Princeton Election Consortium

A first draft of electoral history. Since 2004

High-leverage races for 2014

June 28th, 2014, 2:56pm by Sam Wang


This year, the big political question is who will control the Senate in 2015. I’ve analyzed this briefly and will continue to do so in the coming months. As of the end of June, it’s looking like Democrats/Independents (who vote together) and Republicans are likely to win 47 seats each. The remaining six races are currently on the knife’s edge, and will determine control. They are: Alaska, Colorado, Georgia, Iowa, Kentucky, and Louisiana.

If you want to make the most of your donation, give to your side’s candidates in those six races. As I’ve written in past elections, donations are most effective at the margins. In close races, donations are most likely to move the win probability. In addition, as national politics has become oriented around parties rather than individuals, it is control of the chamber that matters, rather than specific individuals.

Note that this advice is the same whether you support Democrats or Republicans. For your convenience, I have provided links at the left to ActBlue (Democrats) and the National Republican Senatorial Committee (Republicans). The first link will be updated if conditions shift between now and November.

Tags: 2014 Election

19 Comments so far ↓

  • Pat

    I am quite surprised with these ratings. Especially, you say Arkansas is “out of range”, but on which side? Safe for Democrats? or pretty much a sure grab for Republicans? Sorry if this is unclear, because most other analyses actually report Arkansas as a big tossup.
    http://www.nytimes.com/newsgraphics/2014/senate-model/comparisons.html

    • Sam Wang

      For heaven’s sake, just look at the data. It’s the only input, totally transparent, same as always. Take a look at this. Pryor’s been behind for several months.

  • Woo

    About to use the ActBlue page myself. Thanks again.

  • Eric Walker

    Thank you for the Act Blue page setup. Visited and used.

  • Amitabh Lath

    Cochran’s primary win in MS has me wondering if the old models of midterm voter turnout are still valid. Conventional wisdom is that most eligible voters are too disconnected to vote in midterms, and even fewer participate in primaries. The fanatic tail of the distribution determines the outcome.

    But somehow the word got out and this R primary runoff election in MS had turnout that exceeded expectations.

    Does this mean that the general public is now paying attention, they can be reached and turned out to vote? If the turnout is more 2012 than 2010, we are looking at a much different landscape.

    • MAT

      Amitabh,

      I’m not sure we should try to extrapolate a lot out if Cochran’s win in MS. My understanding is that turnout increased in heavily military districts, where the ‘bring home the pork’ msg could resonate. Then also there is the fabled crossover vote. It will be interesting to see the impact that may/may not have once deeper analysis is performed.

      But in any care, this strikes me as an outlier resulting from a rather innovative campaign taking advantage of unique circumstances.

    • Sam Wang

      I think it was basically Cochran’s people turning to Democratic voters out of desperation. Harry Enten has drilled into the topic, and found data consistent with the idea that among Republicans, Cochran lost by 8 points. Heavily African-American counties made up for that, and then some. I view the primary as being basically the same information as Eric Cantor’s race: discontented GOP base voters who are not afraid to eat their own.

    • Amitabh Lath

      That is sort of my point. In the MS R-Senate runoff, somehow apathetic voters who never vote in primaries were contacted, educated, and mobilized.

      Bottom line. Big turnout for one that is historically a low interest election.

      I don’t know if it was the military districts as MAT says or African Americans as Sam (and others say) but somehow these people turned out.

      Whatever the techniques (Facebook? Tweets? sermons in black churches? articles in the Stars and Stripes? good old fashioned door knocking?) turnout increased.

      And we all know what increased turnout means in November.

  • Bill

    Is your 60% estimate based on a meta-analysis of the type you have done for the presidential elections?

    • Sam Wang

      Yes. Mathematically it is the same problem, reduced in complexity because each Senate seat has exactly one vote, whereas states have varying representation in the Electoral College. I’ll lay it all out fairly soon.

      If you can read MATLAB, the algorithm is here.

  • Tony Roberts

    Sam, I am little surprised by your characterization of North Carolina. Most of the recent polls show Senator Hagan ahead. Can you explain further?

  • Bill

    So are you still projecting a 67% chance Democrats will control the senate in 2015?

    • Sam Wang

      More like 60%…but I urge you to think about that as a knife-edge situation. In other words, if it drops to 30% or rises to 70%, to me it’s the same idea.

  • Tony Roberts

    Sam, you do not include Arkansas among your “knife’s edge” list. The RCP average shows a very close race. But you mention “six races” and only list five. Was Arkansas an omission? Thanks for the fine work!