Sports Broadcasts Need a Statistics Ombudsman
September 12, 2011 Leave a comment
In the first set of today’s U.S. Open final the CBS crew stressed the importance of winning the opening set by stating that 19 of the last 21 first set winners have gone on to win the final. Like most sports statistics (which are meant to appear eye-catching rather than impart accurate information), the stat demonstrates correlation, but improperly implies some kind of causality. For example, if a player wins the first set he is likely a better player, and therefore if he goes on to win the match it’s probably due to skill and not because the momentum from winning the first set propelled him to victory.
What made this instance of statistical misrepresentation so frustrating is that there was an easy alternative. CBS could have simply told viewers the ratio of players who won the first set in a tiebreak and then went on to win the match. This sample would only include matches that were relatively even in the first set, and therefore any correlation between first set victories and match victories is much more likely to entail some causality. Of course if it turns out that only 6 of 10 players who won the first set in a tiebreak went on the win the match, then CBS has nothing interesting to fill airtime with.