Republican chances of taking over the Senate have improved again in this week’s forecast. They are now 22 percent — up from 18 percent last week and 15 percent two weeks ago. Republican chances are now approaching the point where they stood before the Delaware primary, when they had peaked at 26 percent before Christine O’Donnell’s victory.
There are six states held by Democrats where Republicans have at least a 75 percent chance of winning, according to the model. These are Arkansas, Colorado, Indiana, North Dakota, Pennsylvania and Wisconsin. Although some of these races — particularly Colorado and Wisconsin — could tighten, the G.O.P. has a fairly strong chance of winning all six if they are having a good night over all. (They will probably win Arkansas, Indiana and North Dakota even if they are having a disappointing night.)
Although earlier in the cycle, it had looked like some Republican-held seats were in play, that is less of a concern for them now. Instead, some races that had once seemed competitive — like Florida, New Hampshire, North Carolina and Ohio — have seen the Republican candidate gain ground in recent weeks.
One partial exception is Kentucky, where some polls show the Democrat, Jack Conway, moving into a somewhat stronger position. It is not that Republicans couldn’t lose Kentucky — Mr. Conway has been running some strong ads lately — but it is unlikely to be the state that prevents them from gaining a majority.
Instead, it is a set of seven states that are likely to determine Republican chances of controlling the Senate. These are California, Connecticut, Illinois, Nevada, Washington and West Virginia and the New York special election. Republicans would need to win four of these seven races to take claim of the Senate.
Illinois and Nevada are tossups, as they have been for months. Both races, of course, merit careful watching. That is particularly so for Illinois, where there remains such a large contingent of undecided voters — and voters telling pollsters they’ll vote for third-party candidates — that the race could break in any direction as these voters finally commit to either Alexi Giannoulias or Mark Kirk.
West Virginia is the next most likely state for a Republican pickup. Our model has not seen quite enough polling to declare the Republican, John Raese, the favorite there. But his chances of winning are up to 37 percent. (We discussed the West Virginia race at some length on Monday).
In California, polls have been moving against the Republican, Carly Fiorina. I would probably take Ms. Fiorina’s side of the odds our model is now offering — it has her as a 5:1 underdog — simply because the Democrat, Barbara Boxer, has had substantially more commercials running in the last two weeks, and that can sometimes produce fairly short-lived bounces in the polling. But Ms. Boxer is now a clear favorite, which was not the case some weeks ago.
Washington is another state that had seemed to move in the Democrats’ direction. But two new polls using Rasmussen Reports methodology — one issued under their own brand and another on behalf of Fox News — now show the race having drawn back to even. Our model still has the Democrat, Patty Murray, as about a 3:1 favorite as it waits to see if the trend shown by Rasmussen is confirmed in other polls.
Connecticut is a state that I have written about extensively. Suffice it to say that I think other observers have overlooked some important pieces of information — like the fact that there are very few undecided voters, and that both polling firms that have been active in the state, Rasmussen Reports and Quinnipiac, have tended to show numbers that are relatively favorable to Republicans. These details might suggest that the race is not quite as close as it appears. Still, this is another case where I would place a bet on the G.O.P. side of the line our model is suggesting (it still has the Republican, Linda McMahon, at under 10 percent to win.)
The New York special election, where the incumbent Democrat, Kirsten Gillibrand, is running, is probably a reach for Republicans — but the polling has been inconsistent in the race and a G.O.P. victory there remains a theoretical possibility. The reason to be skeptical is that the contours of the race are likely to be dictated by the higher-profile governor’s race, and the Republican candidate for governor, Carl P. Paladino, is having a difficult week and may turn out to be a poor standard-bearer for the Republicans after all.
So, the Senate will not come easily for Republicans. But, in contrast to previous weeks, the party seems to have multiple paths toward gaining control of it: one runs through “new” states like Connecticut and West Virginia where the polling has been moving in their favor, and the other through “old” states like California and Washington where the numbers had been running against them, but the momentum could reverse itself.
A third path might involve Republicans winning nine rather than 10 races, which has a 13 percent chance of happening. They then might be able to persuade either Joseph I. Lieberman or Ben Nelson to conference with them. For forecasting purposes, we do not account for this contingency, since there is no robust way to estimate the likelihood of one of these senators defecting. But it should be kept in mind that Democrats could keep nominal control of the Senate on election night, but then lose it before the new Congress convenes in January.
This expanded playing field, among other things, is forcing Democrats to make difficult decisions about where to spend their advertising dollars. Democratic candidates in Kentucky, New Hampshire and Pennsylvania — states where the party’s candidates have some reasonable chances of winning, but are clear underdogs — could see their budgets cut.
*-*
I also wanted to discuss briefly how we are handling Alaska, where the race is difficult to forecast because polls evaluating Senator Lisa Murkowski’s chances as a write-in may be unreliable. The procedure we have adopted is to apply a “write-in penalty” to Ms. Murkowski’s vote share, which varies randomly between 0 percent and 50 percent of the votes she otherwise would have received. The penalty is applied only to polls that explicitly mention Ms. Murkowski’s name, as though she were a listed candidate; polls that make some effort to “hide” her from voters are taken as-is.
As polling has suggested that Ms. Murkowski’s write-in votes would come predominately from Scott McAdams, the Democrat in the race, 60 percent of Ms. Murkowski’s lost votes are reassigned to Mr. McAdams, 35 percent to the Republican, Joe Miller, and 5 percent to the Libertarian candidate.
In other words, the model treats polls that mention Ms. Murkowski by name as reflecting her ceiling, rather than necessarily her most likely performance. And it regards the extent to which she will underperform this ceiling as being uncertain — it will apply a large penalty to her vote total in some simulations, but a small one in others.
Right now, it is not clear if Ms. Murkowski or Mr. Miller would be favored if she were named on the ballot — the polls, in addition to other issues, are of mixed opinion on this question. But if the penalty that Ms. Murkowski suffers for being a write-in is at all substantial, that would make Mr. Miller a reasonably clear favorite — and our model thus gives him an 85 percent chance of winning, versus 12 percent for Ms. Murkowski and 3 percent for Mr. McAdams.
We will remain on the lookout not just for new polls of Alaska, but also other experiments — like the smart one conducted by Ivan Moore for The Anchorage Daily News today — that convey some impression of the potential magnitude of the write-in penalty.
This does not have much impact on the overall Senate forecast, however, because we treat a victory by Ms. Murkowski as being a Republican victory, since she has said she will still caucus with the Republican Party if her write-in bid succeeds.