A recent article in the Wall Street Journal claims an increase in overtime games in college football is explained by greater parity and cites a number of sources to bolster the contention. When the football (not statistics) experts suggest reasons why parity is responsible, this is a case of retrofitting a rationale to a statistical peculiarity.
A closer look at the data indicates it is more likely this season is an outlier than an indication of a trend. Over the past twenty years, there have been four upward spikes at 2000, 2005, 2012, and 2016. During that same period, there have been three downward ones at 2001, 2009-2010, and 2013-2015.
Look at the graph and you will see that at no time after a spike did the percentage go up again. If parity was the real explanation, the change would be ever-upward, not bouncing up and down. To the expert statistician, there is only a slight upward creep over the past twenty years.
This is an example of what experts call regression to the mean, where extreme values are likely to be followed by values closer to the long-term average.
A better measurement of parity than a spike in the incidence of overtime games would be teams’ winning percentages – decreases for historically better teams and increases for historically poor-performing teams. Full parity would be achieved when every team wins and loses half their games.
Think about averages this way.
Washington, DC gets an average of 40.8” of rain per year. This does not mean it rains .1117” every day. Some days it will rain more and some days it will rain less, but it will almost never rain exactly .1117″ any day. But here’s one thing you can count on: it will not rain exactly .1117″ every day.
We found more than 700 songs about rain online. To see the one we chose, click here.