The base file we use, the point of reference, is the last year. We looked back over the past history going back to January 1, 1996, and presented that information. In that way, it was much easier to put it on one chart. We don't have to worry about double-counting persons as we go back in time. We may see someone in 2003 with a conviction, and again in 2002 with a conviction, and if I present the data for 2002, I have effectively counted that person twice. We're trying to remove the possibility of counting the same person multiple times.
On November 22nd, 2006. See this statement in context.