News from the Votemaster

Barack Obama spoke to a large, adoring crowd in Berlin yesterday. His speech was long on vague generalities and short on policy specifics, especially on issues that separate America from Europe such as trade and defense. Nevertheless, he navigated a dangerous shoal well. It is taboo for politicians to criticize the current administration when on foreign soil and Obama said nothing about President Bush's policies. It is also considered poor form to hold a political rally in a foreign capital (unless it is clearly targeted at getting the votes of the 7 million Americans abroad). His speech wasn't political at all. He didn't mention John McCain even once and never said: "If I am elected I will ..." Instead he emphasized that today's problems, such as global warming, terrorism, AIDS, and poverty cannot be solved by any one nation, no matter how large or powerful. What is needed, he emphasized, is for America to work with its allies to tackle these challenges. While he didn't draw any specific contrasts, it is inconceivable that President Bush or Vice President Cheney could have given this speech with its focus on cooperation rather than their usual: "We lead, you follow." One thing Obama didn't do is a supply a nice zinger for the history books like John Kennedy's "Ich bin ein Berliner" or Ronald Reagan's "Mr. Gorbachev, tear down this wall." For Obama, the real goal was to try to convince Americans that he is "presidential." We'll know in a week.

John McCain visited a German restaurant yesterday. There isn't much he can do to get into the news until Obama comes home. Obama will be in France today so McCain is also trying to do something French: he is meeting with seven-time Tour de France winner and cancer survivor Lance Armstrong. McCain joked: "In a scene Lance would recognize, a throng of adoring fans awaits Senator Obama in Paris -- and that's just the American press." McCain is much better than Obama at clever one liners, but he has to be careful because sometimes they are politically incorrect and get him into trouble.

While same-sex marriage is a hot-button issue and frequently put on the ballot by Republican-leaning groups as a way of increasing turnout of like-minded people, practically every state has already had a referendum on this already so there is no gold left to mine here. But a new issue is arising: same-sex divorce. Anyone who never considered this possibility is very naive. Only half of all opposite-sex couples manage to stay together. Why should the odds be better for same-sex couples? While a few states and countries allow (same-sex) couples to come in, get married, and scoot out, few of them allow quickie divorces by out-of-state couples. Will we soon see ballot initiatives banning same-sex divorce? The trouble is that people who rally around "family values" don't like either same-sex marriage or divorce. Is it better for a same-sex couple to stay married (bad) or get divorced (bad)? Maybe the best way to look at this issue is as a business opportunity for some enterprising state.

CQ Politics has changed its ratings on 14 congressional races. Twelve of the shifts favor the Democrats. Some of the more high-profile races are as follows. In CT-04, Chris Shays' race is now rated as a tossup (was leans Republican). This race is an oddity with an established Republican politician facing a rich Democratic businessman, Jim Himes. Usually it is the other way around. Our take is that Shays, the only Republican congressman in all of New England, is likely to go down to defeat.

Another hot race is OR-05, where Democrat Darlene Hooley is retiring in a swing district. At first the Republicans looked like that had a strong challenger in rich businessman and anti-abortion foe Mike Erickson. But then it came out that Erickson had impregnated his girlfriend a few years ago and drove her to an abortion clinic and paid for her abortion. Kind of weakens his anti-abortion credentials in the eyes of some folks. CQ Politics now rates the race as leans democratic (was: tossup). Our take: Erickson is dead meat and state senator Kurt Schrader (D) will succeed Hooley.

One race the Democrats really have to worry about is TX-22, Tom DeLay's old seat. Nick Lampson (D) won that district in 2006 because the Republicans didn't have anyone on the ballot and voters were told to write in the name of dermatologist Shelley Sekula-Gibbs (R)--"typing" with a trackball on a virtual keyboard with no hyphen on it. This time the GOP as an actual candidate on the ballot, veteran Pete Olson. The district is R+15. CQ Politics has changed the rating from leans Democratic to tossup.

There was an Apache problem with the Firefox plugin that displays the electoral college score at the bottom of the browser (in the status bar). Fixed. To get it, visit www.electoral-vote.com/evp2008/Info/electoralcollege@electoral-vote.com.xpi and follow the instructions. This link is also on the Welcome page in case you lose it. If you install it and later decide you don't want it, go to Tools > Add Ons to delete it. If you aren't using Firefox, go get it at mozilla.com. It is a great browser, easy to install, and free. About 36% of the visitors to this site use Firefox and the percentage is growing.

Lots of polls today. Four new polls from Quinnipiac University show John McCain playing catchup. He is even leading in Colorado now by 2%, 46% to 44%. However, Rasmussen had a poll Wednesday showing Obama ahead 49% to 42%. As described on the Map algorithm explained page, all polls taken within a week of the most recent one are averaged weighted equally. Some people have suggested weighting older polls less, but in this case, the Quinnipiac poll was taken July 14-22 whereas the Rasmussen poll was done on July 21 only. Which is older and by how much? Our algorithm uses the midpoint for all polls, so we treat the Quinnipiac poll as having been taken on July 18, so it is within a week of the "newer" Rasmussen poll.

Many people have suggested weighting functions for old polls, but it becomes very arbitrary very quickly. Should a week-old poll be weighted at 0.5? Or maybe 0.4 or 0.3? How much is a three-week old poll worth compared to a current one? The one thing quickly dissolves into a battle over the weighting function. For simplicity, the rule here is that the most recent poll wins, except any other polls taken within a week of it count equally. It is really hard to argue that public opinion in Colorado has changed so much between July 18 (Quinnipiac's midpoint) and July 21 (Rasmussen's one-day poll) that the Quinnipiac result should be partially discounted in favor of the newer Rasmussen result.

If you follow other election prediction sites, you will notice differences with this one. One of them is how much respect old polls get and how they are weighted. Some sites average the last 3 or 5 polls, which means that if the third most recent poll is from February, it gets counted, which strikes us as a bad idea. Another key difference is that we don't count partisan pollsters like PPP (D) or Strategic Vision (R). They can't be trusted in general. Some sites believe them.

None of the other polls are very surprising, although North Dakota and Mississippi are perhaps a bit closer than anyone expected early on.

Quinnipiac University also polled two Senate races. The Colorado poll shows it to be a tied, whereas just about every other poll this year has shown Rep. Mark Udall (D-CO) to be ahead of former representative Bob Schaffer (R). In Minnesota, Quinnipiac has Sen. Norm Coleman (R-MN) way ahead of comedian Al Franken (D). yesterday we had a Rasmussen poll with Coleman ahead by 1%. What's going on here? In virtually all its polls today, Quinnipiac has results more favorable to the Republicans than the other pollsters. It could be that it has normalized the results to a certain assumed partisan distribution and that it is assuming more Republicans than the other pollsters. The underlying model wasn't released.