| 摘要: | 近年來，台灣各產業普遍面臨嚴重的勞工短缺問題，其中以營造業的缺工情況更為嚴重。為解決勞力短缺的問題，政府於2023年6月放寬外籍移工招聘法規，導致外籍移工人數大幅增加。然而，隨著外籍移工的大量湧入，發生重大職業災害事故的人數也可能因此大幅提高，引發社會各界對於外籍移工職場安全的高度重視和關切。本研究旨在透過系統分析重大職災數據，從中觀察外籍移工和本籍勞工在職業災害成因上的差異，並提出因應對策。研究資料共計1857筆重大職災個案，以營造業為主，其中包括58筆涉及外籍移工、1799筆涉及本籍勞工。首先將本籍勞工資料進行抽樣(317筆)，合併外籍移工的資料(58筆)進行卡方檢定，篩選類別。再採用敘述性統計針對災害類型、成因屬性等類別進行分析，初步顯示出兩類族群在職業災害成因上的差異。其次，本研究將運用正規化概念分析FCA (Formal Concept Analysis)技術，從複雜的職災個案中發掘隱藏的關聯規則，並透過概念點陣圖和可信度列表將規則視覺化，方便進一步解讀。再者，研究將進行潛在類別分析LCA (Latent Class Analysis)，根據職業災害的屬性組合模式，了解外籍移工和本籍勞工在工程中存在的隱含分布，得出的結果再進行卡方檢定的複檢和羅吉斯回歸的驗證來支撐分析的產出。後續提取FCA的關聯規則。最後透過FCA和LCA兩種分析的工具應用加上羅吉斯回歸的驗證。當涉及建築工程或營建物及施工設備以及存在潛在危險的作業環境時，可以採用LCA所產出的重大職災類別屬性機率表以及羅吉斯迴歸得出各類別屬性的勝算比方程式，這兩項結果對於安全評估具有重要貢獻。像辨識高風險區域，例如挖掘工地或重型機械施工區，都需格外注意並加強預防措施，避免重大職業災害發生。這些研究成果能作為未來制定勞工職災風險評估的重要依據，進而確保工人們的工作安全。;In recent years, various industries in Taiwan have been facing severe labor shortages, with the construction industry being particularly affected by a lack of workers. To address the labor shortage, the government relaxed regulations on the recruitment of foreign migrant workers in June 2023, leading to a significant increase in their numbers. However, with the influx of foreign migrant workers, the number of serious occupational accidents may also increase substantially, raising widespread social attention towards the workplace safety of foreign migrant workers.
This study aims to systematically analyze data on major occupational accidents to observe the differences in the causes of accidents between foreign migrant workers and local workers, and propose countermeasures. The research data includes a total of 1,857 cases, mainly in the construction industry, of which 58 cases involved foreign migrant workers and 1,799 cases involved local workers. First, the local worker data was sampled (317 cases) and combined with the foreign migrant worker data (58 cases) for chi-square testing to screen the categories. Descriptive statistics were used to analyze categories such as accident types and causal attributes, initially revealing differences in the causes of occupational accidents between the two groups.
Next, the FCA technique was employed to uncover hidden association rules from the complex cases, and the rules were visualized using concept lattices and credibility lists. The research will conduct LCA to understand the latent distribution patterns of foreign migrant workers and local workers in engineering projects based on the attribute combinations, with the results cross-checked using chi-square tests and validated with logistic regression. Subsequently, the association rules from FCA will be extracted.
Finally, through the application of both FCA and LCA analysis tools, along with logistic regression validation, when dealing with construction projects, construction sites, and potentially hazardous work environments, the probability table of occupational accident categories generated by LCA and the odds ratio equations for each attribute derived from logistic regression can be used, as these two results contribute significantly to safety assessment. High-risk areas such as excavation sites or heavy machinery operation areas require extra caution and enhanced preventive measures to avoid serious occupational accidents. These research findings can serve as an important basis for future formulation of occupational accident risk assessments, thereby ensuring the safety of workers in their workplaces. |
