛口距离（km） | 0.05 | 0.96 | 0.48 | 0.50 | 0.26 |
新窗口打开
3.2 OLS回归鉴定结果
OLS估计结果显示（表2）,模型解释了233个研究单元的街头抢劫发生强度（密度）总变异的66.1%,其中八邻域网格交叉口个数变量、土地利用混合程度变量以及距离出岛口最近的距离3个自变量通过了0.01的自变量显著性检验,八网格邻域公交站个数自变量显著性略高于0.01,为0.011,也通过了0.05的显著性检验,表明这4个自变量在一定程度上能够解释对街头抢劫发生强度（密度）的影响。其中,八网格邻域公交站个数、八网格交叉口个数与土地利用混合程度3个变量的回归系数均为正值,表明对街头抢劫发生强度（密度）有显著的正效应,而出岛口最近距离变量的回归系数为负值,表明对街头抢劫发生强度（密度）有显著的负效应,也直接表明了是否能尽快逃离出相对封闭的犯罪区域对街头抢劫犯罪发生有着决定因素。相比之下,截距值和单位网格的路网长度以及距离最近派出所距离两个自变量的显著性均未能通过0.05的变量显著性检验,因此可初步判定OLS模型不是一个合格可用的模型。Tab. 2
表2
表2DP半岛街头抢劫发生强度的OLS全局估计结果
Tab. 2OLS regression model for density of street robberies on DP peninsular
| 回归系数 | 标准差 | t统计量 | P值 | 诊断指标 | 诊断指标值 | |
|---|---|---|---|---|---|---|
| y截距值 | -12.59 | 18.32 | -0.69 | 0.492 | R2 | 0.66 |
| X1 | 7.18 | 2.81 | 2.55 | 0.011* | R2调整 | 0.65 |
| X2 | 4.39 | 0.84 | 5.20 | 0.000* | AICc | 2438.96 |
| X3 | 57.57 | 4.02 | 14.32 | 0.000* | ||
| X4 | -0.008 | 0.03 | -0.311 | 0.76 | ||
| X5 | -16.60 | 15.86 | -1.05 | 0.27 | ||
| X6 | -72.67 | 17.701 | -4.11 | 0.000* |
新窗口打开
对于因变量街头抢劫犯罪强度（密度）的空间自相关分析表明,全局Moran's I指数（P=0.01,排列次数为999次）为0.6955,表明该变量本身存在极强的空间自相关现象。从Moran散点图中可以看出（图4）,街头抢劫犯罪的强度（密度）在局部空间上表现出显著的“同质相聚、异质隔离”特征。第一象限（HH）和第三象限（LL）的散点十分密集,占总数的85.4%,且基本都位于回归线的附近,表明街头抢劫犯罪强度（密度）高的网格单元和街头抢劫犯罪强度（密度）低的网格单元分别形成了显著的集聚区,DP半岛的街头抢劫犯罪发生强度（密度）主要以高—高聚集和低—低聚集分布为主,存在着很强的正相空间依赖性。第二象限（LH）和第四象限（HL）的散点则比较稀疏,表明街头抢劫犯罪发生强度（密度）高的网格单元和较低的网格单元彼此相邻的情况并不多,即表明空间的不稳定性较弱,非典型的网格单元个数并不多。由此可以检测出,DP半岛网格化划分后的233个研究单元的街头抢劫发生强度（密度）在地理分布上并不是随机的,而是呈现出强烈的空间集聚的特征,存在着明显的局域空间相关性,因此无法满足传统的回归模型的建模条件,OLS模型将由于忽略空间效应而可能存在模型的不适用问题。
显示原图|下载原图ZIP|生成PPT图4街头抢劫犯罪发生强度的Moran's I计算
-->Fig. 4Moran's I scatterplot for density of street robbery incidents
-->
通过对街头抢劫犯罪发生强度（密度）的OLS估计结果中的样本残差进行空间自相关分析得出,全局Moran's I指数为0.362（P=0.0000,排列次数为999）,表明区域残差存在显著的正向空间自相关。残差空间分布显示出明显的空间集聚状,这将导致OLS模型中的变量估计系数
3.3 GWR的空间异质性分析结果
DP半岛网格化划分后的233个研究单元的街头抢劫犯罪强度（密度）本身以及多元OLS回归模型计算所得残差均存在显著的空间自相关,表明街头抢劫犯罪强度（密度）无法满足OLS法要求的各个研究单元之间相互独立的假设,因此需要引入空间差异性来进行考虑局域的空间影响,以此来对其进行修正与改良。由于之前OLS回归模型证实了单位网格路网长度和距离最近派出所距离两个自变量的无效性,因此在接下来的GWR模拟分析中将不会继续引入这两个自变量。基于GWR模型能够反映街头抢劫犯罪强度（密度）总变异的17.2%~91.3%,观察局部R2观测值的空间变异,总体表现特征为对街头抢劫发生强度（密度）高的研究单元有较好的拟合度,可以很好地被模拟;相反,对于街头抢劫发生强度（密度）低的研究单元,尤其是半岛边缘的滩涂地区模拟优度较低。但总的来说,GWR模型有效地缩减了残差平方和,R2和调整R2的值也相比OLS模型有了显著的改善。根据Fotheringham等的评价标准[25],如果AICc的下降值超过或等于3就可以比较不同种类的模型拟合显著程度显著性,AICc越小则表明模型的拟合优度越好。本文中,GWR模型的AICc值相较之前的OLS模型的AICc值下降了95.24,AICc统计量的适当收敛说明GWR模型性能更好,大大增强了模型的拟合强度。针对GWR模型的残差值再次进行自相关检测,Moran's I指数为0.128（P=0.000012,排列次数为999）,数值本身通过0.01的显著性检验,相较于OLS结果残差的自相关分析结果大为降低,再次证明GWR局部模型比OLS全局模型性能优越。从GWR模型的回归系数（表3）和局部自变量系数估计图（图5）中可以得出,各个自变量在233个网格研究单元的参数估计结果均有所不同,对街头抢劫案件发生强度（密度）的影响程度也有很大区别,这一结果恰恰说明了影响因素在地理空间上存在不平稳性,尤其是在一些特定区域自变量因子表现出了正负两种不同的效应。
显示原图|下载原图ZIP|生成PPT图5公交站点个数自变量系数估计值的空间分布
-->Fig. 5Spatial variation of the coefficient estimation of the number of bus stops
-->
Tab. 3
表3
表3DP半岛街头抢劫强度的GWR估计结果
Tab. 3GWR estimates of density of street robbery incidents on DP peninsula
| 回归系数区间 | 诊断指标 | 诊断指标值 | |
|---|---|---|---|
| y截距 | -102.60~65.27 | R2 | 0.85 |
| X1公交站点个数 | -23.10~11.03 | R2调整 | 0.80 |
| X2交叉口点个数 | 2.23~18.17 | 局部R2 | 0.172~0.913 |
| X3土地利用混合程度 | 10.98~96.63 | AICc | 2343.72 |
| X4最近出岛口距离（km） | -202.26~25.38 |
新窗口打开
通过对于不同的自变量系数的估计值进行分级后看各自的空间分布特点。从图5中可以看出,公交站点自变量对于抢劫犯罪发生强度（密度）呈现出正反两种不同的效应,其中151个网格中表现出负相关的影响。通过与图3的对比可以发现,A区域和B区域都是研究区内街头抢劫发生强度（密度）较高的区域。两者所处的交通位置也很相似,都是进出半岛的重要通道区域,不同的是A处的通道是联系本地居民与老城区中心商业区地带的日常通道,而B区域的通道则是连接本地区与H市新城区的快速路,因此日常人流量A区域明显高于B区域处。因此虽在这两个区域内同样分布着较多的公交站点,但在A区域内公交车站点变量对街头抢劫犯罪强度（密度）有着很强的正效应,而B处却表现出相反的负效应影响结果。图6显示公交车站点个数变量对于抢劫犯罪发生强度（密度）的影响在岛内只有很少的区域是具有统计上的显著性意义的,但这些区域恰好是A、B两个区域,更进一步说明了上述判断的科学性意义。
显示原图|下载原图ZIP|生成PPT图6公交站点个数自变量系数显著性检验
-->Fig. 6Significance of the number of bus stops
-->
图7显示的交叉口数量自变量系数的空间分布情况。从系数的值可以看出,该变量对于街头抢劫犯罪发生强度（密度）的影响在整个半岛都呈现出正相关效应,但存在明显的空间分异表现。在半岛的中心区域A附近,影响强度呈现出逐渐的递减趋势。B区域的范围基本上包含了半岛上90%的老城区范围,老城区内的路网结构相比岛上的新城区部分明显复杂很多,尤其是城市支路较多,彼此连通性较好,较少出现断头路和死路,因此在B域的交叉口数量明显优于其余区域。虽然B区域的交叉口数量值普遍优于A区域,但在对街头抢劫犯罪发生的影响程度上A区域的交叉口正效应却远远高于B区域,主要原因在于A区域内有一个对外的长途车站,因此在此处实施抢劫行为后更加依赖交叉口来进行逃逸和躲避追捕。从图8显示可以看出,交叉口数量变量对于街头抢劫发生强度（密度）的影响在整个半岛上基本上都是具有统计的显著性意义的,只有在半岛的各别边缘区域不显著。
显示原图|下载原图ZIP|生成PPT图7交叉口个数自变量系数估计值的空间分布
-->Fig. 7Spatial variation of the coefficient estimation of the number of intersections
-->
显示原图|下载原图ZIP|生成PPT图8交叉口个数自变量系数显著性检验
-->Fig. 8Significance of the number of intersections
-->
土地利用混的混合程度自变量对于街头抢劫的发生强度（密度）的影响在整个半岛也是都呈现出正相关效应。图9中A区域的土地利用混合程度的等级数值本身并不高,但该区域的系数值显示出最高的正效应值,说明土地利用混合程度对这该区域的街头抢劫发生强度影响占主导因素。B区域所处位置虽土地利用混合程度相对较高,但本身通达性不高,位于环岛路上,离出岛口的距离较远,不利用犯罪后的逃逸,由于该区域街头抢劫发生强度（密度）本身并不大,因此土地利用混合度在此区域表现出负效应。区域C是半岛上土地利用和混合程度最高的区域,在这一区域街头抢劫的犯罪发生强度（密度）也呈现出与该变量高度的正效应。因此可以判定,一个区域的土地利用混合强度的复杂程度将直接影响街头抢劫犯罪发生的产生,土地利用混合程度越高,街头抢劫发生的强度（密度）越大。从图10中发现,除了极少数半岛的边缘区域外,其余部分土地利用混合强度变量对于街头抢劫犯罪发生强度（密度）的影响都是具有统计的显著性意义的。对比图9与图10,没有统计显著性的区域都是系数估计值最低的区域。
显示原图|下载原图ZIP|生成PPT图9土地利用混合程度自变量系数估计值的空间分布
-->Fig. 9Spatial variation of the coefficient estimation of degree of mixed land use
-->
显示原图|下载原图ZIP|生成PPT图10土地利用混合程度自变量系数显著性检验
-->Fig. 10Significance of degree of mixed land use
-->
图11中显示,对于最近出岛口距离的变量与街头抢劫犯罪发生强度（密度）的影响呈现出正负两种不同的影响。其中221个网格单元呈现出负向影响作用,仅有12个网格单元呈现出正向影响作用。从理性选择理论的角度出发,离出岛口的距离越近越有利于尽快逃离出整个半岛。图11中的A、B、C三个区域都是离开半岛的重要桥梁的出入口,但是却呈现出了不同方向的影响方式。在A、B区域呈现负相关影响,在C区域却呈现正相关影响。但通过图12的统计检验可以发现,在C区域最近出岛口距离变量的系数估计值是不具有统计显著意义的,所以可以确定该变量对于街头抢劫犯罪发生强度（密度）的影响呈现负相关,越靠近离开半岛的重要节点区域发生街头抢劫犯罪的可能性越大。
显示原图|下载原图ZIP|生成PPT图11最近出岛口距离自变量系数估计值的空间分布
-->Fig. 11Spatial variation of the coefficient estimation of the distance to the nearest exit
-->
显示原图|下载原图ZIP|生成PPT图12最近出岛口距离自变量系数显著性检验
-->Fig. 12Significance of the distance to the nearest exit
-->
通过对以上微观空间因素对街头抢劫案件的发生强度（密度）影响的空间异质性表现的分析可以发现,交叉口个数变量和土地利用混合程度变量对街头抢劫案件发生强度影响整体上是正向的,最近出岛口距离变量对犯罪发生强度（密度）的影响整体上是负向的,而公交站点个数变量的影响则呈现出正负效应同时存在。因此可以得到以下结论：在结合“理性选择理论”与“日常活动理论”所选出的微观空间因素变量与最初所设想的对街头抢劫犯罪发生强度（密度）的正向负向效应影响除公交站点个数变量外其余基本相符。公交站点个数变量在微观的空间尺度下,由于受其他因素的影响,对街头抢劫犯罪的发生并不一定都是正向效应的。
4 结论与讨论
应用一系列空间分析方法来计量检验和测算街头抢劫发生强度（密度）的微观空间因素,空间自相关Moran's I指数、散点图和GWR模型由于考虑了空间效应,在很大程度上克服了OLS线性回归模型全局设定同一系数偏差的问题。更值得令人深思的是,在GWR回归模型的基础上对每个空间影响因素估计的回归系数进行局域分解,从而将对研究区全局影响街头抢劫发生强度（密度）的空间因素分解成各个局部研究单元的影响,相比较全局OLS模型只能得出整个区域单一的因素贡献率而言给出了更加细致和准确的信息,更加适合知道微观区域的相应的犯罪防控重点。应用地理加权回归模型对DP半岛上233个有效的网格化研究单元街头抢劫发生强度（密度）及空间影响因素的模拟结果更加真实的反应微观区域的街头抢劫犯罪发生强度（密度）的实际情况。研究表明：（1）DP半岛上的街头抢劫犯罪发生强度（密度）的整体分布存在着明显的空间自相关性规律,说明了街头抢劫犯罪发生的强度（密度）存在空间关联及空间非平稳性。
（2）微观空间因素对街头抢劫犯罪发生强度（密度）的影响存在明显的空间异质性。OLS模型中结果中,公交站点个数变量、交叉路口个数变量以及土地利用混合程度变量对街头抢劫犯罪发生的强度（密度）的影响均为正效应,说明在日常警务防控与居民日常活动出行时应当注意相关的微观空间要素。出岛口最近距离变量始终呈现负效应,说明能尽快逃逸出相对封闭的DP半岛是犯罪者考虑的因素之一。在GWR模型中,交叉路口个数变量与土地利用混合程度变量仍表现出较强的正效应。出岛口最近距离变量表现为较强的负效应。而公交站个数变量表现出正负向近似相当的效应。说明由于局部微观空间环境组合的差异性,源于“理性选择理论”与“日常活动理论”选取的微观空间要素对街头抢劫犯罪发生强度（密度）影响程度存在显著差异,甚至在某些典型区域内的空间差异性影响对案件发生的存在正负两种相反的影响效应。
（3）由于微观空间因素对街头抢劫犯罪发生强度（密度）影响存在空间异质性,因此在进行相应的警务活动时则可以根据不同的空间因素影响程度的空间差异性而更加具体明确。例如某些空间位置的街头抢劫犯罪发生受公交站点影响程度不大,则可以考虑不必在此空间因素的周围分配过多的警力资源,而将警力资源有针对性的布置在公交车站点易于导致犯罪发生的位置,以此来优化警力资源的调配。
本文是基于H市DP半岛2006-2011年的街头抢劫案件数据进行分析的,由于在时间尺度上有一些跨度,所以在空间因素的提取上基本采取的是以2010年的空间环境状况为基础的折中办法,因此有一定的局限性。由于本文研究以微观尺度进行,所以在分析单元上采用了网格化处理的方式,造成分析部分难以有效的用行政单元名称进行描述,在因素估计系数分异现象规律分析描述方面尚有不足。由于研究单元未能与行政单元进行统一,所以难以有效的将社会、经济等相关的空间要素与微观空间因素有效的结合起来,希望在以后的研究中能不断加强相关要素的研究。
The authors have declared that no competing interests exist.
参考文献 原文顺序
文献年度倒序
文中引用次数倒序
被引期刊影响因子
| [1] | |
| [2] | . |
| [3] | . 20世纪60年代初,美国经济学家贝克尔(Gary S.Becker)提出了一个与当时权威观点大相径庭的理论:犯罪是一种理性行为.实践证明,贝克尔的理论是正确的,他的理论对美国的司法政策产生了相当大的影响.本文重点讨论以下三个问题:(1)在贝克尔的理性选择理论的基础上分析了我国当前经济犯罪的行为特征和产生的原因,并进一步对我国经济犯罪活动进行有限理性和非理性分析.(2)打击和预防经济犯罪的实质是司法部门与罪犯之间的博弈.本文第二部分利用博弈论的分析方法,以典型案例为对象,分析我国经济犯罪与打击预防经济犯罪双方的博弈的策略、支付、均衡、结果等.(3)在上述分析的基础上,依据理性选择理论和博弈论思想提出打击和预防经济犯罪的战略原则、制度设计和政策建议. 20世纪60年代初,美国经济学家贝克尔(Gary S.Becker)提出了一个与当时权威观点大相径庭的理论:犯罪是一种理性行为.实践证明,贝克尔的理论是正确的,他的理论对美国的司法政策产生了相当大的影响.本文重点讨论以下三个问题:(1)在贝克尔的理性选择理论的基础上分析了我国当前经济犯罪的行为特征和产生的原因,并进一步对我国经济犯罪活动进行有限理性和非理性分析.(2)打击和预防经济犯罪的实质是司法部门与罪犯之间的博弈.本文第二部分利用博弈论的分析方法,以典型案例为对象,分析我国经济犯罪与打击预防经济犯罪双方的博弈的策略、支付、均衡、结果等.(3)在上述分析的基础上,依据理性选择理论和博弈论思想提出打击和预防经济犯罪的战略原则、制度设计和政策建议. |
| [4] | |
| [5] | In this paper we present a "routine activity approach" for analyzing crime rate trends and cycles. Rather than emphasizing the characteristics of offenders, with this approach we concentrate upon the circumstances in which they carry out predatory criminal acts. Most criminal acts require convergence in space and time of likely offenders, suitable targets and the absence of capable guardians against crime. Human ecological theory facilitates an investigation into the way in which social structure produces this convergence, hence allowing illegal activities to feed upon the legal activities of everyday life. In particular, we hypothesize that the dispersion of activities away from households and families increases the opportunity for crime and thus generates higher crime rates. A variety of data is presented in support of the hypothesis, which helps explain crime rate trends in the United States 1947-1974 as a byproduct of changes in such variables as labor force participation and single-adult households. |
| [6] | ABSTRACT The explanation of crime has been preoccupied with individuals and communities as units of analysis. Recent work on offender decision making (Cornish and Clarke, 1986), situations (Clarke, 1983, 1992), envi- ronments (Brantingham and Brantingham 1981, 1993), routine activities (Cohen and Felson, 1979; Felson, 1994), and the spatial organization of drug dealing in the U.S. suggest a new unit of analysis: places. Crime is concen- trated heavily in a Jew "hot spots" of crime (Sherman et aL 1989). The concentration of crime among repeat places is more intensive than it is among repeat offenders (Spelman and Eck, 1989). The components of this concen- tration are analogous to the components of the criminal careers of persons: onset, desistance, continuance, specialization, and desistance. The theoret- ical explanationfor variance in these components is also stronger at the level of places than it is for individuals. These facts suggest a need for rethinking theories of crime, as well as a new approach to theorizing about crime for public policy. |
| [7] | Recent studies have shown that crime is concentrated at micro level units of geography defined as hot spots. Despite this growing evidence of the concentration of crime at place, studies to date have dealt primarily with adult crime or have failed to distinguish between adult and juvenile offenses. In this paper, we identify crime incidents in which a juvenile was arrested at street segments in Seattle, Washington, over a 14-year period, to assess the extent to which officially recorded juvenile crime is concentrated at hot spots. Using group-based trajectory analysis, we also assess the stability and variability of crime at street segments over the period of the study. Our findings suggest that officially recorded juvenile crime is strongly concentrated. Indeed, just 86 street segments in Seattle include one-third of crime incidents in which a juvenile was arrested during the study period. While we do observe variability over time in trajectories identified in the study, we also find that high rate juvenile crime street segments remain relatively stable across the 14 years examined. Finally, confirming the importance of routine activity theory in understanding the concentration of juvenile crime in hot spots, we find a strong connection between high rate trajectory groups and places likely to be a part of juvenile activity spaces. Though place-based crime prevention has not been a major focus of delinquency prevention, our work suggests that it may be an area with great promise. |
| [8] | The authors review research on police effectiveness in reducing crime, disorder, and fear in the context of a typology of innovation in police practices. That typology emphasizes two dimensions: one concerning the diversity of approaches, and the other, the level of focus. The authors find that little evidence supports the standard model of policing-low on both of these dimensions. In contrast, research evidence does support continued investment in police innovations that call for greater focus and tailoring of police efforts, combined with an expansion of the tool box of policing beyond simple law enforcement. The strongest evidence of police effectiveness in reducing crime and disorder is found in the case of geographically focused police practices, such as hot-spots policing. Community policing practices are found to reduce fear of crime, but the authors do not find consistent evidence that community policing (when it is implemented without models of problem-oriented policing) affects either crime or disorder. A developing body of evidence points to the effectiveness of problem-oriented policing in reducing crime, disorder, and fear. More generally, the authors find that many policing practices applied broadly throughout the United States either have not been the subject of systematic research or have been examined in the context of research designs that do not allow practitioners or policy makers to draw very strong conclusions. |
| [9] | . |
| [10] | The relationship between public transit and crime has been extensively investigated for decades. Compared to rail systems, limited research has been conducted on bus transit and its surrounding robbery influence. In particular, most of the studies have focused on either all types of robbery as a whole or street robbery only. Street and off-street robberies have distinct spatial characteristics, however: The former occurs on the street, whereas the latter occurs not on the street but at indoor places or in (semi)public areas. Therefore, they might have different proximity to bus transit. Furthermore, as bus transit is not operated twenty-four hours a day, it is questionable to connect bus transit with robbery when bus transit is not in service. This study uses exploratory spatial data analysis and statistical modeling to compare and contrast street and off-street robberies to discover how their relations with bus transit vary in space and time. RESULTS reveal that both street and off-street robberies decrease as the distance from bus stops increases. Compared to off-street robbery, street robbery is inclined to occur on the street without bus routes and during nonservice hours, respectively. These findings suggest that the preference of street robbery is related to the excess of suitable targets over capable guardians away from bus routes and when bus transit is not in operation, whereas off-street robbery is connected with the residential and commercial attractions along bus routes and people's frequent activities at these places during transit service hours. |
| [11] | Existing scholarship suggests that crime concentrates in close proximity to public bus stop locations. However, the importance of particular combinations of crime generators and attractors in the proximate environment around public bus stops has not been empirically documented. Drawing on previous environmental criminology research, the current study uses conjunctive analysis of case configurations to address questions about interpersonal violence around bus stops and other activity nodes in the proximate environment in Henderson, Nevada. Findings reveal that street robberies are highly clustered within a relatively small number of environmental contexts that are defined by specific combinations of activity nodes. They also show that bus stops are more likely than any other activity node to be found across dominant situational profiles of robbery, and that the relative risk of robbery associated with the presence/absence of bus stops varies widely on the basis of specific combinations of other activity nodes. |
| [12] | Crime prevention strategists have long been examining the impact of land use on crime, particularly the effects of non-residential land use in the form of facilities. Places like bars, motels, schools |
| [13] | Urban crime has increasingly become a major issue for Chinese cities. Using crime data collected at police precincts in 2008, the main aim of this research is to examine the spatial distribution of property crime which accounted for almost 82% of all crimes in the city of Changchun, and analyze the relationship between the spatial patterns of property crime and neighborhood characteristics. Standardized property crime rates (SCR) were applied to assess the relative risk of property crime across the city. Statistically significant clusters of high-risk areas or hot-spots were detected. A global ordinary least squares (OLS) regression model and a geographically weighted regression (GWR) model were calibrated to explore the risk of property crime as a function of contextual neighborhood characteristics. The analytical results show that significant local variations exist in the relationship between the risk of property crime and several neighborhood socioeconomic variables. |
| [14] | The aim of this paper is to address two critical but largely neglected issues in the spatial analysis of urban crime which are spatial spillover effects of crime penetrating neighborhood boundaries and non-stationarity regarding the relationships between contextual factors and neighborhood crime. We use a GIS-based spatial approach to normalize the estimate of burglary crime at block group level and use the geographically weighted regression (GWR) to investigate the correlates of neighborhood crime. Results suggest that the use of normalized measure of neighborhood crime helps better reveal the spatial patterns of burglary crime and the use of GWR accounts for the spatial variations of relationships between contextual factors and crime. In particular, the normalized measure of crime has implications for improving the measurement accuracy of the risk of crime across urban neighborhoods and can be applied to the spatial analysis of other socioeconomic issues such as housing foreclosures and environmental hazards which are also plagued by the spatial spillover issue when geographically contiguous data are analyzed. |
| [15] | |
| [16] | The geographical weighted regression(GWR) model is used to analyze the relationships between the spatial distributions of crime and geographical factors to find the cause of crime spatial distributions.The GWR model can capture the spatial variations in the spatially non-stationary problems.The GWR procedure is used to build a local model to analyze the effects of population density,road network density,and distance from police stations on the crime rate at the census tract level to give a set of mappable parameter estimates and t-values of significance that vary over space.Analytical results demonstrate that the spatial relationship between crime and geographical factors is a spatially non-stationary process and that the GWR model can help improve the accuracy of parameter estimation. |
| [17] | This article examines the connection of immigration and diversity to homicide by advancing a recently developed approach to modeling spatial dynamics eographically weighted regression (GWR). In contrast to traditional global averaging, we argue on substantive grounds that neighborhood characteristics vary in their effects across neighborhood space, a process of "spatial heterogeneity." Much like treatment-effect heterogeneity and distinct from spatial spillover, our analysis finds considerable evidence that neighborhood characteristics in Chicago vary significantly in predicting homicide, in some cases showing countervailing effects depending on spatial location. In general, however, immigrant concentration is either unrelated or inversely related to homicide, whereas language diversity is consistently linked to lower homicide. The results shed new light on the immigration-homicide nexus and suggest the pitfalls of global averaging models that hide the reality of a highly diversified and spatially stratified metropolis. |
| [18] | In this paper, we compare and contrast a Bayesian spatially varying coefficient process (SVCP) model with a geographically weighted regression (GWR) model for the estimation of the potentially spatially varying regression effects of alcohol outlets and illegal drug activity on violent crime in Houston, Texas. In addition, we focus on the inherent coefficient shrinkage properties of the Bayesian SVCP model as a way to address increased coefficient variance that follows from collinearity in GWR models. We outline the advantages of the Bayesian model in terms of reducing inflated coefficient variance, enhanced model flexibility, and more formal measuring of model uncertainty for prediction. We find spatially varying effects for alcohol outlets and drug violations, but the amount of variation depends on the type of model used. For the Bayesian model, this variation is controllable through the amount of prior influence placed on the variance of the coefficients. For example, the spatial pattern of coefficients is similar for the GWR and Bayesian models when a relatively large prior variance is used in the Bayesian model. |
| [19] | ABSTRACT The present research examines a structural model of violent crime in Portland, Oregon, exploring spatial patterns of both crime and its covariates. Using standard structural measures drawn from an opportunity framework, the study provides results from a global ordinary least squares model, assumed to fit for all locations within the study area. Geographically weighted regression (GWR) is then introduced as an alternative to such traditional approaches to modeling crime. The GWR procedure estimates a local model, producing a set of mappable parameter estimates and t-values of significance that vary over space. Several structural measures are found to have relationships with crime that vary significantly with location. Results indicate that a mixed model— with both spatially varying and fixed parameters—may provide the most accurate model of crime. The present study demonstrates the utility of GWR for exploring local processes that drive crime levels and examining misspecification of a global model of urban violence. |
| [20] | The main aim of this article is to analyze the relationships between the spatial patterns of residential burglaries and the socioeconomic characteristics of neighborhoods in London, Ontario. Relative risk ratios are applied as a measure of the intensity of residential burglary. The variation in the risks of burglary is modeled as a function of contextual neighborhood variables. Following a conventional (global) regression analysis, spatial variations in the relationships are examined using geographically weighted regression (GWR). The GWR results show that there are significant local variations in the relationships between the risk of residential burglary victimization and the average value of dwellings and percentage of the population in multifamily housing. The results are discussed in the context of four hypotheses, which may explain geographical variations in residential burglary. The practical implication of the GWR analysis is that different crime prevention policies should be implemented in different neighborhoods of the city. |
| [21] | Geographically weighted regression (GWR) is drawing attention as a statistical method to estimate regression models with spatially varying relationships between explanatory variables and a response variable. Local collinearity in weighted explanatory variables leads to GWR coefficient estimates that are correlated locally and across space, have inflated variances, and are at times counterintuitive and contradictory in sign to the global regression estimates. The presence of local collinearity in the absence of global collinearity necessitates the use of diagnostic tools in the local regression model building process to highlight areas in which the results are not reliable for statistical inference. The method of ridge regression can also be integrated into the GWR framework to constrain and stabilize regression coefficients and lower prediction error. This paper presents numerous diagnostic tools and ridge regression in GWR and demonstrates the utility of these techniques with an example using the Columbus crime dataset. |
| [22] | |
| [23] | Dealing with physical and social disorder to prevent serious crime has become a central strategy for policing. This study evaluates the effects of policing disorder, within a problem-oriented policing framework, at crime and disorder hot spots in Lowell, Massachusetts. Thirty-four hot spots were matched into 17 pairs, and one member of each pair was allocated to treatment conditions in a randomized block field experiment. The officers engaged "shallow" problem solving and implemented a strategy that more closely resembled a general policing disorder strategy rather than carefully designed problem-oriented policing responses. Nevertheless, the impact evaluation revealed significant reductions in crime and disorder calls for service, and systematic observations of social and physical disorder at the treatment places relative to the control places uncovered no evidence of significant crime displacement. A mediation analysis of the isolated and exhaustive causal mechanisms that comprised the strategy revealed that the strongest crime-prevention gains were generated by situational prevention strategies rather than by misdemeanor arrests or social service strategies. |
| [24] | The study reported here follows the suggestion by Caplan et al. (Justice Q, 2010 ) that risk terrain modeling (RTM) be developed by doing more work to elaborate, operationalize, and test variables that would provide added value to its application in police operations. Building on the ideas presented by Caplan et al., we address three important issues related to RTM that sets it apart from current approaches to spatial crime analysis. First, we address the selection criteria used in determining which risk layers to include in risk terrain models. Second, we compare the “best model” risk terrain derived from our analysis to the traditional hotspot density mapping technique by considering both the statistical power and overall usefulness of each approach. Third, we test for “risk clusters” in risk terrain maps to determine how they can be used to target police resources in a way that improves upon the current practice of using density maps of past crime in determining future locations of crime occurrence. This paper concludes with an in depth exploration of how one might develop strategies for incorporating risk terrains into police decision-making. RTM can be developed to the point where it may be more readily adopted by police crime analysts and enable police to be more effectively proactive and identify areas with the greatest probability of becoming locations for crime in the future. The targeting of police interventions that emerges would be based on a sound understanding of geographic attributes and qualities of space that connect to crime outcomes and would not be the result of identifying individuals from specific groups or characteristics of people as likely candidates for crime, a tactic that has led police agencies to be accused of profiling. In addition, place-based interventions may offer a more efficient method of impacting crime than efforts focused on individuals. |
| [25] | No abstract is available for this item. |
