Princeton Election Consortium

Innovations in democracy since 2004

Outcome: Biden 306 EV (D+1.2% from toss-up), Senate 50 D (D+1.0%)
Nov 3 polls: Biden 342 EV (D+5.3%), Senate 50-55 D (D+3.9%), House control D+4.6%
Moneyball states: President AZ NE-2 NV, Senate MT ME AK, Legislatures KS TX NC

A convention’s place is in the House – and the Senate

September 27th, 2012, 12:00pm by Sam Wang

It’s been kicked around in punditland that party conventions have little value in modern political life. This is contradicted by the lasting change in President Obama’s lead over Governor Romney, as well as the notable increase in his party’s likelihood of retaking the House of Representatives. Now, reader Rick in Miami has done a nice bit of analysis that completes the picture. He finds that top-of-the-ticket moves are closely accompanied by parallel downticket effects in the Senate. Because of the convention and its aftermath in the Presidential race, Democrats are no longer at meaningful risk of losing the Senate.

Last week I posted my code for making the Senate prediction. Rick downloaded it and souped it up. Here’s what he found, in his own words:

I took your file senate.m and upped the “in play” seats to 14, and modified it to plot a time series (attached modification “senate2.m”).

The time series (in attached figure) shows that control of the Senate has been much more volatile than the presidential race, with the % chance of Dem control dropping sharply from about 80% in mid-August to below 50% in late August. That drop ended with Akin’s infamous comment. The meteoric increase in the second week of September – the week after the Dem convention – is extremely dramatic! I remember you posting that the Senate situation was changing, but didn’t appreciate the magnitude of that change until I plotted this time series.

Take a look:

The Democratic control probability is 94%. The median expected split is 53D/I, 47 R (68% confidence interval, 52-54 D/I). Now that the Senate has essentially left knife-edge territory, the one remaining knife-edge issue is the House of Representatives (D – ActBlue) (R – Crossroads GPS).

The remarkable similarity between this plot and the Presidential Meta-Analysis is most easily explained by partisan enthusiasm. Certainly the addition of Paul Ryan as VP was electrifying to the Republican base. In some sense, the Republicans got their convention bounce – two weeks before the convention. (I note that something similar happened in 2004 for the Democrats with John Edwards). Then, with the Akin remark and the Democratic convention, the continuing parallel path of President Obama and his party’s Senate fortunes confirms the idea that some shared cause is at work.

History of electoral votes for Obama

Before, I had been entertaining the idea that the conventions’ main effect was to change the minds of decided voters, à la the the RAND survey. But in the last few weeks, their survey shows a net total shift of only about 1.5% of margin toward Obama. Over the same period, the Meta-Margin grew by 3.0% and is still on the move.

Now we have a quantitative explanation for the remaining change. As I have written, Senate candidates received bigger bounces in strongly Democratic states. Combined with Rick’s analysis, I suggest that at least half of the bounce is caused by increases in partisan enthusiasm – an effect with repercussions up and down the ticket.

This analysis quantifies what the Democrats got out of their convention. For them, it was money and time well spent. As Rick wrote to me, it

puts a rather strong counterpoint to the idea that “the Conventions are outdated and unnecessary.”

Rick’s code is here (some assembly required).


As a consequence of the change in Senate odds, I am unofficially retiring all Senate races as an ActBlue priority. The most effective place to donate is now towards House races, i.e. the DCCC. Senate races will still be shown with a ranking of relative importance, just de-emphasized. For Republicans, I suggest that you trust Uncle Karl.

Tags: 2012 Election · House · President · Senate

20 Comments so far ↓

  • Tapen Sinha

    IMHO that 47% comment was a Lehman Moment. It was just as devastating as the Palin interview four years ago. Obama camp certainly thinks so. They are releasing an ad for seven states (Ohio, Virginia, Florida, Iowa, Colorado, Nevada, and New Hampshire) with just Romney speaking but with a series of pictures



    • Rick L

      I’m sure you’re right that the 47% comment has played a significant role since it’s release. However, it didn’t play much of a role before it was highlighted by David Corn on Sep. 17. The odds of Dem senate control had already soared from ~50% right before the DNC to nearly 90% by then. So I think the meteoric rise was more likely fueled by the DNC, and maintained by subsequent gaffes. There was one gaffe concurrent with the rise: Romney’s much-criticized response to the embassy killings in Libya. But it’s hard for me to imagine that in isolation would have had such an effect. Hence the emphasis on the DNC.

  • Herb

    Could you comment about early voting

    • Sam Wang

      What to say? It’s started. It is likely to help whoever is going to lose ground in the future. We don’t know who that is.

      Also, many pollsters ask whether a person has voted. So it’s baked into the polls.

  • Rick L

    Thanks for posing this, Sam. The shape of the presidential meta-analysis and Senate time series are indeed strikingly similar, but it’s also worth noting the difference in magnitude. While the former never left Obama-leads territory, the latter has gone from >50% chance of R control in late August to >90% chance of D control now. This increased volatility suggests that factors like complacency from a large presidential lead (or, for that matter, despondency from a large lag) could potentially make larger swings in the Senate curve than in the presidential one.

    • Sam Wang

      I don’t think that’s the right explanation – the Presidency and Senate may have similar volatility. The probability zooms up and down because the Senate expected range has a tie near its center, whereas the Obama/Romney race does not.

      You do a raise a good point. A fair comparison would include change in median margins for Senate races. My guess is that the ratio is 1:1 overall, with larger gains in strongly Democratic states. This is based on my previous analysis. I’ll try to put that up soon.

  • Berg

    Prof. Wang,

    Do you believe that it is worth the risk and cost for Obama to shift resources to AZ in the hope that his coattails could improve Richard Carmona’s chances to win the AZ-Sen.

    • Sam Wang

      On the assumption that party enthusiasm underlies the Senate-level bounce, yes. MA and CT might be better travel destinations.

      While we are sending candidates around the country, Romney could visit IN as a side trip from OH.

  • Matt McIrvin

    Yeah, if we’d made a graph of “if the election were held today” win probability for the 2008 presidential election, where the RNC bounce actually had McCain/Palin up over Obama, it’d be a rise as spectacular as this one.

  • Rieux

    It’s probably more suited to navel-gazing than objective analysis (at least as long as n = 1), but one wonders precisely what it was about the DNC that had such positive effects–on Democrats’ enthusiasm or whatever else.

    Was it the full-throated defense of LBGT and reproductive rights? The terrific night-closing speeches (I happen to think Michelle Obama’s was incredible, better than Bill Clinton’s merely very good one, but I sense I’m in the minority in that respect)? The clear communication that foreign policy and national defense are now issues that cut in Democrats’ favor? Or what?

    • wheelers cat

      Dr. Wang pegged it on FLOTUS’s speech.
      But I think its incontrovertible that the DNC bounce is sustained on gaffes.
      just my humble opinion I guess.

    • RadicalCentrist

      I think it was Clinton and 2 points that he made in particular:

      1. Are you better off than 4 years ago? Yes! The Republicants all see themselves as the re-birth of Reagan and see Obama as Carter and thought they would win simply by asking that question. Clinton punched so many holes in that boat that it can now be found at the bottom of the Marianas Trench.

      2. Medicaid and the Ryan budget. For decades the Republicants have done very well portraying social programs as going largely to the poor (i.e., brown people). Clinton made clear that 2/3 of Medicaid goes to the elderly and disabled and cutting it would devastate many middle class families. This point has, never in my mind, been made in such a stark way before to a large audience of non-wonks.

  • Olav Grinde

    Sam, your comments and analyses, and Rick’s time series, are most enlightening! It is striking that you now calculate the odds of continued Democratic control of the Senate to be > 90 %.

    Is there any chance that you will be rating the House races where donations will do the most good, or is that an insurmountable task?

  • Dave Kliman

    Sam, could you comment on the 6 page double sided 8 point type ballots in florida and if taking away a lot of early voting as well as making it take more than 30 minutes per person to vote will have a significant impact, in your opinion?

  • Olav Grinde

    I am puzzled by yesterday’s Bloomberg generic House poll, which gives the Republicans +2% .

Leave a Comment