Nate Silver statistics and the myth of Harry “Kid” Matthews

by Datechguy | November 3rd, 2012

Readability

Nate Silver statistics and the myth of Harry "Kid" Matthews

I may get kicked out of the saber­me­tri­cians union for say­ing this, but it seems to me that we’ve got to start tak­ing (Nolan) Ryan a lit­tle more seri­ously as a great pitcher

Bill James, The Bill James Base­ball Abstract 1986 P420

Doc­tor Who: All ele­phants are pink, Nel­lie is an ele­phant, there­fore Nel­lie is pink. Logical?

Davros: Per­fectly.

Doc­tor Who: You know what a human would say to that?

Davros: What?

Tys­san: Ele­phants aren’t pink.

Davros: Humans do not under­stand logic.

Doc­tor Who Des­tiny of the Daleks pt 4 1979

The more I look at Nate Silver’s rather amaz­ing math­e­mat­i­cal pre­dic­tions con­cern­ing elec­tion 2012 the more I see both Davros and Bill James say­ing these words in my head. Silver’s lat­est pre­dic­tion given events is sim­ply… incred­i­ble:

Pres­i­dent Obama is now bet­ter than a 4-​in-​5 favorite to win the Elec­toral Col­lege, accord­ing to the FiveThir­tyEight fore­cast. His chances of win­ning it increased to 83.7 per­cent on Fri­day, his high­est fig­ure since the Den­ver debate and improved from 80.8 per­cent on Thursday.

The facts on the ground, pffft, the trends, HA these polls have spo­ken and that is that.

My argu­ment, rather, is this: we’ve about reached the point where if Mr. Rom­ney wins, it can only be because the polls have been biased against him. Almost all of the chance that Mr. Rom­ney has in the FiveThir­tyEight fore­cast, about 16 per­cent to win the Elec­toral Col­lege, reflects this possibility.

I haven’t seen such blind faith since the Harold Camp­ing cult insisted the world was end­ing last year.

What Mr. Sil­ver doesn’t seem to real­ize is mea­sur­ing an elec­torate is not any­thing like base­ball stats. Base­ball stats are exact. A per­son is called out or safe a pitch is a strike or a ball and while the deter­min­ing of an error is sub­jec­tive to a degree, once it is ruled it is in the books forever.

But the most impor­tant thing about Base­ball is the sta­tis­tics gen­er­ally reflect what he will do, a sin­gles hit­ter expect in excep­tional cases hits sin­gles, a strike­out pitcher gen­er­ally pro­duces strike­outs etc.

Bar­ring an unex­pected injury, or an excep­tional sit­u­a­tion (World War 2 when the St. Louis Browns won the Pen­nant) once can see the rel­a­tive value of one player vs another pretty clear.

Yet even in this case it doesn’t always work, after all the New York Yan­kees despite an impres­sive pay­roll have not man­aged to win every world series.

How­ever Mr. Silver’s issue is not so much a ques­tion of the vagaries, after all his asser­tion that Barack Obama has an 83% chance of win­ning implic­itly gives Mitt Rom­ney a 13% chance of vic­tory. His real prob­lem is the data he is using, the polls of the MSM are not anal­o­gous to base­ball. They’re anal­o­gous to Box­ing.

But not Box­ing in the way of weight and style, box­ing in the sense of Harry “Kid” Matthews

Harry “Kid” Matthews was the price Rocky Mar­ciano paid for a shot at Jer­sey Joe Wal­cott. There is actu­ally a very inter­est­ing back story con­cern­ing this fight deal­ing with cor­rup­tion in the sport and a fed­eral inves­ti­ga­tion, but that’s more about the WHY of the fight. For our pur­poses we are more inter­ested in Mr. Matthews sta­tis­ti­cal record.

If you knew noth­ing about the fight game and just looked at his record, Harry “Kid” Matthews record of 77 wins 3 losses and 7 draws with 57 Knock­outs would indi­cate a more than for­mi­da­ble foe for Rocky Mar­ciano. If Nate Sil­ver using his math­e­mat­i­cal model looked at this match he would give Matthews an excel­lent chance of winning.

Unfor­tu­nately like the polls the record of Matthews is sus­pect. The polls that he is bas­ing his data on does not accu­rately reflect the elec­torate, either nation­ally or in the states as I’ve already dis­cussed. Harry Kid Matthews was the same as author Everett M. Ske­han put it in his biog­ra­phy of Mar­ciano on page 191:

He had had only one bout in New York, when he won a ten-​round deci­sion over Irish Bob Mur­phy in 1951. A few weeks before the fight, the word in the fight crowd was that Hur­ley had nur­tured Matthews with a steady diet of stiffs , and that Mar­ciano would take him apart.

The word was true, Mar­i­cano knocked him out in the sec­ond round, as Ske­han put it one page later

Rocky was unmarked, unhurt and not even slightly tired. The heavy­weight con­tender Matthews had been eas­ier to defeat than most of the mediocre club fight­ers Rocky had met early in his career at Providence.

And that is Barack Obama, a paper tighter with a paper record. Nate Sil­ver may believe that Barack Obama is rid­ing the great­est wave of his elec­toral pos­si­bil­i­ties but rather than look­ing at the record. Rather than look­ing at the fact that every sin­gle one of these polls show the pub­lic think­ing the coun­try is on the wrong track, rather than look­ing at the reg­is­tra­tion trends over the last 9 years as I did, rather than notic­ing that even poll­sters don’t say their model is likely, he is look­ing at data that is at best sus­pect and insist­ing he is right with Chap­ing like dedication

Yes, of course: most of the argu­ments that the polls are nec­es­sar­ily biased against Mr. Rom­ney reflect lit­tle more than wish­ful thinking.

and on Tues­day he will dis­cover: Ele­phants aren’t pink.

A final thought. The real prob­lem with Silver’s ascer­tains the left. Because of this asser­tion of Mr. Sil­ver the Camp­ing cultists of the left that will not even look at a source of infor­ma­tion that doesn’t pass their test of ide­o­log­i­cal purity will see Gov Romney’s vic­tory as “con­firm­ing” that repub­li­cans some­how “cheated” while the actual evi­dence pro­duced so far shows exactly the oppo­site.

We already see this type of think­ing in the com­ments at the Times.

Mitt Rom­ney is going to win this elec­tion and it’s not going to be close, but thanks to the flawed mod­els pro­mul­gated by Nate Sil­ver this clear-​cut results is going to lead to have con­se­quences far beyond those of los­ing an elec­tion. It will mean that once again the left will not ques­tion the poli­cies and their deci­sions that turned off the public.

I’m sure they will be grate­ful, as I’ve said lib­er­al­ism is a reli­gion and like the Camp­ing cult it can’t bear the thought of being proven wrong.

Update: Stacy McCain does the math:

His head­line sum­ma­rizes exactly what Repub­li­can poll-​mongers have been say­ing since Sep­tem­ber, as they see poll after poll with crazy over­sam­ples of Democ­rats. As for exam­ple, the NBC/​WSJ/​Marist poll that has Obama win­ning Ohio by 6 points — SIX FREAK­ING POINTS! — about which Ed Mor­ris­sey says, “all you need to know is this: the D/​R/​I is 38÷29÷32. In 2008, the exit polls showed a split of 39÷31÷30, and in 2010 36÷37÷28.”

Stop for a sec­ond and think about that: What this poll is telling us is that par­ti­san ID has shifted 2 points toward Democ­rats since 2008, which was the best year for Democ­rats since LBJ won a land­slide in ’64. There­fore, we must choose between two alter­na­tive explanations:

Obama is headed toward a world-​historic vic­tory based upon the remark­able pop­u­lar­ity of the Demo­c­ra­tic Party; or
The poll sam­ple is fucked-​up beyond all comprehension.

Take your pick, eh? When the sam­ple is D+9 and yet Obama is win­ning by only 6 points, why should any­one doubt that the NBC/​WSJ/​Marist lead is entirely a func­tion of the Demo­c­rat oversample?

You would think a base­ball guy would see that sta­tis­ti­cal illu­sion just like why out­field­ers had more putouts when Robin Roberts was on the mount, it wasn’t range (except for Ash­burn) it was a fly­out pitcher.

I may get kicked out of the sabermetricians union for saying this, but it seems to me that we’ve got to start taking (Nolan) Ryan a little more seriously as a great pitcher

Bill James, The Bill James Baseball Abstract 1986 P420

Doctor Who: All elephants are pink, Nellie is an elephant, therefore Nellie is pink. Logical?

Davros: Perfectly.

Doctor Who: You know what a human would say to that?

Davros: What?

Tyssan: Elephants aren’t pink.

Davros: Humans do not understand logic.

Doctor Who Destiny of the Daleks pt 4 1979

The more I look at Nate Silver’s rather amazing mathematical predictions concerning election 2012 the more I see both Davros and Bill James saying these words in my head. Silver’s latest prediction given events is simply… incredible:

President Obama is now better than a 4-in-5 favorite to win the Electoral College, according to the FiveThirtyEight forecast. His chances of winning it increased to 83.7 percent on Friday, his highest figure since the Denver debate and improved from 80.8 percent on Thursday.

The facts on the ground, pffft, the trends, HA these polls have spoken and that is that.

My argument, rather, is this: we’ve about reached the point where if Mr. Romney wins, it can only be because the polls have been biased against him. Almost all of the chance that Mr. Romney has in the FiveThirtyEight forecast, about 16 percent to win the Electoral College, reflects this possibility.

I haven’t seen such blind faith since the Harold Camping cult insisted the world was ending last year.

What Mr. Silver doesn’t seem to realize is measuring an electorate is not anything like baseball stats. Baseball stats are exact. A person is called out or safe a pitch is a strike or a ball and while the determining of an error is subjective to a degree, once it is ruled it is in the books forever.

But the most important thing about Baseball is the statistics generally reflect what he will do, a singles hitter expect in exceptional cases hits singles, a strikeout pitcher generally produces strikeouts etc.

Barring an unexpected injury, or an exceptional situation (World War 2 when the St. Louis Browns won the Pennant) once can see the relative value of one player vs another pretty clear.

Yet even in this case it doesn’t always work, after all the New York Yankees despite an impressive payroll have not managed to win every world series.

However Mr. Silver’s issue is not so much a question of the vagaries, after all his assertion that Barack Obama has an 83% chance of winning implicitly gives Mitt Romney a 13% chance of victory. His real problem is the data he is using, the polls of the MSM are not analogous to baseball. They’re analogous to Boxing.

But not Boxing in the way of weight and style, boxing in the sense of Harry “Kid” Matthews

Harry “Kid” Matthews was the price Rocky Marciano paid for a shot at Jersey Joe Walcott. There is actually a very interesting back story concerning this fight dealing with corruption in the sport and a federal investigation, but that’s more about the WHY of the fight. For our purposes we are more interested in Mr. Matthews statistical record.

If you knew nothing about the fight game and just looked at his record, Harry “Kid” Matthews record of 77 wins 3 losses and 7 draws with 57 Knockouts would indicate a more than formidable foe for Rocky Marciano. If Nate Silver using his mathematical model looked at this match he would give Matthews an excellent chance of winning.

Unfortunately like the polls the record of Matthews is suspect. The polls that he is basing his data on does not accurately reflect the electorate, either nationally or in the states as I’ve already discussed. Harry Kid Matthews was the same as author Everett M. Skehan put it in his biography of Marciano on page 191:

He had had only one bout in New York, when he won a ten-round decision over Irish Bob Murphy in 1951. A few weeks before the fight, the word in the fight crowd was that Hurley had nurtured Matthews with a steady diet of stiffs , and that Marciano would take him apart.

The word was true, Maricano knocked him out in the second round, as Skehan put it one page later

Rocky was unmarked, unhurt and not even slightly tired. The heavyweight contender Matthews had been easier to defeat than most of the mediocre club fighters Rocky had met early in his career at Providence.

And that is Barack Obama, a paper tighter with a paper record. Nate Silver may believe that Barack Obama is riding the greatest wave of his electoral possibilities but rather than looking at the record. Rather than looking at the fact that every single one of these polls show the public thinking the country is on the wrong track, rather than looking at the registration trends over the last 9 years as I did, rather than noticing that even pollsters don’t say their model is likely, he is looking at data that is at best suspect and insisting he is right with Chaping like dedication

Yes, of course: most of the arguments that the polls are necessarily biased against Mr. Romney reflect little more than wishful thinking.

and on Tuesday he will discover: Elephants aren’t pink.

A final thought. The real problem with Silver’s ascertains the left. Because of this assertion of Mr. Silver the Camping cultists of the left that will not even look at a source of information that doesn’t pass their test of ideological purity will see Gov Romney’s victory as “confirming” that republicans somehow “cheated” while the actual evidence produced so far shows exactly the opposite.

We already see this type of thinking in the comments at the Times.

Mitt Romney is going to win this election and it’s not going to be close, but thanks to the flawed models promulgated by Nate Silver this clear-cut results is going to lead to have consequences far beyond those of losing an election. It will mean that once again the left will not question the policies and their decisions that turned off the public.

I’m sure they will be grateful, as I’ve said liberalism is a religion and like the Camping cult it can’t bear the thought of being proven wrong.

Update: Stacy McCain does the math:

His headline summarizes exactly what Republican poll-mongers have been saying since September, as they see poll after poll with crazy oversamples of Democrats. As for example, the NBC/WSJ/Marist poll that has Obama winning Ohio by 6 points — SIX FREAKING POINTS! — about which Ed Morrissey says, “all you need to know is this: the D/R/I is 38/29/32. In 2008, the exit polls showed a split of 39/31/30, and in 2010 36/37/28.”

Stop for a second and think about that: What this poll is telling us is that partisan ID has shifted 2 points toward Democrats since 2008, which was the best year for Democrats since LBJ won a landslide in ’64. Therefore, we must choose between two alternative explanations:

Obama is headed toward a world-historic victory based upon the remarkable popularity of the Democratic Party; or
The poll sample is fucked-up beyond all comprehension.

Take your pick, eh? When the sample is D+9 and yet Obama is winning by only 6 points, why should anyone doubt that the NBC/WSJ/Marist lead is entirely a function of the Democrat oversample?

You would think a baseball guy would see that statistical illusion just like why outfielders had more putouts when Robin Roberts was on the mount, it wasn’t range (except for Ashburn) it was a flyout pitcher.

Buy Raspberry Ketone Here

American 023

From a Former Atheist:

From a Former Atheist:

Try the Double Burger!

nashoba

Annie’s Book Stop of Worcester

Annies Book Stop of Worcester 001

Find Discounts at the Stores you Love

TOP STORES

Listen to your Granny

RWG

Forest of Assassins

Forest of Assassins

DH Gate Dot Com, Online Shopping

ecigarette

Support our favorite Charties

Read me at Examiner.com

Examiner badge2

Only 114 Million Hits to retirement!

Most Innovative Blogger 2013

Most Innovative Blogger 2013

Tags

Help a Brother Knight of Mine who needs a hand