Implicit bias in dressage judging?

            When I was watching the Prix St. Georges class from the Wellington Nations’ Cup last weekend, I was reminded of the familiar claim that it’s better to ride at the end of the class because that is when the scores are higher. I can imagine that this is especially the case when the class is as long as it was for that Prix St. Georges: it was a lot to watch, let alone to judge! But I was suspicious that some of the riders received scores that were a percentage point or two higher than they would have been had they ridden at the beginning of the class. What was my reason for the suspicion? Early on, I noticed that horses that were not terribly expressive, but supple and accurate, were scoring higher than ones that were big movers, but showing tension. It seemed that the tension was less penalized towards the end of the class.

            This is not to say that the judges are blameworthy. I would say that the tendency to score horses higher at the end of the class is an example of unconscious  (or implicit) bias. We have an unconscious bias when we, without being aware of it, have attitudes or behave in ways that favour one group over another. One place that unconscious bias is well documented is in hiring: when women and men are given equivalent résumés, men are still more likely to be invited to interviews. Most of the time, this phenomenon is not the result of explicit sexism, but of certain expectations we unconsciously have about the performance of men versus women, which lead the résumé reviewers to read the men’s more positively.

In cases where the strongest riders are in fact at the end of the class, the bias is useful, since the judges can more easily analyze the movements and give them a score: they have an idea of the (higher) range in which the score will fall, and adjust up or down based on fine details. I think there are two problem cases. The first is when a early rider has a fantastic test, but the judges are operating in lower score range, and don’t give the test high enough marks. In such instances, it is a little more difficult for inconsistent or improving riders to rise in the placings when it is deserved. In the second, a weak ride goes late, but gets scored up because it is among the tests expected to be better and score higher. Here, the best riders end up placing well, even on a not-so-good day. It would be better, I think, if the placings were more flexible, and I think the placings would be more flexible if the unconscious biases could be reduced.

            Judges are well-meaning and want to judge dressage riders fairly and consistently. But intending to be fair is usually not enough to prevent the effects of unconscious biases. In this case, judges must be expecting that earlier riders will be weaker and later riders will be stronger. It is an expectation that makes sense, since classes, especially in championships, are often ordered such that the riders who have previously scored well are placed at the end of the class. For the spectator, this is a great way to arrange a class: it’s exciting to watch the close competition for the win right at the end; but I think this is a likely what has led to this judging bias.