Many studies have been conducted to investigate the diverse human-related factors that contribute to traffic crashes. Human factors have a greater impact on crashes caused by taxi drivers with long driving distances and hours. However, due to issues related to the protection of individual data and the complexity of collecting and processing data, there are limitations in clearly identifying risk factors related to driver characteristics. In this study, we combined in-depth survey data that included characteristics of taxi drivers and the companies they belong to and taxi crash data (2017-2019) in South Korea. However, the combined data showed a high correlation or causality between variables, leading to potential problems, i.e., multicollinearity, hierarchical structure of data, and inefficient analysis. To address this issue, we applied Principal Component Analysis (PCA) to reduce the dimensionality of variables and mitigate the problem. Furthermore, we constructed a hierarchical model considering the hierarchical structure of data in corporate taxis, where drivers are affiliated with specific companies. The analysis revealed that managing fatigue at the company level, managing drivers' diseases, and other intrinsic factors had a significant influence on Fatal-Injury (FI) crashes. These results indicate that taxi crashes are influenced significantly by both company management factors and driver-related factors. Therefore, policymakers can provide customized preventive measures that consider both aspects.