ESPN U Big Ten Preview
- Thread starter Hawkchop
- Start date
tm3308
Well-Known Member
If it's the same poll that took place at the Big Ten media days, I wouldn't put any stock into it at all. That's what, a total of 33 players? Not a very big sample.
MidnightBlueSix
Well-Known Member
Yeah that was my initial thought as well. It's nice to get the respect of our peers but part of me wonders how much of that is due to bitterness toward OSU. If I was at a school that the Buckeyes ran over by 35 points in 2009, I'd say anybody would win just to avoid anointing them for another year. Plus, 33 players doesn't really equate to a landslide. It is good that they respect us though.
SDK46
Well-Known Member
With all due respect, I'd say just the opposite.
Who better than to judge Iowa and have a valid opinion than the players at the league media day? Afterall, they're the ones who've actually played against Iowa and know first hand what we bring to the field.
tm3308
Well-Known Member
My point was that 33 players (3 of whom are Hawks) is not a very significant sample size. If they polled AT LEAST 3 times that many players and came up with the same results, then I would put a little more stock in it. But 30 players (since I'm assuming that our guys did not vote for Iowa) is not much. To get 38%, they only needed to receive 12 votes. I hardly think that 12 votes is a good representation of the Big Ten's players' opinion on who will win the conference
SDK46
Well-Known Member
I'll clarify....in this rare case, the #30 are the returning studs who have played against Iowa and the other league teams....it's quality over quantity...sample size is diminished in this case. I was be confident if extrapolated, the #30's picks would linear-out to a perfect league-player correlation.
tm3308
Well-Known Member
That's kind of reaching, IMO. Just because Terrelle Pryor and Ross Homan say Wisconsin is going to win the Big Ten, and Cameron Heyward thinks Iowa will (I'm just guessing on who OSU actually sent, but those are their best players/leaders), I really don't think that 2/3 of the team is going to think Wisky will win and 1/3 thinks Iowa will win.
How many players are on the active roster for each team? I'll just go with the 85 scholarship players for each team. That comes out to 935 players in the conference. 33/935 is less than 4% of the total players. Again, if you give me a sample size of at least 15%, then I'll start to buy in a little bit. That lets you get more very good players who know a thing or two about what's going on on the field. For example, guys like Posey, Sanzenbacher, Rolle, Chekwa, Saine, and Boren would be included for OSU; Sash, Greenwood, DJK, Ballard, for the Hawks, etc. There would be many more credible sources to draw upon.
MidnightBlueSix
Well-Known Member
I see what you're getting at but it doesn't really fix the problem. If I'm looking for a solution to the country's economic problems, for example, surveying 30 expert economists wouldn't necessarily yield the same quality of results from surveying 300 or 3000 expert economists. Yes, their experience playing against Iowa does give them some unique insight into who may win the conference. But with a sample size so low, the results become pretty irrelevant.
MidnightBlueSix
Well-Known Member
Yes, because statistically they've figured out that at a certain level, it doesn't matter how much larger the sample gets. It's the law of diminishing returns. As the sample size gets larger, the impact of expanding it shrinks. But 33 individual people is statistically insignificant if the population size is bigger than about a hundred (I'm estimating). Even then, it's a population that's easily swayed if only two or three people change their minds.
DFaceHawkeye
Well-Known Member
I'm going to assume this is the same poll that ESPN put in their last magazine. And I'm sure that they polled more than 33 players for that.
I think the players thoughts should be one of the best indications of who's who in the conference. They're the one's going head-to-head each week.
That's exactly what I was thinking.
Also, according to statistical theory, it doesn't take a huge sample size to have a reliable one. Of course that may not apply in this particular scenario.
tm3308
Well-Known Member
I agree that players (and coaches) are probably the best people to ask. I'm not disputing that. I'm only saying that I wouldn't find the results to be that big of a deal unless they polled a much larger portion of the conference, and not just 33 players.
I remember reading (I think on Rittenberg's blog) that the players were polled at Big Ten media days. This may be a different poll, idk. But if it's not, only 33 players were present for the survey.
SDK46
Well-Known Member
That's easy....you need to immediately subtract 20% of your roster number to take into account the true freshman who have never played a down in a league game. 33/748 = 4.4%
Then subtract another 20% who are redshirt freshman this year, who again, have never played against a league opponent. 33/561 = 6%
Then subtract out 5% who are kickers, and such, because they don't go toe-to-toe (pun intended) and really don't have a clue. 33/514 = 6.5%
So now we're left with 52 players per team. Let's subtract another 5% of those who were probably hurt last year and didn't play. 33/467 = 7%
From another angle....if you routinely go 1.5 deep at each position for significant playing time, then you have 33 players/team x 11 = 363 players who, last year, played signficant time. 33/363 = 9.1%
This isn't even factoring in the seniors from last year who graduated...which probably pushes that number into the double digits.
Either way, I'd say you're getting close enough to your requisite 15% of those, on this year's roster, who have previously significantly played against other league teams.
Last edited:
PeoriaHWKI
Well-Known Member
All I know is the Hawks got a lot of love... I hate feeling like there is a set up.
Rawkhawk73
Banned
Normally, statistics dictates that a reliable sample size of any population is 10%. That's assuming that the population you're testing is fairly evenly distributed and the sample size is relatively indicative of that population. In this case, if you take 85 scholarship players per team, that comes to 935 total scholarship players. 33/935 = ~3.5%. Definitely need a larger sample size to get a reliable result. But hey, I like hearing those results as much as anyone!! 
SDK46
Well-Known Member
Normally, statistics dictates that a reliable sample size of any population is 10%. That's assuming that the population you're testing is fairly evenly distributed and the sample size is relatively indicative of that population. In this case, if you take 85 scholarship players per team, that comes to 935 total scholarship players. 33/935 = ~3.5%. Definitely need a larger sample size to get a reliable result. But hey, I like hearing those results as much as anyone!!
Read my previous post where I quantified the numbers into the "correct" demoninator.
