By BrianC
November 25 2004

The London Double Header has caused a headache for this statistician. How do you allocate the crowd? In two, in four, ignore it? Whatever way you chose is imperfect. 

It appears that I am not alone in having difficulties. On Wednesday premier rugby published inaccurate statistics for crowds to date this season. The root cause of their incorrect figures is an error related to this game.   

When our own club published what I considered inaccurate crowd figures I was curious as to the source. Subsequently Premiership Rugby Ltd. issued a gushing press release with figures that were significantly different to my own. This then confirmed the source and sent me hurrying to check my own figures. To be honest my first assumption was that I had somehow made the mistake.

It took several hours to check the individual figures and formulas. I could find no major mistake in my calculations. So why were my figures and those of PRL so different, particularly given that I was using their very own figures as my source? The answer came to me via the size of the discrepancy. The numbers published indicated that they were recording a figure fifty one thousand greater than mine.

The issue here appears to be a simple case of Enron-style double counting, although it has to be said that there is no malice in PRL's error. In terms of developing the sport they have a lot to be proud of. My firm belief is that they have allocated a home crowd of 51,000 for the London double header to both London Irish and Saracens. They have thus counted the Twickenham crowd twice. This has skewed the total they published by some 10% and makes the percentages meaningless.

This has still to be confirmed by PRL, I have had no response from them. They may in fact never acknowledge this mistake. It will be interesting to see what in the way of correction, if anything, comes out of their press office.