IoT-based analysis reveals behavioral differences in public participation in low-value recyclables collection

The global recycling rate for low-value recyclables (LVRs) remains low, with inadequate mechanisms and suboptimal participation rates. Emerging smart recycling systems, powered by IoT-enabled monitoring, now offer researchers unprecedented opportunities to analyze real-time behavioral patterns in household recycling practices. However, key knowledge gaps persist, particularly regarding demographic disparities in participation and the drivers influencing these behaviors. In order to explore policy and behavioral pathways to improve recycling rates, this study analyzes the behavior of 135 thousand residents in a typical Chinese city for LVRs. The analysis includes nearly 2 million pieces of data on recycling behavior and covers 15 months. By using big data statistical methods, multiple regression models (MLR), and regression discontinuity in time (RDiT), this study explores behavioral characteristics, external factors, and the impact of intervention events. Key findings include: (1) With 20% of participants contributing 72% of recyclables and a Gini coefficient of 0.71, residents’ recycling contributions demonstrate substantial inequality; (2) Disposal weights show significant correlations with weather, neighborhood quality, and equipment accessibility, with an inverted U-shaped relationship with temperature, peaking at 13 °C; (3) Equipment downtime induces persistent declines in new-user retention, whereas economic incentives drive only transient participation spikes among regular users, with effects diminishing to baseline within one month. Therefore, we propose a heterogeneous community incentive strategy for groups with different characteristics. Overall, this study provides a data-driven basis and useful insights for the precise design of recycling policies.