DMN Blog: Cowboys visiting with Igor Olshansky

FuzzyLumpkins

FuzzyLumpkins

The Boognish
Messages
36,574
Reaction score
27,856
BrAinPaiNt, Yeagermeister, Sarge, big dog cowboy, adbutcher, MichaelWinicki, Juke99, trickblue, ABQCOWBOY, WoodysGirl, Cbz40

funny I do not see Dallas, summer or BigDFan on that list. If you have an issue then take it up with the people on this list or keep it to yourself.

whether or not you like it tackle stats play an integreal role in determining player comparison. OTOH whining about our choice of topic does not.
 
ethiostar

ethiostar

Well-Known Member
Messages
6,309
Reaction score
46
So, who is this Igor fella really? I hear he has a sever back problem.
 
FuzzyLumpkins

FuzzyLumpkins

The Boognish
Messages
36,574
Reaction score
27,856
JerryAdvocate;2675808 said:
you have to go back a long time since I've questioned anyone's intelligence













except for BigDFan5's
Click to expand...

nah brah i wasnt saying that. i was talking about arguing for the sake of arguing.
 
BigDFan5

BigDFan5

Cowboys Make me Drink
Messages
15,109
Reaction score
546
FuzzyLumpkins;2675816 said:
BrAinPaiNt, Yeagermeister, Sarge, big dog cowboy, adbutcher, MichaelWinicki, Juke99, trickblue, ABQCOWBOY, WoodysGirl, Cbz40

funny I do not see Dallas, summer or BigDFan on that list. If you have an issue then take it up with the people on this list or keep it to yourself.

whether or not you like it tackle stats play an integreal role in determining player comparison. OTOH whining about our choice of topic does not.
Click to expand...


uhh Fuzzy I dont see your name on the list either, so telling me what to do, doesnt phase me even a little bit.

Tackle stats are meaningless because they are subjective, they arent even an official stat

This ludicrous conversation could have been done in private
 
AbeBeta

AbeBeta

Well-Known Member
Messages
35,684
Reaction score
12,393
FuzzyLumpkins;2675790 said:
Took you 20 minutes to come up with that. Nice.

Actually in this case there really is no basis for an acceptable degree of error. Error assumes an ideal like a theoretical prediction or prediction froma schematic. In this case there is no ideal and as such you would be looking at a percent difference but in the end it is one of only semantics.

The person who taught me an acceptable degree of error would have been my statistics professor at college. Its pretty common knowledge in the scientific and engineering communities that 5% is a good rule of thumb. For example neither QA nor peer review will pass something that can not at the very least meet that requirement. If I am not mistaken I believe that polling does that as well.

But you keep babbling about coming up with an acceptable degree of error without any context whatsoever. Error compared to what? There is really no basis for comparison because there is not even any data that is generated by multiple entities from a single source. In other words there is no basis for comparison.

If you want to say that a universal system for determining tackles would make for an acceptable degree of error then even that is not looking at it right. All that does is give a basis for determining what the degree of error would be.

IOW, if you want to come up with a system and then score tackles from all the game tape and then determine what the % difference between what they say and what you say is then we have at least some sort of baseline.
Click to expand...

First your 5% rule you note is specific to the ENGINEERING COMMUNITY. It is not "one of the very first things they teach you about statistics" because it is specific to your field. For other fields, the acceptable degree of error is less relevant in that many fields are not addressing point estimates or results that require the sort of precision you need in engineering. Moreover in experimental fields, the importance of an 'acceptable degree of error' is less relevant because researchers attempt to replicate findings and if you find similar results over and over, you've got strong evidence regardless of the accuracy of the measures.

FYI, Polling data generally have about a 3% MOE -- that's not because 3% is some well regarded level of acceptable error, it is because at about 3% you lose the battle between sample size and accuracy (i.e. you'd need a massively larger sample to increase your accuracy).

And again, it depends what you want to determine. if you want to determine beyond a reasonable doubt that a specific player had more tackles than another player then you'd need a perfect measure which just isn't possible when judgment calls are in play.

However there are ways to reasonably determine whether your measures are accurate enough. For example, does that measure produce outcomes that seem consistent with what you observe (e.g., that Canty and Igor are similar players) or are they inconsistent (e.g., that Hatcher is better than Canty). That's a good litmus test for whether you have enough accuracy. Even better, have two statisticians scoring each game. Their agreement will give you a good idea of the error.
 
fanfromvirginia

fanfromvirginia

Inconceivable!
Messages
4,014
Reaction score
164
FuzzyLumpkins;2675796 said:
you guys love it. admit it. anyway i need to go to the grocery store.
Click to expand...
I for one enjoyed every word of it. It was the greatest moment of my day. No, wait, strike that -- of my life! Yes, my entire life has been elevated by the argument between you and Abe regarding statistics in a thread regarding a Ukrainian 3-4 DE. And why do you ask? Because you both showed us all your inner (and outer) genius and wisdom.

You should turn it into a book. It would sell like hotcakes. You could call it "Abe and Fuzzy's Off topic Pissing Contest about Statistics in a Thread about a Ukrainian Defensive End on a Football Message Board."
 
You must log in or register to reply here.
Top