Of course this difference completely relates to the usage of top computer rankings and both polls in the GPI, and the polls by themselves. Our table at http://www.collegesportingnews.com/c...2-FCS-Rankings displays the top 25 of each side-by-side.
Since we know the components are different, what else may be the reason for the disparity between them?
The disparity mainly lies in the computer rankings and how they factor in each game (and score in those that use margin of victory) and strength of schedule. It would require superhuman abilities to rank all 122 FCS squads so precisely so our trusty computers do their job in tabulating the results in an unbiased method as possible. All the teams are adequately interconnected by opponents to analyze the wins, losses and point differentials in such a way that most humans cannot objectively achieve.
What are the main top 25 differences now between the GPI and the polls?
If you put the GPI next to each poll at a glance there are several points that pop out.
+ The GPI agrees on only three teams ranking in both polls.
+ There are two additional teams where the ranking of the FCP matches the GPI.
+ The GPI ranks three teams that neither poll ranks (each an MVFC squad).
+ There is an additional team in each poll that the GPI ranks and the poll doesn't.
A simple way to get an overall idea how the GPI and the polls differ is to subtract the poll ranking from the GPI to get either a negative number (poll under-ranking) or a positive number (poll over-ranking) in relation to the GPI.
|N Dakota St (7)||1||3||-2||3||-2||0|
|E Washington (2)||3||1||2||1||2||0|
|Sam Houston St||5||5||0||5||0||0|
|S Dakota St||13||N/A||N/A|
As the table illustrates, compared to the GPI the polls most greatly under-rank Indiana State, Cal Poly and Northern Arizona. This does not include the teams the polls do not rank at all including GPI #13 South Dakota State.
Compared to the GPI the polls most greatly over-rank Delaware, Lehigh, James Madison and Towson.
If you add the poll's under and over-rankings together it's shown that they over-rank teams by 55 or more places. And worse, if you look at the "others receiving votes" (each full poll is linked off our http://www.collegesportingnews.com/c...2-FCS-Rankings) it gets more grim with teams being differently ranked by much larger margins.
Conversely, if you do the same exercise and subtract the SNW from the FCP poll you see that they differ by one spot in total. Seems almost like the polls are being voted by the same people.
Does it matter that they differ?
While the GPI and the polls are the only three rankings used by the NCAA Championship Committee officially in any capacity and they might only affect the Pioneer Football League (the only participating conference that has asked for an autobid to the tournament and not receiving one until next year), they do play a lesser role in indicating which at-large teams might be selected. This is where the GPI has historically outperformed the polls. And at this stage of the season you have to wonder what people are thinking to rank a team on their top 25 ballot that is a clear GPI outlier.
Luckily all three components are works in progress that will be closer to the same at the end, and none of them finally determines who gets to play for the NCAA Division I Football Championship.