Princeton Election Consortium

A first draft of electoral history. Since 2004

Alaska and Colorado on the move

September 26th, 2014, 3:57pm by Sam Wang


The Meta-Margin is a powerful statistical measure. It collects all available polls into a simple index that tracks movement in the national campaign. No house effects, no fancy stuff – just a measure of national opinion, calibrated using Senate control to define the zero point.

Recall that the Senate Meta-Margin is defined by how much all Senate races would have to swing together in order to create a perfect tied probability for Senate control between Republicans and Democrats+Independents. Yesterday, the Meta-Margin moved by 1.4% toward Republicans. By the standards of this year’s campaign, that is a huge movement. The change was mostly caused by movement in just two states, Alaska and Colorado. Those were ways for Democrats+Independents to hold 50 votes, and as of today, those routes are currently looking unfavorable. Therefore now the Meta-Margin is R+0.1%. That is a near-perfect tie if polls are exactly calibrated to Election Day behavior (is it? that’s a question for another day).

Our time window is currently to take the last 3 polls or the last 2 weeks of data, whichever is more, for each state. This measure takes a little while to move, but when it does, that’s meaningful. Statistically, we are now at the most Republican-leaning end of the range that we have seen in the entire graph.The dip in June looks better for Republicans, but keep in mind that on September 3rd, the Meta-Margin jumped by 0.8% when Chad Taylor (D-KS) dropped out of the race. Subtracting that gives a better feeling for where we’re at, not counting Kansas. In short: as a group, Republican Senate candidates outside Kansas are at an all-time high.

Will this change stick? There is still plenty of time for movement in most of the races listed in the right sidebar (The Power Of Your Vote). Still, we should take the latest change fairly seriously. Alaska was a hidden bonus for Republicans all along, as I analyzed yesterday. But the change in Colorado polls – I’ve been pondering that all week. Something big happened there: a swing from Udall +3.0±0.7% (mean±SEM, 4 polls) to Gardner +4.5±2.2% (4 polls). As a trigger, I nominate the Udall-Gardner debate, which is the kind of event that can move opinion by showing the two candidates side by side.

As usual, recall that The Power Of Your Vote is a list of states where individual votes are most valuable. Those states are where get-out-the-vote activism and campaign contributions will have the greatest impact.

Tags: 2014 Election · Senate

18 Comments so far ↓

  • Natalie Rosen

    Thanks to Rachel Maddow I am hooked on Sam Wang and the Princeton site. Admittedly I am a Democrat and deep depression awaits and not the economic kind although that can happen too if after the hard hard fought for policy is threatened by a Republican take over of the Senate. Republicans ruin a nation, cost many lives and tons of bucks and then depend on Democrats to pull us through. Hoping the electorate will check out some historical references of the Great Depression, Great recession and others in history. It repeats over and over again, Republicans and their love of the corporation over people kill us in the end!

    I will continue to keep Same Wang as my major source but will look at Nate Silver too.

  • Peter

    There was an article in the NY Times on Saturday that talked about the Colorado democratic governor having problems. This may be moving the Senate race as well.

  • Stuart

    The rise of ISIS will elect Republicans, especially with what happened in Oklahoma.

  • Ben

    Your election day probability is based on current conditions, and variability as seen over prior months, correct? It looks to me, like rather than varying back and forth (especially when factoring I the Orman shift), the variability is around a trend line that is heading R-ward. My rough view of the slope puts the trend (of your meta margin) at about R+1.5% at election day.
    Using that and including the variability, i expect would put the probability much less than 70% that Ds hold the Senate. It would seem such a trend line would capture a changing national mood, But is there a reason it would not be accurate?

    • Davey

      I’m not sure a see the meta-margin as a useful tool for predicting future value of the meta-margin. In other words, I’m not sure why you’d presume this data alone would favor an R +1.5 value in November any more than a possible D +1.5.

    • Sam Wang

      Could be. I wonder how to identify such a trend earlier, rather than afterward.

  • RB

    Nothing for nothing but maybe Udall speaking on behalf of beheaded journalist like he did on his first debate and refusing to debate Gardner(and getting hammered by local media) has been the cause of his decline. Also running singularly on ‘war on women’ when Gardner has already diffused it is cumbersome and boring. Choose what you want to believe but after those two moments Gardner has lead in 4-5 polls on a row. HHR

  • sdf

    Would be very interested to see your analysis of any of the gubernatorial races that are actually close — e.g. Florida, where the RCP poll average now has as a dead heat, although the HuffPo pollster page still has a slight edge to Scott. Any chance you might look at the several states that are still in play?

  • Dean

    Dr. Wang, I believe you stated this year’s Senate races will be very tight overall, like Kerry-Bush 2004. If I didn’t misunderstand or misread, that’s what appears to be happening now.

    I’m sorry if I don’t have all my facts in order, but it’s been a long week.

    I listened to your interview the other day, and you said that races are pretty well set in September, polls-wise. I would like to digress from the Senate and mention a very interesting Illinois gubernatorial race.

    The challenger, Bruce Rauner, was leading Gov. Quinn by 12 points in early August in a We Ask America Poll. His numbers have been tanking in this polling outfit, and the latest poll has him up by 3.

    There were a two pro-Quinn polls lately, and one of them is possibly an outlier (Quinn by 11).

    Quinn and the liberal super PAC’s are hitting Rauner, a multimillionaire private equity guy, for his business practices, such as poorly run nursing homes in which there are lawsuits in the tens of millions of dollars. A bankruptcy trial involving Rauner’s nursing home company is happening right now.

    There are far more issues with Rauner that allege vulture capitalism, fraud, etc.

    Anyway, it’s September, and the polls may or may not be set for Rauner (who’s up by a few but has been dropping). This is a highly-interesting race, and I expect to see more “vulture capitalism” ads.

  • Laurie Roberts

    I don’t know why the change in the polls, but I will point out that this is the first year Colorado has an all mail-in ballot election. I was at a local college registering voters today and easily 95% of the students were already registered, meaning they will all get a ballot in the mail here shortly. FWIW

    • atothec

      If true that should DEFINITELY benefit the Democrats.

    • EmmaAnne

      This factor will start to be captured in the polls shortly when people start to mail in their ballots. It *should* help the Democrats.

  • atothec

    Conservatives that have been on the fence are now starting to consolidate. On the other hand Dems had a long term strategy in place and appear to have the upper hand in upcoming ad time and then there’s the Bannock Street Project ground game.

    If this were the old days I’d be a bit more concerned but Obama revolutionized voter outreach and gotv with his technocrat team. The fact that there’s been no mention of Brannock since summer I take as a good sign.

    That Quinnipiac poll though is an outlier. There are going to be more outliers from the usual suspects like Gravis and Quinnipiac. As long as Udall stays within 2-3pts I’m not worried.

  • Alex

    Do you have any information on key gubernatorial races? As a Texan, I’d like to know how the Greg Abbott vs. Wendy Davis race is doing.

  • Tony Roberts

    In light of these developments, why have you not felt compelled–and I am not complaint–to alter your long term forecast, Professor Wang? I recall vividly during the 2012 presidential race that you were steadfast in your long term prediction even when, during September and early October, polls indicated that Romney might win.

  • Randall

    Obama lost his first debate, too. Then stepped it up. Perhaps Udall will do the same.