From Claiborne Booker - WSJ Article on Election Modeling

         
Greetings, all --
 
Apparently, some were unable to connect to the  WSJ.com link, which is
normally available to non-subscribers for 7  days.  Here's the article below:
 
 
 
PAGE ONE
             
(http://online.wsj.com/article/0,,SB109874593196855291,00.html?mod=home_page_one_us#)  
(http://online.wsj.com/article/0,,SB109874593196855291,00.html?mod=home_page_one_us#)  
(http://online.wsj.com/article_print/0,,SB109874593196855291,00.html)  
(http://online.wsj.com/article_print/0,,SB109874593196855291,00.html)  
(http://online.wsj.com/article/0,,SB109874593196855291,00.html?mod=home_page_one_us#)            advertisement
For Math Whizzes,
The  Election Means
A Quadrillion Options
Close  Race Has Programmers
Predicting the Outcome;
'I'm Just Some  Geek'
By CHARLES  FORELLE
Staff Reporter of  THE WALL STREET JOURNAL
October 26, 2004
To prepare for next week's election, Lawrence N. Allen  taught himself the
Matlab statistical programming language and built a  database of 1,700 state
polls pulled off the Internet. His program runs a  "likelihood analysis" on 15
closely contested battleground states. It  takes 50 minutes to run on an old
computer he got in return for a bunch of  parts from a broken laptop.
The unemployed computer programmer in Oakland, Calif.,  identifies his
politics as "to the left of standard Democratic candidates"  and says he flirted
with voting for Ralph Nader in 2000 before opting for  libertarian Harry Browne.
His calculations, made on Oct. 20, give Mr. Bush  a 78.1% chance of victory.
Mr. Allen says he drew inspiration from Sam Wang, an  assistant professor of
molecular biology and neuroscience at Princeton  University, who devised a
computer program to analyze state polls and step  through all the possible
outcomes of 22 supposed battleground states.         ELECTORAL CALCULATION SITES


Sam Wang: _Electoral College  Meta-Analysis_ (http://election.princeton.edu/)
 
Andrea Moro: _2004  Presidential Electoral College Predictions_
(http://www.econ.umn.edu/~amoro/Research/presprobs.html)
Matthew  Hubbard: _Weekly  Electoral College Status_
(http://binomial.csuhayward.edu/WeeklyStatus.html)
Larry Allen: _Larry Allen's  Election Projection_
(http://www.arrowheadengineering.com/)  
John Denker: _Probability of  Winning the Electoral College Vote_
(http://www.av8n.com/politics/ec-prob.htm)  
ELECTION AT A  GLANCE



Get a _state-by-state  look_
(javascript:window.open('http://online.wsj.com/public/resources/documents/info-elecsnap04-frameset.html','elecsnap04','toolbar=
no,scrollbars=no,location=no,width=730,height=643,left=10,top=10');void('');)
 at the election and track the _latest  national polls_ (javascript:
window.open('http://online.wsj.com/public/resources/documents/info-natpolls04-frameset.
html','natpolls04','toolbar=no,scrollbars=no,location=no,width=790,height=623,
left=0,top=0');void('');) .  


There are 4,194,304 of them. (That's 2 to the 22nd power:  two possible
choices -- Bush or Kerry -- in 22 states.)
As of yesterday evening, Mr. Wang's "median outcome" was a  razor-thin
majority for Mr. Bush -- 279 votes in the decisive Electoral  College, versus Mr.
Kerry's 259, not counting undecided voters. But if the  results followed
historical patterns in which undecided voters generally  break for the challenger,
the Massachusetts senator would wind up with 307  electoral votes and the Oval
Office, Prof. Wang says, based on his  computations.
Messrs. Allen and Wang are among an elite cadre of  political amateurs
unleashing the tools of statistics and mathematics on  an extraordinarily close
presidential race.
Andrea Moro, an economics professor at the University of  Minnesota, uses a
type of simulation known as the Monte Carlo method to  calculate the probable
outcome. John Denker, a physicist and former  AT&T Corp. and Bell Labs
researcher, parses the Electoral College with  a gigantic Microsoft Excel spreadsheet
populated with reams of polling  data that the Founding Fathers couldn't have
imagined.
"Sometimes, brute force has an elegance all its own," says  Mr. Denker, who
describes himself as "polymathic mad scientist."
Mr. Wang says, "Electoral prognostication is just exploding  on the
Internet." He fastidiously updates data and posts comments to his  Web site,
_election.princeton.edu_ (http://election.princeton.edu/) . At 3:30  a.m., Oct. 22, he
had this to say about the big state known for its  hurricanes and hanging
chads: "If Bush wins Florida, his win probability  is 88%; if he loses, it's only
20%."
The Internet explosion was primed by the rise of blogging  and the ready
availability of state polls, manna for numerically inclined  political junkies.
Now, there is an abundance of Web sites dedicated to  drawing electoral maps in
red, for Republican, and blue, for Democrat.
But the statistical modelers contend that isn't enough to  go one-by-one
through the states and call them for Mr. Bush or Mr. Kerry.  That, they say,
misses substantial nuances that are greatly magnified by  the large number of
states in play. For example, a candidate polling slim  margins of victory in a
number of small states is less likely to win them  all than an opponent who has
larger leads in fewer, but larger, states.  The distinction is hard to represent
on a color-coded map; it is more  easily captured by statistical software.
So the new wave of amateur prognosticators employ a  technique widely used in
the physical sciences, known as likelihood  analysis or "probabilistic
modeling." The idea is to understand complex  events by breaking them into simpler,
discrete events and assessing the  probabilities of those events' actually
occurring. Physicists, for  instance, express the positions of subatomic
particles as probabilities.  Astronomers and cosmologists use likelihood analysis to
generate estimates  for quantities such as the age and expansion rate of the
universe.
Mr. Denker started building his model in August, after  watching a TV
commentator botch an explanation of probable outcomes. "I'm  watching the news and
slapping my forehead and saying, how can these guys  be so silly?"
Matthew Hubbard, a math lecturer who teaches at California  State University
in Hayward, notes that many pre-election polls four years  ago incorrectly
predicted that Mr. Bush would win the popular vote. That  led Mr. Hubbard to
believe the data weren't being processed properly by  the media. "We weren't
really getting what I thought was particularly  interesting or particularly good
information," he says. "So many people,  when they talk on TV and write in
newspapers, their mathematics are so  bad."
So this year, Mr. Hubbard, a computer programmer who once  wrote games for
Atari, created his own model, with its own _Web site_
(http://binomial.csuhayward.edu/WeeklyStatus.html) , that  chomps through 16.8 million possibilities in
the Electoral College in 72  seconds. His Oct. 23 prediction gave Mr. Kerry a
73.9% chance of reaching  the winning threshold of 270 electoral votes, with
Mr. Bush at 24.6%. He  rated the probability of a 269-269 tie at 1.6%.  
Mr. Hubbard, who is a Democrat, isn't placing his bets  quite yet, since the
numbers have been shifting over the past several  weeks. "I've played enough
backgammon and poker to know that you don't  celebrate too early," he says.
So it is with this agonizingly close election, where the  slightest tips and
swings have broad influence, and the combination of  possible results is vast.
Each of the models uses slightly different  polling databases and different
methods for assessing the errors in those  polls, but all follow a similar
scheme.
A statistical formula can transform one or more poll  results and margins of
error into a probability of victory. To take a  simple example that avoids the
complications of undecided voters and  independent candidates, assume that
Mr. Bush garners 55% and Mr. Kerry 45%  in a poll with a three percentage-point
margin of error. The wide gap and  low margin yield a probability of more than
99% that Mr. Bush wins. A poll  that's 51%-49% for Mr. Bush, with a
four-point margin, on the other hand,  would yield only a 68.5% probability of victory.
Armed with the probabilities of victory in each state, the  computers go to
work crunching the probability of each separate Electoral  College outcome. In
a two-party system, there are quadrillions of them.  Cutting the number of
states down to a dozen or two battlegrounds makes  the number more manageable --
between about 30,000 and 16 million. (That  ignores the fact that two or three
states may split their electoral  votes.)
The wide variation in the expected outcomes is due largely  to the
differences in selecting what polls to use, how to treat undecided  voters, and what
method is used to derive the statistical error in poll  results. The modelers
say, too, that they don't have a way to take into  account "external factors" not
relating to the sampling error in  poll-taking -- including turnout and
polling bias.
As a domain, politics is far messier than physics, says  Alan Abramowitz, a
professor of political science at Emory University and  an expert on polling.
The data are "subject to all kinds of error," he  says. "Trying to impose some
of these very high-powered methods may be  overwhelming the data." What's
more, state polls generally draw from a  smaller sample than national polls and
are as a result less reliable.
The race is so close and so difficult to assess that Mr.  Wang is split with
himself: His model predicts a very narrow Bush win when  he makes no
assumptions about the behavior of undecided voters, and a  larger Kerry victory if he
allocates more undecideds to the Massachusetts  senator.
When, last week, he switched his calculation to include the  Kerry-boosting
undecided formula, he got scores of protests from readers  of all political
persuasions who found the commingling of current survey  data with historical
trends to be an affront to the model's purity.
He was slightly taken aback by a flood of e-mailed  criticism. "I'm just some
geek posting numbers," he  says.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: /pipermail/friam_redfish.com/attachments/20041026/0a01eac6/attachment-0001.htm