Autonomous driving technology is transforming transportation systems, with automated mobility-on-demand (AMoD) services emerging as a promising mobility solution. This study investigates how spatial characteristics of residential areas and latent attitudes influence preferences for AMoD services. Utilizing a stated choice experiment with 2,258 respondents (9,032 observations), the study estimated a hybrid choice model incorporating road area ratio, degree centrality, and closeness centrality as spatial indicators. Results demonstrate that spatial centrality is significantly associated with AMoD service preferences and latent attitudes. Road area ratio is positively related to preferences for buses and AMoD services, while degree and closeness centrality exhibit mode-specific associations with taxi and bus preferences, respectively. Spatial characteristics are systematically related to latent attitudes, suggesting that AMoD services are perceived as closer to public transportation than to private cars. These findings provide insights for effective operational strategies for AMoD services, particularly in regions with high public transport orientation.