Replication Data for: A National Study of Dockless Transportation: Land Use and Demographic Correlates of Trip Hotspots and Mode Shift

This research builds a land use regression model to explain dockless scooter trip generations. We use publicly available scooter trip generation data for Louisville, KY and Minneapolis, MN and publicly available data on land use characteristics. The model shows that scooter trip generations are associated with higher employment densities, higher densities of entertainment land uses (bars and clubs), and in some specifications higher densities of eating establishments and university buildings. We establish that using the regression results to predict out of sample gives predictions that correspond well to observed scooter trip generations in Austin, TX. Because scooter trip data are not available for research in California, we use the Minneapolis model to predict scooter trips as a function of land use characteristics in California census tracts. The results yield a promising screening method that can highlight census tracts with land use characteristics that are potentially supportive of micro-mobility and non-automobile short-trip travel. We recommend that such a screening method can be a first step in more detailed analyses of planning programs or infrastructure that could support non-automobile short-trip travel.

本研究构建了土地使用回归模型（land use regression model）以阐释无桩滑板车（dockless scooter）的出行生成量（trip generations），采用了美国肯塔基州路易斯维尔市与明尼苏达州明尼阿波利斯市的公开滑板车出行生成量数据，以及土地使用特征的公开数据集。模型结果显示，滑板车出行生成量与更高的就业密度、娱乐用地（酒吧与夜总会）密度显著相关；在部分模型设定下，其还与餐饮场所及高校建筑的更高密度存在关联。我们验证发现，利用该回归结果开展样本外预测（out of sample），所得结果与德克萨斯州（TX）奥斯汀市观测到的滑板车出行生成量匹配度较高。鉴于加利福尼亚州暂无适用于相关研究的滑板车出行数据，我们借助明尼阿波利斯市的模型，基于土地使用特征对加利福尼亚州各人口普查分区（census tracts）的滑板车出行量进行预测。本研究所得结果提供了一种颇具应用前景的筛查方法，可识别出土地使用特征潜在适配微移动出行（micro-mobility）与非机动车短途出行的人口普查分区。我们建议，该筛查方法可作为开展更精细化分析的第一步，用于评估可支持非机动车短途出行的规划方案或基础设施建设。