Evaluating Shuttle Feeder Service Preferences for MRT Lebak Bulus Station Using Scenario-Specific Binary Logit and Generalized Cost Analysis

Abstract

First-mile access remains a critical challenge in connecting residential areas with urban rail stations. This study evaluates a proposed shuttle feeder service for MRT Lebak Bulus Station, Jakarta, using 249 valid stated-preference survey responses. Respondents evaluated Shuttle A (30 minutes, Rp10,000), Shuttle B (25 minutes, Rp15,000), Shuttle C (20 minutes, Rp20,000), and their existing access mode. The analysis integrated observed stated-preference choices, access-mode segmentation, data diagnostics, chi-square association tests, scenario-specific binary logit models, generalized cost sensitivity, willingness to pay, an affordability proxy based on existing access expenditure, and residential corridor synthesis. The findings show that 89.6% of respondents selected a shuttle alternative, with Shuttle B being the most preferred scenario (43.8%). Existing access mode was significantly associated with shuttle choice (p = 0.034), while data diagnostics indicated no serious multicollinearity between time and cost differences. Among the scenario-specific logit models, Shuttle C versus existing mode produced the strongest result (Omnibus p = 0.010; Nagelkerke R² = 0.152), with cost difference being statistically significant (p = 0.007). The significant Shuttle C model should be interpreted as evidence of a cost-sensitive segment rather than as the sole basis for selecting the initial service package. Nevertheless, Shuttle B remains the most realistic initial operational package because it combines the strongest stated preference, a middle fare, and consistency with dominant willingness-to-pay and affordability ranges. Cluster-based synthesis further indicates that the southwest corridor, particularly the Ciputat-Rempoa-Legoso cluster, should be prioritized for initial implementation.