(10/21) Election Forecasts 2010: Senate and House

In a previous post on 9/29 I developed my election forecasts for the 2010 Senate Election; this is an update to that post. But first, let's talk the U.S. House.

For those of you who are tuning in late to politics, remember that there are 435 House seats, apportioned to states according to population; all 435 seats are up for election every 2 years. This makes the House a volatile and sometimes rapidly changing group of people. Unlike other major forecast sites, (see CQ Politics, Cook Political Report, Real Clear Politics, Pollster, Fivethirtyeight/NYT, etc) I don't have the resources to make forecasts on individual House races. But I do have the time to make general statements about the House based on these forecasts, where it's likely Democrats will lose somewhere between 25 and 75 "net" seats. That may seem like a huge margin of error, but there are substantial issues with forecasting House races this year, as has been mentioned and reiterated almost constantly by Nate Silver; his argument is hard to refute. Still, Republicans need to take 40 seats from their current position to obtain a majority, which seems increasingly likely. Of the 150 or so House seats that are being "seriously" contested this election, 90% are Democratic incumbents or Democratic seats. This really gives some perspective on why Democrats have such a problem this year in the House; it may turn out to be the greatest turnover of House seats in history.


The Senate race actually tells us a slightly different story. Democrats are actually well positioned in the Senate comparative to the House, meaning that they are at this time very unlikely to lose control of the Senate chamber.  Again, for those of you who are coming into politics late, remember, each state gets 2 Senate seats, each Senator serves for a 6 year term, and only 1/3 of the Senate gets elected every 2 years. Thus only 2/3 of states have senate elections every 2 years, in this case there are 37 Senators being elected (2 from NY). Republicans now at 41 seats, need to have a net gain of 10 seats in order to be reasonably confident of taking control of the Senate with 51 seats. (In the case of a tie, 50-50, Democrats would still be considered "in control" because Vice President Joe Biden would cast any tie-breaking votes).

In the coming paragraphs, I use the term "competitive" to mean that either party has a chance of winning, and the term "almost certain" to mean that I am as certain as is possible, under the circumstances of prediction. I use the term "tossup" to mean that based on polling, spending, and state specifics it is virtually impossible to predict a winner- it might as well be a coin toss. In this update, Democrats have made some slight gains since my last forecast; let's take a look.

Last time I pointed out that Democrats "almost certainly" will not take any Republican-held seats (Florida is a possible exception that I will discuss at the end), so every "competitive" seat is Democratically held (Democrats have everything to lose, nothing to gain this election). This factor is unchanged. Moreover, I pointed out that Republicans were "almost certainly" going to win North Dakota, Arkansas, Indiana, and Pennsylvania, turning those states Red. Pennsylvania has been removed from this column in this update. This leaves us with a baseline of 48 Democrats and 43 Republicans. I am "almost certain" that Democrats will not have less than 48 seats, and that Republicans will not have less than 43 seats after this election.  

Last time I also established that there were 9 seats that I considered "competitive", where either party has a notable chance of winning. These seats were Washington, California, Nevada, Colorado, Illinois, Wisconsin, West Virginia, Florida, and Connecticut. Since that time, Connecticut has proven non-competitive; an apparent surge by Republican Linda McMahon has evaporated (it may have just been coincidental polling noise, media attention, or a temporary effect of McMahon's sizable personal contributions to her own race), leaving Democrat Richard Blumenthal with a commanding lead.

Meanwhile, as mentioned earlier, I would like to add 1 new state to the "competitive" list: Pennsylvania, which was formerly an "almost certain" win by Republicans.This leaves us with 9 "competitive" Senate seats.

Washington and California have been polling Democratic, so we'll put them in the D column. Florida, Pennsylvania, and Wisconsin have been polling Republican, so we'll put them in the R column. This leaves Colorado, Illinois, Nevada, and West Virginia as "tossups". I gave Republicans a 2/3 edge amongst the tossups last time; this time, coin-flips as these races are, I'll assign these seats half and half for this forecast. This leaves us with 52 Democrats and 48 Republicans, a one-seat gain for Democrats since my last forecast. Remember, though, I still think there is a reasonable margin of error with this forecast of four seats, which means the numbers could range from 50D-50R to 54D-46R. This good news for Democrats, but not great news- still, it seems highly unlikely that Republicans will take control of the Senate at this time. 

I want to make an additional note on a few individual races.

First, Wisconsin has shown stronger and stronger polling for Russ Feingold's (D) opponent Ron Johnson (R). If Johnson continues in this fashion, I will remove Wisconsin from the "competitive" races, despite my tendency to give an incumbent Senator the benefit of the doubt.

Second, Pennsylvania has shown some very good numbers recently for Democratic candidate Joe Sestak. Since Pat Toomey (R) has held such a strong lead for so long, I am dubious about these numbers. I'll be paying close attention to this race in the future.

Third, Kentucky's race has seemed to be tightening recently. Kentucky is the one Republican-held seat that might seriously be vulnerable this year, besides Florida which I'll discuss below. Despite Rand Paul's (R) libertarian-leaning positions on a lot of issues, I have remained highly skeptical that conservative Democrat Attorney General Jack Conway can beat him. While RCP continues to consider Kentucky to be a tossup, and other forecasters say it only leans Republican, I haven't been convinced by the polling or the other dynamics of the race. Thus I still do not consider Kentucky in my list of "competitive" races. The same remains true for my forecast on Missouri, which is considered by most to be less close than Kentucky, but still in the running. I am not convinced by Robin Carnahan's (D) numbers either.

Fourth, Colorado, California, and Washington have consistently shown very close numbers. Despite some reservations, I have moved Colorado into the "tossup" section because the race seems to have tightened even further. California and Washington's numbers are not so different from Colorado, but due to the incumbency effect, deep respect for Fivethirtyeight's methods, and some local dynamics like, say, Prop 19 in California, and mail-in voting in Washington, I have retained these West coast states' positions on the D side.

Fifth, and finally, Florida has shown a consistent double-digit lead for Republican Marco Rubio for many weeks now. Florida's race is interesting because popular incumbent Republican Governor Charlie Crist has entered the race as an Independent. Crist (I) has been polling around 30%, Rubio (R) around 40% and Kendrick Meek, the Democrat at around 20%. Polls have shown that at this point Crist is getting a lot of his support from Democrats, rather than Republicans, with about 50% of Democrats supporting him, and only 10% of Republicans. Three-way races are hard to predict, as people tend to vote strategically, rather than with their conscience, and it's hard to reflect this in polls. Also with three-way races, just a little gain at the expense of one candidate can cause another candidate to spring, suddenly, into the lead. It is for these reasons that I have kept Florida on the list of possible Democratic takeovers. It is highly unlikely that Meek can beat Rubio, but Crist still might be able to beat Rubio if a lot of Meek voters defect from him at the last moment and vote for Crist in order to keep Rubio, seen as an ultra-conservative, from winning. It's hard to know if, and to what degree, this might happen. Right now it doesn't look likely. But if it were to happen, and Crist won, he would have to decide to caucus with the Democrats or the Republicans. Since, at that point, it would have been Democrats who had propelled him into office, I think there's a good chance that Crist would caucus with the Democrats, thus sending Florida into the D column. There are a lot of "what-ifs" in this scenario, which is why I keep Florida on the "competitive" list, but I also still assume it'll go Republican for now, since the chances are slim.