A new Slop University research poster finds that a junior sports association's most-improved award selects the players with the shortest first half rather than the largest gain, and that attendance at the mid-season clinic the award is cited to support makes no detectable difference.
A research poster recomputes ten seasons of a junior sports association's award rule and recovers the same winners from records with the games shuffled.
A trophy for the most improved player is meant to reward a season’s work. The Trophy Goes to the Smallest Sample, released today by the School of Emergent Priorities, finds that the rule a junior sports association uses to award one selects instead on the shortest record in the squad.
Dr Ingrid Vasseur, Dr Lindiwe Achterberg and Dr Sten Okwuosa recomputed the association’s published rule against ten seasons of per-game returns from 24 clubs and 1,914 registered players. Recipients’ first halves sat a median of 2.3 points below their own squad’s median and rested on a median of four recorded games. Their second halves came back to within half a point of that median in nine of the ten seasons.
To test what the rule was reading, the team shuffled the order of each player’s own games ten thousand times over and ran it again. It named the same recipient in 64 per cent of draws.
An award becomes evidence surprisingly quickly. By the time a name reaches an annual report it has stopped being something read out in a hall and started being a result.
— Dr Sten Okwuosa, Senior Research Fellow and Convenor of the Impact Pathway Atlas, School of Continuous Improvement
The association’s annual coaching report cites recent recipients as evidence that its mid-season skills clinic is working. Attendance at the clinic made no detectable difference to the figure the award is decided on.
“The presentation night was never the part we were looking at,” said Dr Ingrid Vasseur, Lecturer and Convenor of Futures in Committee. “It is the calculation that comes before it, which nobody attends.”
The University offers the work as an example of the questions it prefers to ask: not whether an award is deserved, but what the arithmetic behind it is able to see. The poster closes by proposing a minimum-games floor and a shrinkage-adjusted difference, to be trialled across a single award cycle.
The full poster is available from the University’s research repository under an open licence, doi:10.5555/slop.yhp74e.