This is a response to an ESPN.com article that claimed that the NCAA men's basketball week 6 AP poll is historically the most predicative week of the eventual national championship, noteworthy because of how early it's conducted in a season of 17 polls. The two metrics used to bolster that claim were that all champions have been in the top 12 of the week 6 AP poll and that "the average Week 6 ranking of the eventual national champion has been 4.7", the lowest of any other week.
By computing the arthimetic mean on ordinal data, the author fails to support this claim; I'd like to revise his calcuations using tallied votes to do so. More broadly, I'll analyze historical weekly polling data against all tournament results to answer a larger question: "Is polling on early season play (mostly non-conference games) undervalued when we pick our brackets?"