Asymmetric effects of attribute performance on transit satisfaction among older adults: machine learning regression with dummy variables
摘要
As the global population ages, it is essential to discourage older people from relying heavily on cars and to encourage them to use low-carbon travel modes, such as transit. However, previous studies have paid limited attention to older adults’ satisfaction with transit services (transit satisfaction) and its determinants. More importantly, they typically assume that the effects of service attributes on transit satisfaction are linear and symmetric. This study analyzes questionnaire survey data from 928 older transit users in Chengdu, China. It employs random forest-based regression with dummy variables to examine the effects of service attributes on satisfaction with three types of transit services: metro, bus rapid transit (BRT), and conventional bus services. Additionally, this study incorporates the three-factor theory to identify the factor structure of service attributes, categorizing them into basic, performance, and exciting factors, and determines priorities for service improvement. The results indicate that certain attributes (e.g., seat availability) have consistent effects across all three transit modes, whereas others (e.g., comfort) vary significantly by mode. The key service attributes requiring priority improvement include seat availability and crowding (for metro, BRT, and buses); station accessibility and information at stops (for metro and BRT); and waiting time (for BRT). This study offers mode-specific service enhancement strategies and provides valuable theoretical and empirical insights for optimizing age-friendly transit services.