A new Slop University research poster compares the reason code a site manager ticks at an exit interview with the code the same leaver selects six months later, and finds the two agree on fewer than half of 604 matched departures.
604 matched pairs put agreement at 46.9 per cent, with the disagreement running toward the codes that name nobody
Fewer than half the reasons recorded on an employer’s exit forms survive contact with the person who left.
Recorded as Personal, deposited today through the Office of Research Outputs, follows 1,214 departures across 38 sites of one multi-site employer. Each exit interview was conducted by the departing employee’s own site manager, who ticked one of eleven reason codes. Six months on, the University wrote to every leaver and offered them the same eleven. On the 604 pairs that came back, the two selections agree 46.9 per cent of the time, and where they part they part in one direction: reasons that name the site are recorded on the form as personal ones roughly twenty times more often than the reverse.
We would not want this read as a finding about managers. The instrument asks one person to write a sentence about themselves, in a room with the person leaving, and then reports that sentence upward as though it came from somewhere else. That is a design question, and the School is glad to have it in front of it.
— Associate Professor Kwame Lindqvist, Director of the Master of Priority Studies, School of Emergent Priorities
Three quarters of the forms also carried free text. Read blind by two coders, those notes line up with what the leavers said six months later rather than with the code beside them. Dr Iben Chikere, a Lecturer in the School, said the material had been “not missing from the system at any point, only from the report”.
The University counts the work among the plainer contributions it is pleased to have on the record, and part of a longer interest in what an institution learns about itself, and from whom. The full poster is available from the University’s research repository under an open licence, doi:10.5555/slop.ih1b2w.