Tuesday, October 28, 2008

The Only Way McCain Wins (and how to stop him)

The Current Polls & Predictions
With one week to go, here's an overview of my various sources and what they say about the upcoming election. Everything points to a significant Obama victory next week, unless something very odd or untoward happens.

Pollster.com EV map:

Image Source: Pollster.com, 2008.10.27

Pollster.com basically takes all the poll data from all the polls and displays it as an EV map. Yellow states are "tossups", while lightly colored (blue and red) states "lean" to their respective parties. The critical piece of information here is the number in the upper right next to the dark blue legend square: 268. Current polling shows Obama has "strong" polling numbers representing 268 EVs, leaving him only two short of election to pick up over the the "lean democratic" and "tossup" states, while McCain must run the table of all "lean republican", "tossup", AND "lean democratic" states to win.

Electoral-vote.com map:

Image Source: Electoral-Vote.com, 2008.10.27

Electoral-vote.com shows us basically the same data in a slightly different form. Pure white states (MT and ND) are "tied". Light blue and red states (CO, VA, OH, SD, WV) are "weak" for their respective parties while light blue and red outlined states (FL, NV, NC, AZ, GA) are "barely" for their respective parties. You'll notice that the outlined and white states are the "tossup" states in the pollster.com map. Behind the scenes these sites use very similar data, and the conclusion is almost the same. Electoral-vote.com says Obama has 260 EVs locked up in "strong" states, with another 57 "weak" and 58 "barely" to McCain's 134 "strong", 8 "weak" and 15 "barely", with 6 EVs "tied" in MT and ND. Again, Obama needs to pick up very few EVs (10) from the entire range of "weak democratic", "barely democratic", "tied", "barely republican" and "weak republican" in order to win.

Five-thirty-eight.com EV map:

Image Source: FiveThirtyEight.com, 2008.10.27

Five thirty eight.com is not just reporting aggregated polling data. The FAQ explains how they are different:

Firstly, we assign each poll a weighting based on that pollster's historical track record, the poll's sample size, and the recentness of the poll. More reliable polls are weighted more heavily in our averages.

Secondly, we include a regression estimate based on the demographics in each state among our 'polls', which helps to account for outlier polls and to keep the polling in its proper context.

Thirdly, we use an inferential process to compute a rolling trendline that allows us to adjust results in states that have not been polled recently and make them ‘current’.

Fourthly, we simulate the election 10,000 times for each site update in order to provide a probabilistic assessment of electoral outcomes based on a historical analysis of polling data since 1952. The simulation further accounts for the fact that similar states are likely to move together, e.g. future polling movement in states like Michigan and Ohio, or North and South Carolina, is likely to be in the same direction.

Five thirty eight.com uses statistical techniques in an attempt to determine which polls are more valid than others, and to extract trend information. Using the processed data, they run Monte Carlo simulations to "predict" the outcome of 10000 simulated elections. The aggregated information from those simulation runs provides them with a "most likely" result, a probability of Democratic or Republican victory, and an estimate of the popular vote. As of 27 October, the most likely result is 351 EV for Obama and 187 EV for McCain, a 96.7% chance that Obama will win, and a popular vote split of 52.4% for Obama to 46% for McCain.

Electoral-vote.com, all EV polling:

Image Source: Electoral-Vote.com, 2008.10.27

Electoral-vote.com also shows us two graphs. This lead graph provides the total EV count polling to each party. Obama has polled above 270 EVs very consistently since he clinched the nomination.

Electoral-vote.com, EVs polling more than 5% only:

Image Source: Electoral-Vote.com, 2008.10.27

This second graph shows only those EVs which poll a 5% or greater difference. Even with that, Obama is above the 270 needed to win the presidency.

Pollster.com gives us a national trend graph showing all their polls as individual dots with a line showing a smoothed average trend. I've included three versions to show just how tricky this kind of analysis can be. The sensitivity of the smoothing indicates how quickly the trendline responds to changes. More sensitive means more quickly. However, more sensitive also means more vulnerable to noise in the data.

Pollster.com More Sensitive Smoothing

Image Source: Pollster.com, 2008.10.27

Pollster.com Medium Smoothing
Image Source: Pollster.com, 2008.10.27

Pollster.com Less Sensitive Smoothing
Image Source: Pollster.com, 2008.10.27

Notice how the trendlines are more jagged in the "more sensitive" graph and how they smooth out as the sensitivity goes down. The "medium" and "less sensitive" graphs show us a pretty consistent picture of Obama running about 50% and McCain somewhere in the 42-43% range. The "more sensitive" graph suggests a hint that Obama's support is weakening and that we might be seeing a significant narrowing (perhaps as much as 3%) of the national gap.

That's not very likely, especially not this close to the election. The gap changes we see now are most likely to be undecideds (although how you can be undecided 19 1/2 months into a 20 month campaign is a little beyond me) making up their minds at the last minute, NOT people suddenly deciding that the candidate they're with has become unacceptable to them. If the change continues over the next couple of days, it may represent an actual shift, but I wouldn't call it a real trend until Thursday and then only if it starts to show up in the less sensitive trend analysis. If I had the actual numbers to play with I might be able to call it sooner, but I don't.

With all the polling and trending suggesting bad national news for McCain, how does he win the election? Watch the video for the answer.

Source Video: Heroes, Season One, Episode "Landslide"
Video Production: Evan Robinson, Group News Blog

Obviously I'm not suggesting that John McCain can win the election by finding an 8-year old African-American child with the power to "talk" to computers. But I am suggesting that the most believable way McCain wins the presidency in 2008 is through election fraud.

Voting Fraud
According to the Brad Blog, we are already seeing "vote flipping" happening during early voting in West Virginia and Texas. Candidate names disappeared from the final review screen in South Carolina. We've even seen at least one case of vote flipping from R to D in Tennessee (although that report appears to be exaggerated -- at least)!

Ultimately, computer voting using machines created by private companies expose our democracy to unacceptable risks. Companies making electronic voting machines have committed to deliver votes for one side. Unvetted software has been installed to run elections. Computers count votes in secret and many make meaningful re-counts impossible.

Vote Suppression
Every election cycle, Republicans launch massive efforts at suppressing voter turnout. They do so by a variety of methods, including not mailing out ballots, circulating flyers with false information about election dates or locations or requirements, purging voter rolls, issuing spurious challenges to voters on election day, even threatening voters with deportation or arrest.

It has been part of the conservative (aka Republican) canon since at least 1980 that low turnout benefits the Right:

How To Stop McCain From Winning

The only way John McCain wins in 2008 is to cheat. Don't let him. We must win by an overwhelming majority in order to take the election. No 2000-style sportsmanship in 2008 if the election is stolen.