Republicans in Georgia have been trying to attach Michelle Nunn to unpopular Democrats, mostly because she has never been in politics before so the oppo researchers came up more-or-less empty. In particular, they have been accusing her of being too close to President Obama and Senate majority leader Harry Reid. Nunn is apparently worried about these attacks so she said she might not vote for Reid for majority leader. The idea is to convince voters that she is not like a Washington Democrat. Senate elections are secret, so nobody will ultimately know how she voted, but the possibility of her voting for someone other than Reid in an uncontested election for majority or minority leader is ridiculous. If one of the 16 Democratic women in the Senate were to challenge Reid, then her statement makes sense, but there is little reason to believe this might happen.
It is common that campaigns attack candidates by trying to attach them to well-known but unpopular figures in their party, saying "Smith is just like Jones." To the extent that the country is getting more polarized, this strategy is getting more common. Sometimes candidates are tied not to a specific national politician, but to a locally unpopular policy of the national party. For example, Republicans have attacked Democratic Senate candidates Natalie Tennant in West Virginia and Alison Lundergan Grimes in Kentucky as being anticoal, even though the candidates themselves have all but said that they heat their homes with coal.
The Export-Import Bank is an obscure federal agency established by President Franklin Roosevelt to make loans to foreign companies that want to buy U.S. products buy don't have the credit to borrow money from commercial banks. It has functioned quietly for 80 years helping companies like Boeing, Caterpillar, and General Electric sell their products abroad. All of a sudden the tea party has made it the poster child for government meddling in the economy. Its charter needs to be renewed now and many rigid conservatives want to kill it, even though big business and the Chamber of Commerce strongly support it. Consequently it has become a flash point in states whose manufacturers made heavy use of it. In North Carolina, for example, Sen. Kay Hagan (D-NC) strongly supports the bank and is attacking her opponent, Thom Tillis, as someone who doesn't care about the $3 billion in sales it has brought to her state and all the jobs created by those sales. It puts Tillis in the awkward position of supporting ideology against jobs. In Kentucky it is also an issue, with Mitch McConnell on record as wanting to kill the bank. It is also affecting the race in Louisiana and other states.
Sen. Bernie Sanders (I-VT) is planning to visit New Hampshire and South Carolina, two early primary states, instantly leading to speculation that the only member of Congress to call himself a Socialist is thinking about challenging Hillary Clinton for the Democratic presidential nomination in 2016. Unlike fellow Vermonter Howard Dean who ran for President in 2004 with the intention of winning, Sanders knows very well that he has no chance (unless Clinton declines to run). Nevertheless, a primary challenge could be useful to him as it might nudge her a bit to the left by forcing her to talk about Dodd-Frank and the regulation of big banks. Ironically, it could also be useful to Clinton as it would give her practice debating on the national stage and give her an excuse to travel around the country in full campaign mode, nominally against Sanders, but really against the still-unknown Republican nominee. It would also allow her to position herself to the right of Sanders on some issues and later, when the Republicans call her a Socialist, to be able to say: "I ran against an actual Socialist and opposed him on issues X, Y, and Z." Sanders is technically an independent but caucuses with the Democrats. To run in the Democratic primaries, he would have to switch to become an actual Democrat. Now that Sen. Elizabeth Warren, the darling of the left, has said about 100 times that she has no intention of running for President (certainly not against Clinton), people who think Clinton is too cozy with Wall Street are beginning to look closely at Sanders as an alternative.
In 1936, the magazine Literary Digest ran a poll sampling an astronomical 10 million people, of whom 2.4 million responded. It predicted Republican Alfred Landon would crush Franklin Delano Roosevelt. Landon got 37% of the vote and won only Maine and Vermont with a total of 8 electoral votes. The magazine became a laughingstock and soon folded. In retrospect, the problem was sampling. They got the names of the people to poll from their list of subscribers, lists of people who had registered a car, and lists of people who had telephones. In 1936, at the height of Great Depression, that meant they only sampled rich people, and indeed rich people overwhelmingly preferred Landon. Since then, the gold standard for all polling has been random sampling, for example, just picking the last four digits of a telephone number at random and calling it. Starting around 1950, that scheme worked fine for at least 5 decades. Now it doesn't work any more.
There are multiple problems. First, the model assumes everyone has one telephone. That used to be true but isn't any more. Some people have multiple telephones and some have none. Also, polling with live interviewers has gotten to be expensive (> $10,000) so media outlets have turned to cheaper pollsters like SurveyUSA, Rasmussen, and PPP that have computers dial the numbers and ask the questions. However, federal law prohibits the automated dialing of cell phone numbers. With more than 30% of the population no longer having a landline, automated pollsters are missing a sizable chunk of the population, and a chunk that is younger, poorer, busier, and more Democratic than landliners. Furthermore, response rates have dropped to about 10% because people are too busy to answer 10 questions. If nothing is done, pretty soon pollsters will be talking to only elderly widows with nothing to do all day.
There is some innovation in the polling business, but many people are skeptical of it. For example, If SurveyUSA cannot get an answer on the number it called, it tries to locate the callee on the Internet and sends him or her a link to the questions online. Picking people to interview at random using random digit dialing and then contacting them some other way (email, in person, etc.) preserves the random character of the sample.
Other firms are going further and allowing people to sign up online to be polled and then constructing an electorate that matches the census data. So if only 20% of the people in the online survey are women, each one counts for 2.5, to get the weight up to 50%. Similar adjustments can be made for age, income, race, partisanship, and other things. Nevertheless, this method is yet to be proven as accurate as live caller polling. It is a real problem.
Nate Silver has written a long piece on the subject.
While there have been polls of the Alaska Senate race between Sen. Mark Begich (D-AK) and Dan Sullivan all year, today we have the first poll taken after Sullivan has become the actual nominee rather than a hypothetical nominee. The poll is from Rasmussen and puts Sullivan ahead by 2 points. Historically, Rasmussen has had a bias of several points in favor of the Republicans, so we should wait for confirmation before drawing too many conclusions. Another poll is from Minnesota, where counting the ballots here in 2008 went almost into the summer. Ultimately Franken won by 312 votes. This year he will coast easily to reelection.
| State | Democrat | D % | Republican | R % | I | I % | Start | End | Pollster |
| Alaska | Mark Begich* | 45% | Dan Sullivan | 47% | Aug 20 | Aug 21 | Rasmussen | ||
| Minnesota | Al Franken* | 51% | Mike McFadden | 42% | Aug 19 | Aug 21 | SurveyUSA |