Ed: Good stuff here, and lots of hard work. Can't imagine why a Chelsea fan would want to watch that first leg again, though...
I put my back out yesterday, and have therefore spent most of today lying in bed with nothing but my laptop for company. To give myself something to do to take my mind off of the pain, I decided to have a re-watch the game from last week. Further to the recent discussions regarding possession stats and the inaccuracies in the way that they're reported, I decided to track the amount of time that each team actually had possession of the ball to see how it compared.
Obviously there are many inherent inaccuracies in this kind of data collection, most of which Graham has already outlined here so I won't bother going over them all again. For the sake of clarity, however, I will outline the exact method I used to collect this data. A team was deemed as being in possession of the ball when a player had the ball at his feet (or in his hands, if it's the goalkeeper) or was in the process of passing to a teammate. Both clocks were stopped when the ball went out of play or if there was no clear team in possession (goalmouth pinball from a corner, long clearances etc).
The possession split as reported by WhoScored was Barca 79%, Chelsea 21% which tallies quite well with the previously suggested passes/total passes algorithm (which gives 79.5% and 20.5%).
For some reason I can't seem to get images to show up in this post, but a table with the actual possession times of each team (in minutes and seconds) can be seen here*. *(Now seen below...)*
And here it is in pie chart form...
The thing that really jumps out at me about this number is just how much time over the course of a match neither team is in possession of the ball. The following two links show the % of possession for each team and the % of game time that each team possessed the ball.
I'm not entirely convinced that the differences in these numbers is statistically significant, especially considering the potential inaccuracies involved in the data collection, but it's interesting to see the comparisons.
I'm not really sure what analysis is appropriate for this data (if any) so do with it what you will :-)