A School of Emergent Priorities paper finds that crossing a shift-rostering platform's weekend-shift fairness-credit threshold is followed by a sharp fall in how often a worker's appraisal uses a flexibility-adjacent term, an effect three placebo cutoffs fail to reproduce.
A retail chain's weekend-shift fairness threshold predicts what a supervisor writes about a worker, not how available they actually are.
A weekly shift tally decides two things at one national retail-and-hospitality chain: whether a worker is offered another weekend shift, and, this study finds, how a supervisor describes that worker three months later. Dr Lindiwe Achterberg of the School of Emergent Priorities led the work with Associate Professor Casimir Beng and Dr Anneke Tolan, treating the rostering platform’s fixed weekend-shift threshold as a dividing line between otherwise near-identical workers.
Staff whose rolling tally sat just under the platform’s six-shift cutoff kept a steady rate of flexibility-adjacent language in their next appraisal; staff who crossed it saw that rate fall by more than a third within the same quarter, with no matching change in rostered hours, shift-swap requests, or logged incidents. Three placebo cutoffs, set where the platform enforces nothing, produced no comparable shift, which the team reads as ruling out a general trend in the underlying tallies rather than the threshold itself.
A threshold built to even out the weekend roster has ended up deciding what a supervisor writes down, and the worker never sees either number.
— Dr Lindiwe Achterberg, Lecturer, School of Emergent Priorities
“The chain’s own people-analytics team read this the opposite way to how we did,” said Associate Professor Beng. “Where we saw a signal worth removing from the appraisal, they saw one worth building in properly.”
That team has since proposed writing the rostering threshold directly into the appraisal platform’s suggested language, a change the School is now tracking as a case study in how quickly an informal pattern can become declared policy. The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.gj2xye.