This study conceptualises algorithmic punishment as a multidimensional construct and develops a validated scale capturing its three core features on food-delivery platforms. It addresses a critical theoretical gap in platform governance research by providing a granular instrument to assess how couriers perceive and respond to platform-imposed algorithmic punishment.
Drawing on algorithmic management research and qualitative interviews, we identify three dimensions of algorithmic punishment: systematicness, reasonableness and understandability. A multi-stage procedure uses three samples for item generation, exploratory and confirmatory factor analyses, and validation within a nomological network.
Results support a stable three-factor structure with satisfactory reliability, convergent validity, and discriminant validity, and further support modelling algorithmic punishment as a higher-order construct in a hierarchical confirmatory factor analyses. In the nomological validation model, higher-order algorithmic punishment is positively associated with standardised empowerment and negatively associated with standardised pressure. Empowerment is positively related to service quality commitment, whereas pressure is positively related to workaround behaviours, providing criterion-related validity evidence for the scale.
Data were collected from Chinese food-delivery platforms, indicating the need for further validation in other contexts. The cross-sectional design limits causal inference, highlighting the value of longitudinal or experimental research.
The validated scale provides platform managers with a diagnostic tool for assessing couriers’ perceptions of algorithmic punishment. Enhancing the systematicness, reasonableness and understandability of algorithmic punishment can help balance control and support, reduce conflict and promote sustainable gig work.
Design and communication of algorithmic punishments that strengthen perceived fairness, lower courier stress and ease platform–courier tensions can contribute to more equitable and sustainable forms of digital labour.
This study is among the first to conceptualise and operationalise algorithmic punishment as a multidimensional construct. It offers a context-sensitive scale that advances theory on algorithmic management and supports future empirical research on gig work.
