運用空間統計方法探討法律扶助申請案件之區域樣態與影響因子

Using Spatial Statistics to Explore the Distribution and Influential Factors of Legal Aid Cases

方啟元(Chi-Yuan Fang) ； 林易霆(Yi-Ting Lin) ； 林鼎智(Ding-Chih Lin) ； 鄧雅云(Ya-Yun Teng) ； 黃怡婷(Yi-Ting Hwang)

《中國統計學報》 59卷4期 (2021/12) Pp. 226-254

摘要

司法差距（Justice Gap）是各國政府長期困擾的社會議題之一，國內針對此議題成立財團法人法律扶助基金會。該組織礙於不諳統計方法，難以量化服務需求、找出潛在之重點服務區，本研究期望以視覺化初探法律扶助案件之分布，利用空間統計方法，建構貝氏階層線性模型探討法律扶助案件數於北北基各經濟二級發布區之分布樣態，以推論影響案件數的因子及其效果，並標示出法律扶助申請案件之離群區。利用Moran's I指標，本研究發現案件數分布具空間自我相關性，再由條件自我相關模型設定研究區域之空間結構，將隨機效果項納入廣義空間線性混合模型，比較Besag-York-Mollie模型、Leroux模型及Locally Adaptive模型等3種條件自我相關模型之空間結構設定，並以集成嵌套拉普拉斯近似（INLA）方法進行變數效果估計。其中以Besag-York-Mollie模型為最適解釋法律扶助案件申請區域樣態，供法扶基金會作為後續擬策參考。

關鍵字

法律扶助 ； 區域資料分析 ； 階層線性模型 ； 貝氏推論 ； 條件自我相關模型 ； 集成嵌套拉普拉斯近似

並列摘要

＂Justice Gap＂ has long been a troubling matter for countless nations, The Legal Aid Foundation of Taiwan is the designated countermeasure organization for the subject. Due to the organization's unfamiliarity with statistical practice, quantifying sectional demand and spotting sectors with exigent needs remains an arduous task. This paper scrutinizes the visualization of legal aid cases in the Taipei Metropolitan Area and constructs a Bayesian hierarchical linear model with spatial statistical methods to estimate the distribution of legal aid cases within the research area. Through calculating the Moran's I test, the existence of spatial autocorrelation in the geographical distribution of legal aid cases is found. The spatial structure of our research area is assumed under the conditional autoregressive (CAR) model, which is regarded as the random effect component of a generalized linear mixed model (GLMM). Three distinct CAR models including the Besag-York-Mollie model, Leroux model, and the Locally Adaptive model are considered. Using the integrated nested Laplace approximation( INLA) method to estimate variable effects, the best-suiting spatial structure design in this study is the Besag-York-Mollie model. We highlight regions with ineffective policy implementation and provide inference on variables that significantly affect the allocation of service demand. Final results serve as a reference to future policy designing for the Legal Aid Foundation of Taiwan.

並列關鍵字

Legal aid ； Areal data analysis ； Hierarchical linear model ； Bayesian inference ； CAR ； INLA

參考文獻

Banerjee, S., Carlin, B. P., and Gelfand, A. E. (2015). Hierarchical modeling and analysis for spatial data. CRC press.

Besag, J., York, J., and Mollié, A. (1991). Bayesian image restoration, with two applications in spatial statistics. Annals of the institute of statistical mathematics, 43(1), pages 1-20.

Bivand, R. S., Pebesma, E. J., Gómez-Rubio, V., and Pebesma, E. J. (2008).Applied spatial data analysis with R. Springer.

Cressie, N. (2015). Statistics for spatial data. John Wiley & Sons.

Gelman, A., Hwang, J., and Vehtari, A. (2014). Understanding predictive information criteria for Bayesian models. Statistics and computing, 24(6), pages 997-1016.