Recently, Dr Liam Lenten was featured on ABC’s Radio 666 Canberra, talking about his research on the voting process in the AFL’s Brownlow Medal (the ‘Best and Fairest’ award). In his research, Dr Liam Lenten analysed how umpires vote and whether they have any biases.
During the segment, Dr Lenten says he became motivated to test for how Indigenous players draw votes from umpires early on in the project – a time at which there were a number of unsavoury incidents involving adverse fan behaviour (particularly) towards now-retired Sydney star Adam Goodes.
‘As economists, we are interested in voting biases whenever and wherever they occur, as they violate the notion of pure rationality. There are many examples in the literature involving data from various competitive environments, including many from sports, but also political elections, and even song contests such as Eurovision or Idol’, Dr Lenten says.
In this case, he employs econometric modelling and regression techniques, as well as a comprehensive database of over 100,000 player-within-match observations, in all of the 2,288 home-and-away AFL matches from 1998-2010.
Lenten says ‘…when I compared the number of votes between groups, Indigenous players quantitatively poll 6.6% more average votes per match than non-Indigenous players – this difference was statistically significant, but only weakly so’.
He qualifies this by saying that this alone does not imply a voting bias: “The comparison does not account for various factors, which could be driving the result. One possible explanation is that the pool of Indigenous players in the AFL is simply drawn from a higher-distribution of talent than non-Indigenous players. Aspiring Indigenous footballers are more likely to be from regional and remote areas, consistent with being at an elite-level recruitment disadvantage.”
According to Lenten, if this explanation is true, then such differences in talent levels should be picked up by using player performance measures. Specifically, he uses a range of 13 well-known player-level match statistics; such as kicks, handballs, goals scored, and tackles.
The model results show that controlling for player performance explains most (but not quite all) of the difference in average votes awarded. Indigenous players still poll approximately 1.2% more votes than non-indigenous players.
Is this remaining difference evidence that umpires unfairly give more votes to Indigenous players? As a former 200-game umpire himself (at suburban level), Dr Lenten is open to this possibility, but not committal.
‘There could be other productive actions off the ball for which Indigenous players outperform their contemporaries that are not captured within the statistics I use, yet are nonetheless noticed by umpires’. Also, he says: ‘…with respect to goals, it may be that Indigenous players often kick goals with higher degrees of difficulty, which would imply that not all goals have the same vote impact’.
Listen to the full fragment on the ABC website, here (Dr Liam Lenten speaks around the 3:26:00 mark).