A new Slop University research poster finds self-reported wellbeing scores rise once a workplace pulse-survey bot's escalation rule begins, while independently coded interview sentiment stays flat.
An interview study at three open-plan employers finds self-reported wellbeing scores climbing only after an automatic escalation rule begins
A number typed into a workplace chat bot each week is supposed to describe how someone is doing. A research poster published today by Slop University’s School of Continuous Improvement finds that once the number can trigger a manager’s phone call, it starts to describe something else.
The poster reports an interview study of thirty-four staff at three open-plan employers already using the same weekly wellbeing pulse-survey bot, led by Dr Dagny Okafor of the School’s Evaluation of Evaluation program. Participants’ typed scores were checked against blind thematic coding of their own concurrent interview answers across a sixteen-week window spanning each site’s introduction of an escalation rule: two consecutive weekly averages under three summon a scheduled check-in with a rostered manager.
People start answering the bot instead of the question it was supposed to be asking. Once you can see that happening from the inside, it’s hard to call it measurement anymore.
— Dr Dagny Okafor, Lecturer and Convenor, Evaluation of Evaluation
Self-reported scores rose after the rule began; the same participants’ independently coded interview sentiment did not move. Co-author Dr Joost Nwosu, who studies what happens to a performance target once it has been met, said the finding fit a familiar shape: “the number stops tracking the thing itself right around when someone starts checking it.”
The School frames the result as one more entry in its long-running interest in what a self-report instrument teaches the people filling it in. The University’s own Adaptive Metrics Lab has adopted the study’s escalation rule for its internal staff wellbeing log.
The full research poster is available from the University’s research repository under an open licence, doi:10.5555/slop.1qchl4.