原住民族障礙統計的資料治理困境：從資料落差到制度性不可見性

本研究著眼於臺灣官方統計系統中原住民族障礙盛行率不一致現象，關注其對人口健康監測與政策制定的影響，採檔案分析法與次級資料分析，探討制度性統計落差的生成機制與政策影響。運用衛生福利部統計處、社會及家庭署「身心障礙人口檔」，以及原住民族委員會《原住民族人口及健康統計年報》三套官方資料，進行跨資料比對與族群分類的落差檢視。研究發現如下：一、原住民族障礙盛行率在不同資料來源中呈現顯著落差，資料呈現方向不一致。二、「一人多筆」紀錄對統計結果產生偏誤，宜採歸人處理予以修正。三、來自同一資料庫的原始資料，各主管機關因資料串接與處理方式不同，導致統計結果存有差距。四、現行資料缺乏完整族群欄位與可再利用格式，限制原住民族障礙研究之發展。原住民族障礙統計資料的不可見性，反映出資料治理制度中族群敏感度之不足，宜建立統一的跨部會資料整合機制，制定具文化敏感度的資料分類架構，以落實族群正義與政策回應性。

Keywords

原住民族；身心障礙；資料治理；暗數據

Parallel abstracts

This study examines the inconsistencies in official statistics regarding the prevalence of disability among Indigenous populations in Taiwan, focusing on their implications for population health surveillance and policy formulation. Using document analysis and secondary data analysis, the research explores the institutional mechanisms behind statistical discrepancies and their policy relevance. Three official data sources are compared: (1) statistical reports from the Ministry of Health and Welfare, (2) the Disability Registry maintained by the Social and Family Affairs Administration, and (3) the Population and Health Reports published by the Council of Indigenous Peoples. The major findings are as follows: (1) significant discrepancies in Indigenous disability prevalence across sources, often with contradictory trends; (2) duplicated individual records that distort prevalence estimates unless corrected through proper deduplication; (3) divergent statistical outcomes resulting from differences in data linkage and processing practices across agencies, even when drawing from the same source data; and (4) the absence of standardized ethnicity fields and consistent formats in existing databases, which limits the development of Indigenous disability research. The statistical invisibility of Indigenous persons with disabilities reflects a lack of ethnic sensitivity within Taiwan's data governance framework. The study calls for the establishment of an integrated, inter-agency data coordination mechanism and the development of culturally responsive classification systems to support data justice and responsive policymaking.