Caesar's View, Part Two

[VolunteerGeneral]

Posted by silverpie:

(which is a more rigorous form of what lazarus did)

 

Note that there may be errors in what I did.

 

When analyzing by private/public, private/private, and public/public, the differences turn out not to be statistically significant (to put it in plain terms, you'd expect to see a difference at least that significant 11% of the time, a number we call alpha--to be considered significant, that percentage should be less than 5%). Adding in the Division II games makes the differences even less significant (alpha goes up to 17%). I think the structure of D2 distorts the data.

 

The difference by classification is significant--alpha there is just 2.2%. Again, adding the three D2 classes (which averaged 27.3 in small, 15.1 in medium--counting the overtime final as a margin of 0--and 17.4 in large) reduces the significance.

 

I also analyzed the D1 data by round. This turns out to be the most significant factor of all (alpha of just 1.6%); averages are 25 for the first round, 19 for the second, 16 for quarters, and 15 for semis.

 

 

OK...It`s time for more than one of us to go to BED!!!

 

VG

[Indian]

I can see how such a system might work, maybe take the records the last five years, schedule strength and classify the teams that way. Just from this past year's results, not looked at very closely but just what I remember a 1A group around Chattanooga might have Lookout Valley, Bledsoe County, McMinn Central, Walker Valley, Sequoyah and Copper Basin. A 2A group might have Brainerd, Sequatchie, Meigs, Tellico Plains, Notre Dame, Whitwell. A 3A group might have Marion, Polk County, Rhea County, Chattanooga Central, Bradley, Red Bank. A 4A group might have McMinn County, Soddy-Daisy, East Ridge, Hixson, South Pittsburg, Grundy. A 5A group might have Ooltewah, Tyner, Baylor, Boyd-Buchanan, McCallie and Cleveland. Those could be switched but it might look similar. The main problem I see, just like before any split, what happens if McCallie is at the top of the local region each year, or one of the current Super 7 teams wins the state every year, a real possibility. It would go back to the same complaints.

[lazarus]

i thank you for your response, which is better than i had hoped. as a casual reader of science publications i have seen that good science requires opposing viewpoints and close scrutiny. researchers develop a hypothesis, then perform studies/experiments & collect data, then finally present their results. at that point their methodology is questioned & their conclusions challenged, others even seek to duplicate their results. the world of science actually looks like a pretty rough & tumble place from where i sit.

 

well, this is hardly science, and i am probably as unqualified a person as there is to advance a hypothesis, but what the heck, no one else was doing it. before i try to answer some of your questions, i would have to say that i would not act on the results of just this year's data. altho i feel that 150 games are enough to produce numbers worth following up, i would be very interested in having prior years playoff results added, if anyone knows where i can get them?? possibly they would show this year to be an anomaly, but if the results were similar...

 

Q1) obvious how? that is a very good question, and one for which i can only offer an inadequate answer. since i do not know the formulae for calculating significance, i rely on looking for numbers that swing so far from the norm that i consider it apparent that there has to be a reason. and, while i must confess to educational deficiencies, sifting thru piles of numbers to see what they tell me (especially sports statistics) has been a lifelong passion, and i have a pretty fair track record for being able to predict based on those conclusions. my interpretation is also biased by the memory of prior years information. i have played with the playoff results every year since they started. unfortunately that information is gone. my unreasonable termagant wife will not allow me to fill the house with stacks of paper covered with columns of numbers (a pass-time that she considers as falling somewhere between foolish and mentally ill), and she controls enough critical assets to get her way.

 

as regards the private-public and private-private data, most of my reservations are with the sample size. 18 games arent very many, altho the essentially complete sweep is pretty compelling-i CAN calculate basic probabilities. as an interesting aside, the 6 private-private games matching higher & lower seeds resulted in a 50-50 split. between private-public with the private a lower seed, the lower seed won 80%, of course with privates having the higher seed it was 100%. the overall ratio from as long as i can remember, lower seeds always win around 20% (this year was 22%). i wouldnt be comfortable drawing conclusions based on that information alone, due to the sample size. however, as far as the public-private comparisons, those, along with the margin, are just a continuation of the long-term trends that i have noticed (but as i have said, cannot currently document).

 

q2) private-private margin. that s a good question. to be honest, i have never looked at that before. the sample-size question is there, and the 50-50 split between higher/lower seeds raises doubts about the significance of the margins. however, i did not expect find the higher number. i would really like to see how that has played out over time. however, i already felt that there was sufficient evidence that an all-private league would cover too wide a field, due to the small number of teams, which range from the super-7 to the lighthouse christians. but i have to confess to a little intellectual dishonesty in including a number that i did not myself believe to be conclusive.

 

q3) arguments rendered moot due to a lack of evidentiary statistics. to quote my kids, "no duh"! i never intended that to be percieved as anything but opinion. remember my objective is to create discussion, this was more of a theme with data, than a study full of opinions.

 

q4) 14 point blowouts. no, i didnt mean 14 points was a blowout. i just sequenced the material poorly and gave that impression. over 28 points was meant as the blowout. that wouldnt be much for a regular season, but in a playoff that is a pretty sizeable margin to occur so frequently. of course, it is equally easy to attribute that to having too many teams in the playoffs. what i really needed here was some sort of a "control" group. i am still rasslin with what would make a good set of control data, since simple probability wouldnt really fit, and i cant think of what set of data that does exist would be comparable, yet exclude the factors i wish to examine. not to mention that i only have so much time to cover sheets of paper with columns of numbers. ultimately i gave more weight to how the margins compared between the different groupings.

 

q5) access to original data. its there for you any time: i got the results from the tssaa site, playoff brackets. as for documenting my methodology, i am handicapped by the lack of formal training. what is needed is for an educated person to pick up the ball and prove or disprove the conjecture...hint, hint

 

q6)the magic 2-touchdown margin-i said up front that was subjective. but i did make note that there is NOT a wider range of enrollments in 5-A. every way i could think of looking at 5-A it actually fell in the mid-range of the classes.

 

q7) statistical significance of the 55% higher margin in 5-A compared to the other classes. again, i really dont know. what i am more interested in finding out is whether that difference would hold, given a larger sample size (more years results). finding the highest margin in 1-A, being the predicted result if my hypothesis for the reason for the 5-A margin, gave it some support in my mind. and experience tells me that the size of any variance, as it holds thru a larger and larger sample size, becomes less and less likely to swing back to the norm. 150 games is enough to really pique my interest.

 

q8) do the data support the conclusions. good question. i hope that smarter people than me will take an interest in this. i have had the feeling, since the whole classification thing started, that there has to be a better way. however, my initial thoughts on the subject have been considerably altered by the discussions here (in the few thinking posts or threads). i continue to be prepared to learn more from others in order to reach more accurate conclusions. i believe it to be inevitable that major changes occur in the structure of hs sports in tennessee. the easy way, the "safe" way from an administrative point of view, is to merely split and run each division the same old way. the same way it has always been done, and the same way everybody else does it. i dont think that is good enough. however, before we could expect there to be the least consideration of a change, we would have to be able to present compelling evidence that such a change would be desirable. to produce such evidence, there is nothing better than to have the data examined by qualified individuals who disagree with the conclusions. so, nonews, i hope you will pick me apart on this.

 

q9) could the data support a split. we all know that data can be manipulated to support anything. my livelihood has depended to a great extent on my ability to plow thru piles of numbers and predict outcomes of future actions based on that data (yes, my hobby and my work are just alike, except no-one is usually interested in what i do with sports numbers). i have found that there are two different types of people whom i serve. those who say "this is what i want the numbers too say" and those who say "what do the numbers say". the oddest thing is, the first group generally ends up failing to meet their objectives, while the second group usually succeeds. if i dont find some surprises while i play with numbers, i know i am doing something wrong.so, i eagerly anticipate seeing what you can come up with.

[nonews]

(post trimmed for brevity)

..."The problem I have is that you never make a point"...

 

...good point! I'm glad you noticed! The only point I was even trying to make was that IMO the data does not adequately support his conclusion(s), and that both a larger sample size and population (most likely in the form of similar statistics from earlier years) would be necessary to account for deviations from standard variance...

 

..."you disingenuously ignore what are generally regarded as de facto (forgive me; I have yet to learn how to use the italics, bold, etc. buttons below) statistical givens"...

 

...I stand corrected, I, like Lazarus and many others here am not a professional statistician, and any knowledge I have of the subject is admittedly rudimentary in nature and based on long-forgotten and little-used courses taken eons ago ...

 

..."Do you not think we could follow your reasoning? Begin with the basic assumptions and show why they are false before you launch into the more calculated and intricate obfuscations of statistical analysis"...

 

...quite often I'm not sure I can follow my reasoning, as you so eloquently pointed out at least some of it is inherently flawed... BTW, that's a great quote from your SA class instructor...

 

..."What post are we reading? He did this, albeit not in the stasis of a formal paper. Read all his posts before positing a defamation that is inaccurate"...

 

... we are reading the same post, I wasn't intentionally defaming Lazarus or anyone else, merely disagreeing with the conclusions presented... as for the statement you question, it was poorly worded, the implication was that an analysis of the same MOV numbers from pre-split games would (again, IMHO) provide a more accurate comparison and would in addition provide a sort of "control" group with which to compare the more recent data... note that in his later post Lazarus bemoans that same lack of a control group for comparison...

 

..."Come on, dear sir. This is the equivalent to asking a student to prove that it is a discrepancy between a corporation earning a 26.7% dividend as opposed to an industry-wide 6.5. As a professor of mine once said, "Sometimes it is true: the numbers speak for themselves apart from some purely academic proof of their veracity"...

 

... no disagreement here, but after re-reading the original post I'm not altogether certain that these numbers will support any conclusion except that, particularly in 1A, some very weak teams are presently included in the playoffs, evidence of which is the shrinking MOV found in the later rounds in all classes... it would be interesting to compare these numbers with similar data from the pre-split days...I do have one remaining question though, namely that 3 and 4A have higher MOV numbers than 2A, yet these classifications do not currently include any private schools, shouldn't their MOV numbers be closer to those of 5A if the conclusions are supported by the data?...

 

..."Thank you for the humor"..."I wish I could retake some of my final exams under the dictates of your axioms of statistical relevance. Instead of a MAGNA I could, rather easily I think, get the coveted SUMMA"...

 

...you're welcome. Like you, I don't think my chances of teaching statistical analysis anytime in the near (or far) future are very good, however I would assert that at least some valid points were raised by my over-lengthy post...

 

[Guest Dingo]

pujo and VolunteerGeneral,

 

OK, we are all "good ol' boys," but I have read too many posts from both of you to buy into this "Gosh-dang, we just don't understand" rhetoric. Yeah, you do.

 

So you have to read a little more slowly. Big deal. Do it and join in. You both have germane comments to add and you know it. Get out of the "I ain't no geek" mode for which you would excoriate your own child (VG; I don't know if you have kids, pujo).

 

Quit pretending.

[Guest Dingo]

silverpie,

 

I will respond to your 1-3 points at later date (good points all, but I want to view this proposal in detail before shooting holes through it). I want to ask if my Shakespearean minutiae is still intact.

 

Ok, then. Is this when the Friar is telling Romeo about the strengths and weaknesses of the various herbs at his disposal? Something like, when you take this one in moderation it relaxes you, but if you take too much it kills you. Am I right? Do I win a prize?

 

And, no, I am not surprised to read a Shakespearean allusion on this thread. On other boards, yes. But not here.

[pujo]

A salesman goes to a ladies home one day,and you know how they like to tell jokes to get their foot in the door.So he says to the lady,let me tell you a polock joke.The lady replies ,I'll have you know right now, I'm polish. The salesmam replies,I'll tell it slow. So me and Vol Gen might be able to follow along if we take our time.

[Guest Dingo]

silverpie,

 

I am trying desperately to understand your post. Your conclusions are based on a positive designation of an Alpha. If you posit a negative correspondence, the very percentages you refer to overwhelmingly suggest statistical anomalies of significant proportions. At least, it seems that way to me.

 

If I am reading the threads correctly, it is the position of the rural public schools that the statistical evidence of the negative results (losses) prove that there is a deficiency in the equity of the playoff system as it now stands (I defer to RHSfan and others for the pertinent numbers). My question is how does positing a positive Alpha answer their charges?

 

What comparison of contests are you using to decide that DII results skew (your word: distort) the data? I am befuddled here, I admit.

 

I guess I am asking this: for the sake of argument only, if you posited a negative Alpha, would not some of the very figures you quote suggest a statistical superiority that would not occur in a non-controlled significant sample? I am not disagreeing with you, necessarily. I just do not understand why you are beginning where you are.

[Guest Dingo]

pujo,

 

OK. If I understand the meaning of this post, you just fortify my point.

 

You understand. You comprehend. Trying to pretend that what I say is over your head is like the eagle telling the goose, "Your flight in flock is more beautiful than mine; you are so logical in your formations, so supportive of your mates. Obviously I am inferior to you, and you have nothing to fear when I soar above you and look down upon you. What threat can a clumsy glider like me pose to such a strong wing man as you?"

 

Only the most stupid of geese would respond with anything other than, "Yeah, right."

[Guest Dingo]

lazarus,

 

Thank you for your post. I want to continue with your ideas, and then I have some questions for you.

[Guest Dingo]

nonews,

 

Thank you for a terrific and humbling response; I told you that you have found the thread that would welcome your ideas, even as I attacked what I thought some unwarranted points. I apologize for my too-forward response; I honestly thought you were claiming to be a statistician, and I feel I responded too harshly.

 

As a matter of fact, you DID make some very pertinent points that deserved better treatment than I gave them. Actually, as you already know, I gave those points no treatment at all. I just went for the jugular, as is my wont.

 

I must, as VG suggested, get to bed myself, now. Still, I want to say that I will respond in more detail tomorrow and that I would hope to see more of your voice on this thread instead of less.

 

"If you listen only to those who agree with you, if you take counsel only of those who share your views, if you avoid facts that make you uncomfortable, then you are constrained to view your foes as unworthy baboons, their advice as trivial, and their views as self-evidently wrong; no institution you construct will serve our great nation; you will have sealed your doom as an independent thinker."

[VolunteerGeneral]

LOL Dingo!

 

OK...I won`t complain about your posts anymore. Maybe I can dig out my old Thesarus and join in the fun. I`d hate for coacht to be forced to start an intellect board. Man...could you imagine all the threads that would get locked up on that board?

 

VG(whose endeavors to verbalize comparably with the lofty measure of Dingo may go awry, but would epitomize my fondest expectations should I be able to author posts that are half as exemplary.)

[Edited by VolunteerGeneral on 12-6-02 5:49P]