Can geographically weighted regressions improve regional analysis and policy making?
Can geographically weighted regressions improve regional analysis and policy making?
复制标题
DOI：
10.1177/0160017607301609
复制
发表时间：
2007-07-01
影响因子：
2.3
通讯作者：
Olfert, M. Rose
中科院分区：
文献类型：
--
作者：
Ali, Kamar;Partridge, Mark D.;Olfert, M. Rose
在区域范围内的政策设计需要明确认识到社区特征的空间异质性以及这些特征如何影响目标变量的异质性。通过仅提供整个空间的“全局”度量，标准方法（如普通最小二乘法或（大多数）空间计量经济学模型）倾向于折衷空间异质性，以利于平均估计和效率。需要更多地评估简化和统计效率的收益是否抵消了忽视空间异质性的损失。作者使用大约1,900个加拿大农村社区的数据作为背景，使用地理加权回归方法解决这个问题。作者发现，对于大约两个变量，标准方法会大大低估选定变量影响的空间差异。标准的分析不会发现这些信息，这表明随后的政策推论不适合许多地方环境。
Policy design in a regional context requires explicit recognition of spatial heterogeneity in community characteristics as well as in the heterogeneity of how these characteristics impact the target variables. By providing only a "global" measure for the entire space, standard approaches such as ordinary least squares or (most) spatial econometric models tend to compromise spatial heterogeneity in favor of average estimates and efficiency. More assessment is needed of whether the gains of simplicity and statistical efficiency offset the losses from ignoring spatial heterogeneity. Using data for about 1,900 rural Canadian communities as a backdrop, the authors address this issue using a geographically weighted regression approach. The authors find that for about two-thuds of the variables, standard approaches would have significantly understated the spatial differences in the impact of selected variables. Standard analysis would not have uncovered this information, suggesting that subsequent policy inferences would be poorly suited to many local settings.