Algorithmic Platform Management and Risk-Taking Behavior among Chinese Food Delivery Riders

1. Research Overview This dataset explores how algorithmic platform management shapes food delivery riders’ risk-taking behavior (e.g., traffic violations). Based on Cognitive Appraisal (CAT) and Persistent Cognition Theories (PCT), we proposed a dual-path model: algorithmic management influences behavior via an "online" path (Perceived Algorithmic Control during work) and an "offline" path (post-work Work Rumination), with Self-Control Resource Depletion (risk factor) and Learning Agility (protective factor) as boundary conditions. 2. Key Findings Data supports the dual-path model, revealing a double-edged sword effect: Algorithmic management increases risk-taking via Perceived Algorithmic Control and Emotional Rumination, but reduces it via Problem-Solving Contemplation. Riders’ risk-taking is "silent compliance," driven by both active reward pursuit and passive penalty avoidance. Self-Control Resource Depletion amplifies Perceived Algorithmic Control’s negative impact; Learning Agility mitigates the effects of Perceived Algorithmic Control and Emotional Rumination. 3. Data Interpretation Algorithmic management’s influence extends beyond work hours via offline cognition. Behavioral outcomes are co-determined by algorithmic pressure, individual cognitive resources, and adaptive capabilities. Enhancing rider safety requires algorithm optimization (e.g., flexible time buffers) and programs to boost Learning Agility. 4. Data Collection Source: Two-wave matched questionnaire survey of food delivery riders (Meituan, Ele.me, etc.) in Changsha, China. Timeframe: T1 (Apr-May 2024), T2 (3 weeks post-T1). Sample: 320 valid matched responses (full-time/part-time riders). Design: Multi-wave offline collection (via station managers, field visits) to reduce common method bias. 5. Data Usage Variables: Constructs (APM, PAC, ER, PSC, RTB, etc.) measured with adapted 6-point Likert scales. Application: Ideal for gig economy, algorithmic management, occupational safety, and work psychology research (verification, secondary analysis, methodological reference).1. 研究概述 本数据集聚焦算法平台管理如何影响外卖骑手的冒险行为（如交通违法）。基于认知评价理论（Cognitive Appraisal Theory, CAT）与持续性认知理论（Persistent Cognition Theory, PCT），本研究提出双路径模型：算法管理通过“线上路径”（工作期间感知到的算法控制）与“线下路径”（工作后反刍）影响骑手行为，其中自我控制资源损耗（风险因子）与学习敏捷性（保护因子）作为边界条件。 2. 核心发现 数据验证了双路径模型，并揭示出双刃剑效应： 算法管理通过感知算法控制与情绪反刍提升骑手的冒险行为，却通过问题解决沉思降低其冒险行为。 骑手的冒险行为属于“隐性顺从”，同时受主动追求奖励与被动规避惩罚的双重驱动。 自我控制资源损耗会强化感知算法控制的负面影响；而学习敏捷性则可缓解感知算法控制与情绪反刍的不良影响。 3. 数据解读 算法管理的影响可通过离线认知延伸至工作时长之外。行为结果由算法压力、个体认知资源与适应能力共同决定。提升骑手安全水平，需要优化算法设计（例如设置弹性时间缓冲）并开展提升学习敏捷性的相关项目。 4. 数据采集 数据来源：针对中国长沙地区美团、饿了么等平台的外卖骑手开展的两波配对问卷调查。 调研周期：T1（2024年4-5月），T2（T1结束后3周）。 有效样本：320份配对有效问卷（涵盖全职与兼职骑手）。 调研设计：通过站点管理人员、实地走访开展多波线下数据采集，以降低共同方法偏差。 5. 数据使用 变量设置：所有构念（算法平台管理（Algorithmic Platform Management, APM）、感知算法控制（Perceived Algorithmic Control, PAC）、情绪反刍（Emotional Rumination, ER）、问题解决沉思（Problem-Solving Contemplation, PSC）、冒险行为（Risk-Taking Behavior, RTB）等）均采用经过改编的6点李克特量表进行测量。 应用场景：本数据集适用于零工经济、算法管理、职业安全与工作心理学领域的研究（可用于假设验证、二次分析与方法学参考）。