As we pointed out in March, there are two different approaches to predicting election outcomes: model-based and data-based approaches. The former looks at fundamentals, including the parties' respective approval ratings, generic congressional polls, how may people self-identify with each party, and so on, and comes to a probabalistic estimate of the election results. The second approach just uses current polling data and assumes it takes into account all those factors. This site is data based but others use models. We now have the odd situation that two of the better known modeling sites are far apart in their estimates of the probability the Democrats will hold the Senate. The New York Times model gives the Democrats only a 35% chance of holding onto a Senate majority whereas Sam Wang at the Princeton Election Consortium says it is 70%. That is a rather large difference. How can this be?
The model-based predictors all use different factors in their calculations. Does the historical tendency of the President's party to lose seats in Congress in the sixth year of his term matter? How much? Does the unemployment rate matter? How much? Does the President's approval rating matter? How much? It is possible to collect dozens of factors, assign each one a weight, and run simulations of previous elections to get the parameters that best predict the average election going back two, three, five, or eight cycles, but how much does that say about this one? Using different factors and weights gives very different results, as is apparent today.
Strictly data-driven approaches (like we use) also have methodological issues. Do polls from blatantly partisan pollsters count? How about polls from small colleges that have never done a poll before but would like to be in the news? Should you just use the most recent poll, the most recent three polls, all the polls this week, or something else? Each site has its own secret sauce. Except that ours isn't secret. It is clearly stated here. It was determined by some trial and error in earlier years and has worked pretty well. Here is the track record.
In our view, talking about probabilities only make sense when applied to a testable hypothesis. If a weather forecaster says there is a 70% chance of rain tomorrow, one can go back for a year or more and collect all the days the prediction was 70% and then look at the next day to see if it indeed rained on 70% of those days. In theory, this can be done with the final prediction on election day and then go back many years to compare the prediction to the results but there isn't much data available. In any event, we will stick with looking at real data and leave models to others.
In the early days of the Republic, senators got on their horses or in their carriages, trotted off to D.C. to pass a few laws, and went home. Nowadays, being a senator is a full-time job and all senators de facto live in D.C. or the Virginia or Maryland suburbs around it. Almost none of them except the senators from Virginia and Maryland live in their home states, although when Vice President Joe Biden was a senator he commuted to Delaware. Senators from the East Coast with enough money sometimes have a house in their home state that they sometimes go to on the weekend, but the reality is that all of them live in or near D.C.
Nevertheless, the issue of where a senator lives has come up a number of times in recent years. This year, senators Pat Roberts (R-KS) and Mary Landrieu (D-LA) have been accused of not living in their respective home states and it has become a campaign issue. In 2012, former senator Richard Lugar was also accused of not living in Indiana, the state he represented in the Senate. Senators who are not wealthy often don't see the point of buying and maintaining a home they never use--until it becomes a campaign issue.
While we are not tracking the governors' races in detail, two of them have enough national and future implications to mention from time to time. One is in Florida, the mother of all swing states, where Gov. Rick Scott (R-FL) and former governor Charlie Crist, now a Democrat, are in a neck-and-neck race. The other, which gets less attention but is probably more important, is in Wisconsin. Gov. Scott Walker (R-WI) has been a dream governor for conservatives. He signed laws that crippled public employees unions so badly that they petitioned for and got a recall election, which Walker won. However, there are many people who felt he was elected governor properly and thus entitled to serve out his term but that doesn't mean he should get a second term, which he is now attempting. However, the latest poll, from Marquette University, shows him trailing Mary Burke (D), an executive at a bicycle manufacturer, 49% to 47%.
If Walker wins reelection and also survives a scandal he is now embroiled in, he will probably run for President in 2016. Also, other Republican governors are likely to imitate his aggressive anti-union tactics. On the other hand, if he loses to a political newbie, other governors are likely to see such tactics as dangerous and be more cautious in pushing controversial items on the conservatives' wish list. This is a race to keep a close eye on.
Every since the Supreme Court ruled that super PACs could spend unlimited money to support candidates provided they didn't talk to each other, the super PACs and candidates have been looking for ways to communicate without breaking the law. A commonly used technique is for the super PAC to announce its advertising plans to the media in advance. The candidate can then plan accordingly. For example, in the Colorado Senate race, when the Koch brothers super PAC, Americans for Prosperity, announced it would spend $970,000 over a three-week period to attack Sen. Mark Udall (R-CO), the Republican candidate, Rep. Cory Gardner (R-CO), knew that he wouldn't have to spend any money in that period on advertising and could devote his time to fundraising and his ground campaign.
None of the potential 2016 Republican presidential nominees has any foreign policy experience, so the Republicans are a little worried that if Hillary Clinton is the Democratic nominee when the topic turns to foreign policy (and there is usually one presidential debate on the subject), her four years as secretary of state will make her look strong. To weaken her, the committee investigating the incident at Benghazi, where the America ambassador and three other Americas were killed, is planning to continue its work and keep the investigation in the news for 2 more years.
Historically, politics ended at the water's edge. For example, when 241 Americans were killed in Lebanon in the Beirut barracks bombing in Oct. 1983, Democrats didn't use that incident in their campaign to unseat then-President Ronald Reagan in the 1984 election. Using an incident in which Americans were killed overseas for political ends would have been considered unpatriotic then. Now, anything goes. Whether the voters will be interested in the story in 2016 is another matter, but Rep. Trey Gowdy (R-SC), who is chairman of the House select committee on Benghazi, feels it is worth trying. Nevertheless he is taking a risk because there is no evidence that Clinton had anything to do with the incident and she is surely going to attack the Republicans for exploiting the deaths of four Americans for political purposes.
A variety of factors, not the least of them intense gerrymandering, has split Congress into two groups, one from blue America and one from red America. They differ hugely in their demographics (blue is young, urban, and multicultural; red is older, rural, and unicultural) and agree on practically nothing. McClatchy has an interesting piece on how this is happening.
The U.S. has a very low election turnout rate, with barely 60% of the eligible voters actually casting a ballot in presidential elections and even fewer in other elections. From time to time somebody suggests Internet voting as the solution. Mark Ambinder has written a good piece describing why this is a terrible idea. The biggest objection is security: it might be possible for highly motivated people to hack an election. Witness how often banks and stores and other companies are hacked, despite their best efforts to prevent intrusions. Can you imagine the social consequences of having an election in which half the country doesn't believe the reported results? A secondary objection is that not everyone has a computer, giving an advantage to computer-literate over the computer-illiterate voters.
| State | Democrat | D % | Republican | R % | I | I % | Start | End | Pollster |
| Michigan | Gary Peters | 45% | Terri Land | 39% | Aug 22 | Aug 25 | EPIC MRA |