Top 25 talk

[dansand]

Albion is obviously outstanding this year and Brandon Crawford is putting up Jeff Gibbs-type numbers. The question I have is, how in the world did they lose to UM-Dearborn? Maybe some of the MIAA posters could offer some insight.

[Gregory Sager]

I'm just not following the logic of your argument.  If you acknowledge that history matters, then why should it NOT be a legitimate factor in voting?  I'm not talking about in the 12th poll - by then the current season provides adequate data.  I'm talking about in the early polls, when there IS little data from this season.

History -- or, to be more precise, a proven coach or an experienced team or great crowd support, or some sort of combination of these things -- matters in that it may have an effect upon the outcome of a game. But, again, you're confusing primary causes with secondary and even tertiary causes. The outcomes of the games themselves are what ultimately matters. Anything that influences the outcomes of those games -- whether it be a proven coach facing an inexperienced one, the best player on one of the teams suffering from the flu, a ref who calls a bad game, whatever -- is important only inasmuch as it affects the game. They're factors that are part of a greater whole, and the greater whole -- the outcome of the game -- is what is considered.

See where I'm going with this? I'm saying that the judge at a baking contest compares the taste of one pie to another. You're saying that the judge should take all the pies back to the lab to find out how much monosodium glutamate each pie contains, or whether or not the contestants are using confectionary sugar or corn syrup, before actually biting into the pies. Plus, you're saying that last year's contest winner should get extra credit in this year's contest based upon the tastiness of last year's pie, while the novice baker should be penalized for not having paid his or her dues.

The early polls (say, weeks two and three) are distorted by their reliance upon the preseason poll in addition to the games that have been played. They constitute the acme of what you desire -- a polling situation in which factors other than this season's results count. They're always the most inaccurate polls of the year, in that they usually "bounce" the most wildly in terms of teams moving up and down the ladder and in that they tend to be the ones that least reflect the final poll. As I said earlier, I argued several years ago with Pat that he should dispense with the first couple of polls, since they're tainted by the influence of an extraneous, non-relevant factor (the preseason poll, which shouldn't matter at all once games begin). Pat's not going to give up running those first couple of polls, and I can understand why -- the interest is there among people to see those polls. But those first couple polls or so are tainted to a degree, and I regard them with a lot of suspicion.

Now, on top of that, you enter the picture and start arguing that extraneous factors other than actual game results should be valid aspects of balloting until midseason, and that just baffles the heck out of me.

We assume that a victory over, say, a Wooster in game three is much more impressive than a win over, say, Caltech, but afterall, neither Wooster or Caltech would have enough of a record THIS season to say that for sure.  We base it on history (and usually correctly).  Later on, when more data is in, we can reconsider.  Lawrence's meteoric rise in the rankings was largely due to beating #2 Oshkosh.  In retrospect, it was a good win, but not nearly the win we all though at the time.

Chuck, don't make me quote The Odd Couple on the subject of the word "assume".  

You're lumping a lot of things together under the word "history" here. Yes, a win on Opening Weekend over Wooster is considered more solid than a win over Caltech. But this goes back to the preseason poll, in that a win over Wooster is a win over a team that returns a lot of players from a 20-win team the previous season, while a win over Caltech is a win over a team whose players, even if they are returnees, are carried over from a nearly-winless outfit the season before. If Wooster does not return a big chunk of the previous season's rotation, that Opening Weekend win means a lot less. Look at it this way: While people will make a fuss over beating Illinois Wesleyan early in the season in 2006-07 based simply upon the program's reputation, it won't mean nearly as much in terms of the poll as a win over Illinois Wesleyan would've meant early this season, because the Titans are going to lose so much to graduation in the intervening period. It's the reason why, although I'm sure Lakeland and Clarke fans were delighted to beat UWSP before the holidays, pollsters and national observers took little note of it. The Pointers are not the same team as the one that won the last two national titles.

Speaking of WIAC teams, Wisconsin-Oshkosh is a prime example of why the early polls should be viewed with such great suspicion. I'm not knocking the pollsters for ranking the Titans so high in the preseason, since the evidence indicated that they were a good educated guess for success in 2005-06. But their #7 ranking in the first week's poll and #14 ranking in the second week's poll were clearly unmerited by actual performance; they were up that high only as a residual of their preseason ranking, and their presence in those lofty perches taints those first two polls. By week three they were #25, and based upon their current status that appears to be more justifiable.

Ranking teams upon criteria other than actual wins and losses this season is, as Q said the other day, inherently unfair. It's also an invitation to court inaccuracy, as UWO so amply demonstrated in the first two polls. Top 25 rankings should be handed out upon the basis of the Smith-Barney principle. In other words, teams should be ranked the old-fashioned way ... they earn it.

[Titan Q]

Pat, you are correct.  Pat Cummings was basically saying that just based on the North Central loss, IWU would have been #2 or #3, but when combined with the close game vs Carthage 3 days later, it resulted in the drop all the way to #5.  As I stated, I think Pat Cummings is absolutely correct about the Week 6 poll dynamics and how the voters evaluated IWU when they voted Monday.

This is where I wish we (D3 fans, poll voters, etc) could all see more games.  I think most people who watch 4-9 Carthage play leave saying, "Wow do they have a lot of talent."  Lawrence head coach John Tharpe told me exactly that two weeks ago.  Again, Carthage followed up the near upset over IWU by nearly winning at Elmhurst Wednesday.  In all 9 of their losses (all vs good teams), they've been competitive:

vs Concordia (CA) - 75-78
vs Westmont - 69-72 (OT)
vs Calvin - 78-82 (tied with :50 to play)
@ Hope - 61-75 (17 pt lead in 1st half, tied with 13:55 in game)
vs Carroll - 74-82 (tied with 3:19 to play)
vs Lawrence - 65-78 (down 4 with 5:56 to play)
vs Gustavus Adolphus - 70-77 (2 pt game with :58)
vs IWU - 79-80
@ Elmhurst - 73-78 (led by 5 with 3:28)

Now, the 2005-06 Red Men obviously can't close games, but based on the above, a close win over Carthage shouldn't be viewed as something bad just because they have a bad record.  My basic point was that when evaluating opponents, we all need to make sure we're not just looking at the opponent's record.  

[Mr. Ypsi]

Greg,

I have concluded that we are not really in disagreement, we're just using terms a bit differently.  I think I caught on to that with your baking contest - what YOU are not taking to account is that with early polls, the judges (voters) are NOT tasting pies, they are tasting ingredients!  (Or, at most, half-baked pies!)  And what I'm calling history is far more than just the preseason poll - we would also no doubt give more credit for an early win over, say, Millikin than an early win over, say, Eureka.  These assumptions are based (largely) on what I am calling history (after all, I'm sure even Pat's most conscientious voters don't go over all 400+ rosters with a fine-tooth comb).  We assume (usually correctly) that even a mid to lower level CCIW or WIAC teams is PROBABLY better than even a higher level (not necessarily #1) SLIAC or LMC team (before you jump on my example, I'm aware that Eureka is not a higher level team even in the NIIC).

[BTW, I'm pretty sure that "when you assume you make an ass of u and me" predates the Old Couple - I knew that gag in the 1950s.]

That 'history' does make up an improperly large factor once this season's data has become more telling is, no doubt, exacerbated by overly-early polling and the 'human nature' tendency that voters are NOT going to look at things totally from scratch each week.  Quite naturally, they are going to take their ballot from the previous week as a template, and move teams up or down from that starting point.  Polling before most teams have played at least 8-10 games is not really justifiable logically.  BUT, I'm not arguing for Pat to abandon the pre-season and/or early season polls - like I said, what else would we have to talk about in December?!

[Pat Coleman]

Albion is obviously outstanding this year and Brandon Crawford is putting up Jeff Gibbs-type numbers. The question I have is, how in the world did they lose to UM-Dearborn? Maybe some of the MIAA posters could offer some insight.

It was mentioned just a few posts ago -- NAIA teams can start practice and play games earlier than D-III schools can. It makes no sense for D-IIIs to play NAIAs early in the season because they come in at a disadvantage. Apparently it was UM-Dearborn's fifth game while it was Albion's opener.

[smedindy]

I agree. Sometimes Wabash brings an NAIA team for their tip-off tourney, and their record is like 3-2 or 4-1 already.

[dansand]

Albion is obviously outstanding this year and Brandon Crawford is putting up Jeff Gibbs-type numbers. The question I have is, how in the world did they lose to UM-Dearborn? Maybe some of the MIAA posters could offer some insight.

It was mentioned just a few posts ago -- NAIA teams can start practice and play games earlier than D-III schools can. It makes no sense for D-IIIs to play NAIAs early in the season because they come in at a disadvantage. Apparently it was UM-Dearborn's fifth game while it was Albion's opener.

Good point. I've seen the same thing when Augustana plays their cross-river rival St. Ambrose. Quite often it's their second or third game while the Fighting Bees have played eight or ten. Still, it's surprising considering how bad Dearborn is.

[smedindy]

I think Albion's loss to them could be just 'one of those things'.

[augie_superfan]

I know there are some people that dont necessarily agree with the built-in historical bias of the early season polls.  Also, there are some that say there shouldn't be a poll for a few weeks until we have some more info on the current season. 

This past year, the BCS used a new poll, the Harris Interactive poll, which didn't come out until late September when most teams had already played 3 or 4 games.  I thought I remember them saying that their reasoning for this was to take out that built-in bias that comes with a preseason poll.  So I compared the 1st Harris poll to the AP poll at that date and here are some of my findings:

24 out of 25 teams were in both polls
17 out of the 25 teams were ranked the same in both polls
4 were ranked differently by 1 place
3 were ranked differently by 2 places

Looking at this, I don't think that releasing early season polls will create a worse poll once some games are played.  These results tell me that either the Harris voters maybe were making their own early season polls eventhough they weren't being tallied yet or maybe they kept seeing the AP & Coaches polls which influenced them.

I don't know if the same dynamics would happen with a D3 poll as they did with an NCAA football poll (we all know that the NCAA has some ranking problems in football, with the BCS and all).  I just thought I'd bring this up as another source to generate some discussion, if we don't already have enough of that.

[cardinalpride]

Gentlemen, as long as humans are in control of the polls, "historical bias" will creep in.  It's just that simple.  Computers don't have an opinion.  People do.  Don't get me wrong, I'm not saying computers are the way to go.  However, because we base opinions on what we see, hear, read, feel, and conclude, "historical biases" are bound to creep in.  Right or wrong.

Greg, you used the "judge in the pie contest" example.  Nice example.  The difference is the judge actually gets to taste All the pies.  Many pollsters don't actually get to see all the teams play and they rely on information that sometimes isn't pertinent.  "Historical Bias" falls into that category.  Whether it's right or wrong is as much someones opinion as the actual D3 poll or any other poll for that matter.  Basically, what I'm saying is its all an opinion.  Why should we analyze someone elses opinion unless it's just complete nonsense.

[David Collinge]

Earlier today, in this swift-flowing stream of posts, a poster wondered aloud how Hope could have passed Albion in the poll without beating anyone noteworthy.  Only the pollsters and poll compiler can answer that, of course, but in playing around on my spreadsheet, I came up with a plausible mathematical way for something like this to happen.  If you are interested, and are ready to be bored, read on; if not, I suggest you move on to the next post. 

Suppose we live in an 8-team universe, and those 8 teams, listed alphabetically, are Albion, Caltech, Hope, Illinois Wesleyan, Immaculata, Lawrence, Wittenberg, and Wooster.  There are 8 pollsters who rank these 8 teams every week.  The polls are scored with 8 points for a first-place vote, 7 for second, and so on.  In Week N, here's the 8 ballots:

Wittenberg: 1-1-1-2-3-4-2-3
Wooster: 3-2-3-4-2-3-1-4
IWU: 4-3-2-3-5-2-3-2
Albion: 2-5-6-5-6-5-7-1
Hope: 8-8-7-1-1-1-6-7
Lawrence: 5-4-5-8-4-7-4-5
Caltech: 6-6-8-7-7-8-5-6
Immaculata: 7-7-4-6-8-6-8-8

The Week N poll therefore looks like this:
1. Wittenberg (3) 55
2. Wooster (1) 50
3. IWU 48
4. Albion (1) 35
5. Hope (3) 33
6. Lawrence 30
7. Caltech 19
8. Immaculata 18

In Week N+1, everyone plays the only other team in our universe, let's call them Duke, and everyone except Caltech beats them.  Duke, however, upsets Caltech.  In the next poll, every voter decides to drop Caltech to #8 and otherwise leaves their ballots unchanged.  So the Week N+1 ballots are

Wittenberg: 1-1-1-2-3-4-2-3
Wooster: 3-2-3-4-2-3-1-4
IWU: 4-3-2-3-5-2-3-2
Albion: 2-5-6-5-6-5-6-1
Hope: 7-7-7-1-1-1-5-6
Lawrence: 5-4-5-7-4-7-4-5
Caltech: 8-8-8-8-8-8-8-8
Immaculata: 6-6-4-6-7-6-7-7

The Week N+1 poll is
1. Wittenberg (3) 55
2. Wooster (1) 50
3. IWU 48
4. Hope (3) 37
5. Albion (1) 36
6. Lawrence 31
7. Immaculata 23
8. Caltech 8

Despite the fact that no poster changed their opinion about how the the seven teams east of Pasadena rank vis-a-vis one another, Hope passes Albion in the poll. 

This may be an extreme (and extremely silly) example, but I'm trying to make a point: that it's not just what you do that affects your ranking; who is around you in the poll, and how they do, also affects your ranking.

We now return you to your regularly-scheduled programming...

[Mr. Ypsi]

Gentlemen, as long as humans are in control of the polls, "historical bias" will creep in.  It's just that simple.  Computers don't have an opinion.  People do.  Don't get me wrong, I'm not saying computers are the way to go.  However, because we base opinions on what we see, hear, read, feel, and conclude, "historical biases" are bound to creep in.  Right or wrong.

Greg, you used the "judge in the pie contest" example.  Nice example.  The difference is the judge actually gets to taste All the pies.  Many pollsters don't actually get to see all the teams play and they rely on information that sometimes isn't pertinent.  "Historical Bias" falls into that category.  Whether it's right or wrong is as much someones opinion as the actual D3 poll or any other poll for that matter.  Basically, what I'm saying is its all an opinion.  Why should we analyze someone elses opinion unless it's just complete nonsense.

Your first paragraph is utterly correct, though I would change 'historical biases' to 'judgment'!

Greg's use of the 'judge in the pie contest' WAS good, because it illustrated WHERE I feel he went astray!  You added that the d3 judges (voters) don't get to taste all the pies.  Correct.  I observed that (in early polls) they are tasting ingredients (or, at best, half-baked pies), not pies.  Also correct.  But I think you went astray in your conclusion.

Unless you PREFER a strict computer analysis (e.g., Massey), don't denigrate the human element in polls.  I PREFER (good) polls BECAUSE they include judgment, and that judgment is (early on) for the most part what I mean by 'history'!

[Coach C]

I htink what most people are talking about here is not really historical bias as much as it's a preference given to voting for a proven coach and program over an unproven coach and program. 

Let's also remember that part of the reason to do a poll is to generate interest in our sport.  A pre-season poll does that really well since it usually incluses a large number of teams and gets people focused on the coming games.  There IS an element of FUN involved !

C

[bamm]

The City of Rochester gets the Chase Championship game we all were hoping for.  Tonight @8 #20 St John Fisher will take on U of Rochester.  Both teams blew out their tourney opponents yesterday (SUNY Brockport and Roberts Wesleyan). 

Edge to UR this year as the championship is in their gym... which will be packed.

[hugenerd]

Rochester has lost to two ranked opponents in their gym this year, let's see if they can get one against Fisher...it is a big game for them after struggling last weekend (18 point loss to NYU, 1 point win vs. Brandeis).