Princeton Election Consortium

A first draft of electoral history. Since 2004

With Perdue in, Nunn now an underdog (GA-Sen)

July 23rd, 2014, 11:08am by Sam Wang


Nate Cohn chose today to drop his story on how demographic trends bring Georgia surprisingly within reach for Democrats. I guess that is true, but the timing is funny. The race was close to begin with, and then yesterday’s primary was a boost for the GOP. In my view, that is the real story.

Like many, I expected Rep. Jack Kingston to be the Republican nominee in the Georgia Senate race. But businessman David Perdue bested him by about 1 percentage point. This was a substantial miss for polls, which showed a median 5.5-point lead for Kingston. However, that is par for the course for primary-season polls (dissected here). It is hard to predict which partisans will turn out.

However, general-election polls are more reliable in the aggregate. What they tell us is that the Democratic candidate, Michelle Nunn, probably got the toughest opponent possible for November. Out of six survey organizations who polled Georgia in 2014, only one gave Kingston a lead over Nunn (median margin, Nunn +2.0±2.9%). In contrast, three out of six gave a lead to Perdue over Nunn (median margin, Perdue +1.5±2.9%). That means Perdue is a stronger candidate, by 3.5% in the general election. (Note that there is a small chance, 1 in 10, that actually Kingston is stronger.)

To back up a bit, Nunn’s chances in November are still not that bad. Perdue, despite being cousin to a two-term governor, is himself new to running for office. He could make another misstep, like denigrating people who “just” went to high school. But all the same, voters coalesce around their nominee after the primary season. In the coming weeks, we should expect movement toward Perdue. At this point I would call Nunn the underdog, by a small margin.

This season, the Republican party has mostly picked their strongest Senate candidates during the primary season: Mississippi (Senator Thad Cochran over Tea Partyite Chris McDaniel) and Alaska (Dan Sullivan over Mead Treadwell). Georgia adds to that list. The one misstep has been Iowa, where Mark Jacobs probably would have been a better choice than Joni Ernst, who subscribes to fringe ideas like the threat of United Nations Agenda 21. Tea Party-like Republicans aren’t a major factor in the 2014 Senate campaign. However, they are keeping the GOP in line, both in agenda and tone. This is more apparent in the House, where they continue to go after – and win – nominations in heavily Republican districts. Whatever happens in November, we will have pretty much the same Republican party next year.

Tags: 2014 Election · Senate

20 Comments so far ↓

  • Zathras

    Sam,

    Are you going to post something on the methodology behind the NYT/Yougov flurry of polls released yesterday? I would be very interested to hear your take on this.

    • Sam Wang

      Briefly: I think the proof is in the outcomes. YouGov has an excellent record, and Doug Rivers has a sterling reputation in this domain. I think it requires a level of sophistication other pollsters do not have – yet.

  • Tony Roberts

    Professor Wang, The RCP generic ballot average is currently Democrats +2.5. Yet, the likes of Rothenberg and Sabato maintain that, based on present conditions, the Republicans are likely to gain a handful of House seats. I understand the effects of gerrymandering, but do their predictions comport with the RCP average? Do you think the Democrats have a chance, based on an RCP average of 2.5, of gaining seats? Thank you, as always.

  • Bill

    I am curious about today’s (July 28) snapshot at the top of the page showing a 75% probability of a Democratic Senate majority. If you click the link to the right hand side of the page (today’s Senate snapshot) you see a Republican probability of Senate majority of .55. Why the difference? is this an error, or am I misunderstanding what the different numbers mean?

  • Bob Grundfest

    A new Upshot today says that the math favors the Republicans, but the methodology involves adjusting numbers based on the pollster’s lean. This is decidedly NOT Wangian analysis.

    http://www.nytimes.com/2014/07/28/upshot/republicans-senate-chances-rise-slightly-to-60-percent-.html?hp&action=click&pgtype=Homepage&version=HpSum&module=second-column-region&region=top-news&WT.nav=top-news

    • Sam Wang

      Yes, I saw that. It provoked huge suspicion in me.

      On the other hand, the big reveal – which they totally buried – was a Begich lead of 12 points in Alaska! If that holds up, it is a game-changer.

  • Amitabh Lath

    Do Tea Partiers vote strategically? To put it another way, will all the Kingston voters now switch their votes to Perdue because he is the more conservative choice in the general?

    Usually, ballot papers have several fringe parties’s candidates listed. If enough Kingston voters go to one of those, Nunn could pull it out.

    Again, a likely voter model that actually worked in these troubled times would be a great help.

    • ArcticStones

      It would indeed be worthwhile to distinguish between “likely voter” and “registered voter” polls. The question is whether that would leave enough data points…

  • Tony Roberts

    Professor Wang, when you say “a bit higher” because of Colorado and Iowa. Are you suggesting that the Democrats vying in those states have reclaimed the lead? Thank you, as always.

  • Ron Pitts

    Agreed, without as many TeaPublicans Engaged the R voters are sending the Senate Candidates that seem to have less of “‘teh crazy”. But the real test will be seeing if even a single one of these candidates implodes. Although you can’t poll it, odds are at least one will.

  • Aaron

    It should be noted that the Alaska GOP does not technically have a nominee for Senate yet.

  • ArcticStones

    Another discouraging development, is the NYT article documenting Senator John Walsh’s plagiarism. While plagiarism is never good, I think voters will be especially unforgiving that this happened at a military college – whatever the circumstances (such as a fellow soldier recently having committed suicide).

    Senator Walsh also recently admitted being on anti-depressants. That does not win him any points either, although I am sure that is a question of culture…

    While Soviet politicians in bygone eras might disappear for a month “with a cold”, and American politicians might go on “extended holidays”, it raised eyebrows in the international community some years ago when Norwegian Prime Minister Kjell Magne Bondevik openly admitted to being depressed, and taking anti-depressants, after the death of a close friend.

    As I understand it, the governor-appointed Senator was really struggling to become competitive in the upcoming election. The question now is whether the Democratic Party has essentially handed a victory-by-walkover to Republican candidate Steve Daines.

  • Tony Roberts

    Professor Wang, can you update your “probability of Democratic Control of the Senate” analysis in light of recent Colorado, Iowa, Georgia and North Carolina Senate polls? Thank you!

    • Sam Wang

      Currently the median is still 50D/I, 50 R. Pr(Democratic control)=75%. A bit higher, because of Colorado and Iowa I think.

      Possible changes in the near future: Begich (D-AK) up, Nunn (D-GA) down. Those will cancel one another to an extent.