I’ve been writing for BI media for about two and a half years now. I enjoy the work, but often fret over its effectiveness. In my darkest days, I wonder if anyone actually reads the stuff. Web visit statistics from both the media companies and the OpenBI web site suggest readers are out there, and I generally get an acknowledging email at least once per column, but still I wonder. One time I plan on posting pure jabberwocky to test for attentive readers. Of course all that’ll do is test for an attentive editor.
For all my writer’s doubts, I was quite surprised by the enthusiastic response to the late August article: “The SEC vs. the Big 10 – A Tongue in Cheek Look at Football and Academics.” Pentaho CEO, University of Florida alum, and friend Richard Daley posted a witty response defending the Southeastern Conference (SEC). My nephew, an honors student at Atlantic Coast Conference (ACC)stronghold University of Maryland, edgily opined that ACC, with North Carolina and Duke providing both sports and academic leadership, trumps both the Big 10 and the SEC, and challenged me to expand my analysis.
An Iowa State neighbor dissed the Big 10 as overrated and underperforming in sports contrasted to the Big 12, home of 5 of the top 11 current football teams and even more top quarterbacks. And a UCLA friend of another neighbor haughtily dismissed consideration of any conference other than the Pac 10. After all he notes, UCLA became the first D1 school to win 100 National Collegiate Athletic Association (NCAA) championships a few years back, while Stanford is second in total championships and perennially wins the award for top overall sports school, even as it sets the standard for academics.
So I decided to accept the challenge of expanding the analysis to other conferences, and began searching for new ways to compare the prowess of Division I football programs. The big problem is one of research design. It would be nice to have a random or at least systematic inter-conference schedule of games involving top programs to test conference claims of supremacy. As it stands now, most games are intra-conference, so very good teams might beat up on each other, ending with mediocre records, even if their conference is superior. And rather than test themselves with tough non-conference games, top football powers often schedule weak teams from lesser conferences to enhance their records – and their chances for top year end bowl games. Much as I’d like to use existing inter-conference performance to rank teams, the data are just too sparse. And, unfortunately, the Bowl Championship Series (BCS) to determine college football’s national champion doesn’t address the problem. James Carville makes a good point.
If scheduling doesn’t promote an ideal approach for determining the best football conference, what’s an alternative that might make sound methodological sense? Colleges often evaluate themselves on graduate success getting desirable jobs or placements into prestigious medical or law schools. So why not evaluate football schools on their success getting “grads” to the next desirable level – the NFL? It certainly seems sensible that “better” football schools would be well represented in the NFL, and that top conferences would have higher numbers of NFL players than lesser ones. A reasonable approach would then be to contrast the performance of the major football conferences getting players into the NFL.
There must be demand for this information, because I was able to readily find a website, CBS Sports, with current data on the colleges of NFL players at the start of the 2008 season.1 A site detailing membership in the NCAA Division I college football conferences was also accessible from ESPN2, so I had most of the information needed. I then had to decide on a definition of an NFL player, and settled on those with a roster position or an injured reserve slot, excluding the practice squad. If a player had no number on the website listing, that often meant he was a late cut and was therefore not generally included. I then created a file detailing the number of NFL players by school within conference for ten D1 conferences and used the R statistical platform to examine the data.