Spatial Analysis of Indonesia’s 2019 Provincial Voter Turnout Using Geographically Weighted Regression

Ernawati Pasaribu (Department of Statistics, Politeknik Statistika STIS, Jakarta 13330, Indonesia)
Ega Eugenia Naomi (Eureka Research Consultant, South Jakarta 12610, Indonesia)
Arya Candra Kusuma (Directorate of Crops, Horticulture, and Plantation Statistics, BPS-Statistics Indonesia, Jakarta, Indonesia)
Firman Emmanuel Declarantius Parulian (BPS-Statistics of Kotawaringin Barat Regency, Kotawaringin Barat 74112, Indonesia)

Zero : Jurnal Sains, Matematika, dan Terapan
Vol 10, No 1 (2026): Zero: Jurnal Sains Matematika dan Terapan

This study analyzes spatial heterogeneity in voter behavior using the Geographically Weighted Regression (GWR) model, estimated via GWR 4.0 software to ensure reproducibility. Using 2019 provincial-level data, we compare an OLS model against GWR to evaluate the mean voter turnout across Indonesia’s concurrent elections. The GWR model significantly improved estimation performance, evidenced by a reduction in AICc from -140.960 to -146.581 and an increase in Adjusted R² from 36.48% globally to a local range of 65.30%–82.60%. Statistical testing via F-statistic yielded a p-value of 0.0633, significant at the 10% level acceptable given the sample size (n=34). Findings reveal significant spatial non-stationarity: the Gini ratio shows a pervasive positive mobilization effect, while education and infrastructure display region-specific impacts, with infrastructure accessibility strongly influencing turnout in eastern Indonesia. These results underscore the mathematical necessity of incorporating geographic variance into predictive electoral models to capture localized socio-political dynamics.