Democrats are now favored to retain control of the Senate when the new Congress convenes in January, according to the FiveThirtyEight forecast, breaking a summer stalemate during which control of the chamber appeared about equally likely to go either way.
An unusually large number of Senate races remain competitive, meaning that a wide range of outcomes are still possible. Republicans have about a 10 percent chance of winning a net of at least six seats from Democrats, according to the forecast, which would give them control of at least 53 seats next year. However, there is also about a 20 percent chance that Democrats could actually gain Senate seats on balance, giving them at least 54. The only thing that seems completely assured is that neither party will control enough seats next year to hold a filibuster-proof majority.
But the odds of a favorable overall outcome for Democrats have increased in recent weeks. The forecast model now gives them a 70 percent chance of controlling the chamber, either by having at least 50 seats and the presidency, or 51 without it.
Although this represents the first official FiveThirtyEight forecast for the Senate this year, I ran backdated forecasts to July 1 based on the polls that were available at that time. Two weeks ago, for example, the model would have given Democrats a 52 percent chance of retaining Senate control — and four weeks ago, it would have given them a 39 percent chance.
The trend toward Democrats is a relatively recent one. Part of the shift may reflect the bounce President Obama received from the Democratic convention. If so, it could recede, especially if Mr. Obama’s poll numbers do so, too.
But our analysis also suggests that the Democratic advantage has probably been building over the past few weeks, and may not have any one root cause. Instead, Republicans risk death by a thousand cuts, with a gradual deterioration in their standing in several important races, and their inability to field optimal candidates in others.
Democratic Seats Where a Republican Takeover Is Favored
There is, however, some good news for Republicans: they are still playing plenty of offense. In fact, they are favorites to pick up four seats currently controlled by Democrats, according to the model.
The Republicans’ easiest pickup opportunities are in Nebraska and North Dakota, where the Democratic senators Ben Nelson and Kent Conrad are retiring.
Democrats have capable candidates in each race, in the form of Bob Kerrey, the former governor and senator of Nebraska, and Heidi Heitkamp, the former attorney general of North Dakota.
However, the model considers several statistical factors in addition to the polls, all of which have some power to predict Senate races. These include the ideological positioning of the candidates relative to that of the state, the projected result in the presidential race there, the generic Congressional ballot (a measure of the overall disposition of voters to cast their ballot for Democrats or Republicans for Congress), the amount of money raised by each candidate and the candidates’ respective credentials, as measured by the highest elected office that they have held in the past.
In addition, for elected incumbent senators who are running for another term, the model considers their previous margin of victory relative to the overall partisan climate at that time. And it uses Congressional approval ratings — which are historically low right now — to gauge the overall degree of pro-incumbent or anti-incumbent sentiment. (A more complete description of the model’s methodology will be provided in a separate article.)
In North Dakota, the polls show only a modest lead for the Republican candidate, Representative Rick Berg. However, these fundamental factors favor him. In particular, North Dakota is a conservative state that Mr. Obama is likely to lose badly. And Mr. Berg has raised considerably more money than Ms. Heitkamp, with $2.6 million in individual contributions to her $1.3 million. When it combines the polls and these other factors, the model makes him about an 80 percent favorite.
The story is slightly different in Nebraska, where Mr. Kerrey trailed his Republican opponent, Deb Fischer, by double digits in the last polls of the state. The model expects that the polls are more likely than not to tighten some: Mr. Kerrey has raised about twice as much money as Ms. Fischer as of the last reporting period. And he is the more experienced candidate, with his former tenure as senator and governor. (Ms. Fischer is a member of Nebraska’s unicameral legislature.) Still, Nebraska’s red-leaning partisan gravity is probably too hard for him to overcome in the current political environment, and Mr. Kerrey is not a true moderate. Between Mr. Kerrey’s disadvantages and her lead in the polls, Ms. Fischer is about an 80 percent favorite.
Republicans have a 60 percent chance of winning in Montana, where the Democratic incumbent Jon Tester is up for re-election. Although the polls are close to being exactly tied there, the model gives the Republican, Representative Denny Rehberg, an edge on the basis of the fundamentals. Mr. Rehberg is a strong candidate, with streaks of moderation in his voting record, which could play well to center-right Montana. And Mr. Tester won his election by an extremely slim margin in 2006 — during a year where the overall partisan climate strongly favored Democrats — which often bodes trouble for re-election.
Republicans also rate as very narrow favorites in Wisconsin, where the model estimates that they have about a 55 percent chance of winning. Here, the nonpolling factors have been more equivocal. The Democratic candidate, Representative Tammy Baldwin of Madison, rates as very liberal, which could hurt her even in a slightly blue-leaning state. However, she has raised considerably more money than her opponent, Tommy Thompson, the former Wisconsin governor and health and human services secretary under President George W. Bush. The polls have been tricky to read: the candidates were roughly tied in the polls before Mr. Thompson won the Republican primary and pulled ahead, but a Public Policy Polling survey published on Monday showed Ms. Baldwin in back in the lead. The Public Policy Polling survey was conducted on behalf of Democracy for America, a liberal group, but I’d like to see its result confirmed by a nonpartisan pollster before concluding that the composition of the race has changed again. For now, Ms. Baldwin looks like a slight underdog.
Democratic Seats Where a Republican Takeover Is Possible
The next best pickup opportunity for Republicans is in Virginia, where Jim Webb, the Democratic senator, is retiring. Both parties have well-qualified candidates: for the Democrats, the former Gov. Tim Kaine, and for the Republicans, the former Senator George Allen. But the model gives Mr. Kaine a very slight edge in the race — assigning him about a 60 percent chance of victory, partly on the basis of two new polls, published on Monday, that gave him leads of one and two percentage points. A two-point lead is what counts as a noteworthy result in a race where many polls have shown the candidates tied — but this should be thought of as a tossup for all intents
After that, however, Republican pickup opportunities start to look more problematic. In Missouri, their candidate, Representative Todd Akin, considerably damaged himself with his comments about rape — and also has some other hallmarks of being a poor candidate, including sluggish fund-raising and a Congressional voting record that rates as being very far from the center.
Although the polls of the state have been varied since Mr. Akin’s remarks, some show him still remaining competitive or even tied in the race, and the model gives him about a 40 percent chance of victory. Perhaps that’s a slight overestimate: there’s empirical evidence that remarks of that nature can cost a candidate about 10 points in the polls in the end, but the model does not account for them directly.
The fundamentals for the Democratic incumbent, Claire McCaskill, are also quite weak, in that she barely won her race in 2006 and is running for re-election after having supported significant parts of Mr. Obama’s agenda in a red-leaning state where he is not popular. In a sense, this is a race that neither candidate ought to win.
One race where Republicans have favorable momentum is Connecticut, where their candidate, Linda E. McMahon, the former chief executive of World Wrestling Entertainment, has drawn a near-tie in the polls with the Democrat, Representative Chris Murphy. However, Ms. McMahon faded down the stretch as the Republican nominee in 2010, and by the model’s calculations, that could happen again, as she has never held elected office and Connecticut is a solidly blue-leaning state. The model gives her about a 25 percent chance of a win.
Another group of states, however, are more clearly disappointing for Republicans. Foremost among these are Ohio and Florida, where the polls have shifted sharply toward the Democratic incumbents, Sherrod Brown and Bill Nelson, in recent weeks.
Mr. Brown and Mr. Nelson each lead by seven to eight points in the average of recent polls, an advantage that is much more difficult to overcome with 50 days left in the race. Historically, candidates trailing by that margin come back to win the race only about 10 percent of the time.
It is the Republican candidate in Ohio, Josh Mandel, the state treasurer, who probably has a better chance of beating those odds. He has raised a considerable amount of money, and Mr. Brown’s positions are quite left-leaning for moderate Ohio. On the other hand, Mr. Mandel’s ideology rates as being strongly right-leaning — this race features perhaps the largest ideological gulf of any major contest this year — and Mr. Brown has also raised a significant amount of money.
In Florida, however, it is harder to see the Republican case for victory. Their candidate, Representative Connie Mack, has a strong family name, but he has raised only about $3 million, an inadequate sum in a state as large as Florida. And Mr. Nelson won re-election easily in 2006 and his approval ratings remain decent. This race, like Ohio, is one that Republicans might win in a wave election year like 2010, but it will probably not come together for them this year.
Republicans have strong candidates in two open-seat races, in the form of former Gov. Linda Lingle of Hawaii and former Representative Heather Wilson of New Mexico. Both rate as being more moderate than most Republicans who are running this year. Ms. Wilson has raised about as much money as her Democratic opponent, Representative Martin Heinrich, while Ms. Lingle has raised more than hers, Representative Mazie K. Hirono.
But the Democrats, Mr. Heinrich and Ms. Hirono, have built up leads in the polls, and they have the partisan tide at their back, since New Mexico has become quite blue-leaning in recent years, and Hawaii will be overwhelmingly so with its native son, Mr. Obama, on the ballot. The chances of a Republican pickup in each state is now only about 10 percent, according to the forecast.
In other cases, Republicans have a candidate who either campaigned poorly, like Representative Peter Hoekstra of Michigan, or a candidate who has underwhelming credentials, like the businessman Tom Smith in Pennsylvania. The Democratic incumbents in those states, Senators Debbie Stabenow of Michigan and Bob Casey of Pennsylvania, are clear favorites, each with about a 95 percent chance of keeping their seats.
Republicans also fielded weak candidates in Washington, West Virginia, and California. Instead, their long-shot hopes might rest on New Jersey, where the approval ratings are tepid for the Democratic incumbent, Senator Robert Menendez, and where Republicans have a relatively moderate candidate in Joe Kyrillos, a state senator. However, Mr. Kyrillos has yet to make much of a move forward in the polls. A true long shot might be Rhode Island, where the Democratic incumbent Sheldon Whitehouse’s fund-raising has been poor and where Republicans have an interesting and unorthodox candidate in the libertarian-leaning Barry Hinckley. But Rhode Island is strongly Democratic and Mr. Hinckley has received little support from the national party.
Republican Seats Where a Democratic Takeover Is Favored
One problem Republicans are having is simply that Mr. Obama is now favored to win the presidency, meaning that Democrats would control the Senate in the event of a 50-50 tie. The Senate model incorporates the current FiveThirtyEight presidential forecast in order to resolve these cases, which now gives Mr. Obama a 75 percent chance to win another term.
Still, even if Republicans required four Democratic pickups rather than three, their odds would not be too bad. As I mentioned, the model has Republicans favored in four Democratic-held seats, and they have a decent opportunity in a few others, even if other states like Florida seem to be increasingly off the table.
However, Republicans are not purely playing offense. They are now likely to lose two of their own seats, and three others are quite vulnerable. Thus, they will probably need at least five Democratic pickups, and possibly as many as eight.
The seat that Republicans are most likely to lose is Maine, where an independent candidate, former Gov. Angus King, has an overwhelming lead in the polls and is a 95 percent favorite to win, according to the model. He would replace the retiring Republican incumbent, Olympia Snowe.
Mr. King has not said which party he would caucus with in the Senate. However, the statistical measures of his ideology that the model uses rates him as being reasonably liberal — more so than about one-third of current Democratic senators. And Mr. King has endorsed Mr. Obama for president.
Still, there is no hard and fast way to estimate the chances of Mr. King caucusing with either party. So last week, I asked my Twitter followers to provide their best guesses. The consensus estimate was that Mr. King had a 75 percent chance of caucusing with the Democrats, a 10 percent chance of doing so with the Republicans, and a 15 percent chance of caucusing with neither party, which would effectively be the same as his caucusing with the majority since it would give the control of the Senate to whichever party had a 5
0 to 49 lead in the other races.
The forecast model randomly assigns Mr. King a party in each simulation based on these probabilities, meaning that Maine is likely to be a de facto Democratic pickup.
A more straightforward case is Massachusetts. In this race, the momentum has shifted numerous times — but the latest shift is toward Elizabeth Warren, the Democratic candidate who has made gains in three polls published on Sunday and Monday to take the lead over the incumbent Republican senator, Scott P. Brown.
The model also estimates that the fundamentals slightly favor Ms. Warren. She has some weaknesses as a candidate — her ideology rates as being very liberal, even by Massachusetts standards, and she has never held elected office before. However, her fund-raising has been prodigious, as she has already brought in more than $27 million, and it is simply very difficult for a Republican to win in Massachusetts under most circumstances. The model lists Ms. Warren as about a 65 percent favorite. But this is a very rare case in today’s politics where both candidates are actually quite popular.
Republican Seats Where a Democratic Takeover Is Possible
A race of underrated importance is Indiana, which has been sparsely polled. But the polls that have been published have shown a near tie between the Democrat, Representative Joe Donnelly, and the Republican, Richard Mourdock, the state treasurer who defeated longtime incumbent Senator Richard G. Lugar in the primary earlier this year.
The model’s view is that a tie is a reasonable estimate of the state of the race. Mr. Donnelly’s voting record is quite moderate, whereas Mr. Mourdock’s policy views are very conservative. The difference is enough to counteract Indiana’s Republican lean and make the race close to a tossup, although the model gives Mr. Mourdock an incremental edge, with about a 55 percent chance of victory.
Republicans also have a 55 percent chance of retaining Nevada, where their candidate is the appointed incumbent Dean Heller. The model treats states with appointed incumbents as tantamount to open-seat races, as the historical evidence suggests that they do not enjoy much of the traditional incumbent advantage until they win an elected term for themselves. In Nevada, this may be evidenced by the fact that the Democrat, Representative Shelley Berkley, has raised more in individual contributions ($5.3 million) than Mr. Heller has ($3.6 million). And Nevada has increasingly become blue-leaning. Still, Ms. Berkley’s candidacy has been clouded by ethics accusations and she is at a slight deficit in the polls there.
The final realistic pickup opportunity for the Democrats is in Arizona, where polls have shown a tight race between their candidate, the former United States surgeon general, Richard H. Carmona, and the Republican, Representative Jeff Flake. In this case, however — unlike in Indiana — the model calculates that the fundamentals favor Mr. Flake, since he has raised about twice as much money and since Mr. Carmona has never held elected office before. Mr. Carmona has about a 25 percent chance of winning, according to the model.
The model also gives an outside chance for an upset, about 8 percent, to the Democrat in Texas, Paul Sadler, although that is mostly because his race against Ted Cruz, the Republican candidate, has not been polled adequately, meaning that the model treats the race as having more uncertainty. This upset is highly unlikely to come to fruition in practice unless there is a large and unexpected Democratic wave.
But while a large wave toward either party is hard to conceive of at this stage, Democrats do seem to have a gentle tide favoring them. They have gained ground on the generic Congressional ballot in recent weeks and now lead it by about two percentage points on a likely-voter basis, according to our estimates, meaning that they might have a slight edge in races that might otherwise be split 50-50. And with Mitt Romney having penetrated little into traditionally blue-leaning states, Democrats should have a clearer edge in those.
Between those factors, and subpar Republican candidates in some purple and red-leaning states, the Democrats are in the best shape they’ve been in all year in the Senate. The model now gives them the equivalent of a one-seat advantage, meaning that the most likely composition of the new Senate is 51 Democrats and 49 Republicans, although some seats are almost certain to change hands on both sides.
Republicans still have a multitude of paths toward 51 seats, but they risk being overextended, having poured party committee funds and independent expenditures into races like Florida and Ohio where the polls have shifted against them.
Their decision may be whether to wait out any potential effects from the convention bounce and see if Democratic momentum fades at all, or target a narrower set of races.
One sign that may indicate that Republicans recognize their predicament is if they quietly provide more financial backing to Mr. Akin, who still has a shot in Missouri. It isn’t an ideal circumstance for Republicans, but the battle for Senate control could run through the state. The model estimates that Missouri has an 8 percent chance of being the tipping-point state in the Senate, meaning that race determines the majority. It ranks third by this measure, just behind Virginia and Massachusetts.