NH-Gov: Lynch (D) Totally Romping

Public Policy Polling (PDF) (3/31-4/3, New Hampshire voters, no trendlines):

John Lynch (D-inc): 57

Jeb Bradley (R): 33

Undecided: 10

John Lynch (D-inc): 57

Ovide Lamontagne (R): 33

Undecided: 10

John Lynch (D-inc): 57

John Stephen (R): 29

Undecided: 14

John Lynch (D-inc): 54

John Sununu (R): 36

Undecided: 11

(MoE: ±3.5%)

John Lynch is nobody’s idea of a partisan progressive fighter, and we don’t even know if he’ll seek an unprecedented fifth two-year term, but after last November’s utter wipeout in New Hampshire, these numbers have to be heartening to any Democrat. Lynch faced a competitive race from John Stephen in 2010, ultimately winning by eight, so it’s pretty remarkable to see Stephen not even cracking 30. The best performance comes from ex-Sen. John Sununu, but at 18 points back and terrible favorables, Bununu doesn’t look so hot, either.

The obvious question to ask here is what does PPP’s sample look like, since they’re showing such a big reversal of fortune. It’s 35 D, 29 R and 36 I, which makes it much more Democratic (and much less independent) than the last two exit polls have shown: 29 D, 27 R & 44 I in 2008, and 27 D, 30 R & 43 I in 2010. However (and this is an important however), respondents say they supported Obama in 2008 by a 51-42 score, very close to Obama’s actual 9.6% margin.

In a number of states, PPP has been showing a 2012 electorate that’s quite similar to the 2008 voter universe, something that’s been greeted with some skepticism given what seemed like record-high enthusiasm for Barack Obama last time out. We’ll see if history repeats, of course, but PPP nailed things last cycle, and unless and until another pollster (who actually shows their work – I’m looking at you, Quinnipiac) comes along to contradict their work, I’m going to give them the benefit of the doubt.

27 thoughts on “NH-Gov: Lynch (D) Totally Romping”

  1. The partisan breakdown is a bit too rosy. PPP should take off their “rose colored glasses” and reexamine this one. Normally I think that people bitching about them having too many Democrats is ludicrous, but this time I think they’d be right.

  2. because I’m by no means proficient at this, but wouldn’t the fact that PPP’s Obama numbers so clearly mirror his ’08 victory be more evidence the poll is skewed? We’ve seen universally that Obama’s numbers have dropped, especially in New Hampshire. If he’s still holding that same margin, he’s either seen a huge turnaround recently or the poll includes too many Democrats, right?  

  3. Except 57% over Sununu. It wouldn’t make a massive difference to the topline anyway. I would also note their final pre-election day poll of the state last year had Lynch winning 53-44 (he actually won 53-45) with a 28d, 37r, 36i party ID sample which was obviously nothing like the real deal. Once again, they constantly hit the mark even when the crosstabs are off.

  4. The Republican legislature is acting like a parody of what liberals think of when they think of what a legislature dominated by the far right. Every day they are in session generate either a new piece of hilariously impractical legislation or a freshl over-the-top offensive and/or braindead remark.

    It would not surprise me at all that Lynch’s numbers shoot up as swing voters are in no mood to risk giving a rubber stamp to that bunch.  

  5. I think PPP is going to be the Ras of this cycle.  Up to this point, they’ve done a good job, but now they are seeing too partisan of an electorate.  I don’t think they are deliberately trying to engage in ‘narrative setting’ but I don’t think they are reading the picture right.

  6. Having a sample with the same number of Dems and Indies in New Hampshire, especially coming off of a year like 2010 is definitely misguided.

  7. The partisan breakdown in 2010 was 27/30/44. Against John Sununu, Lynch’s breakdown is 84/13/57, which if my math is right gets him to 51.66 percent. PPP has him at 54 percent in this poll.

  8. I’ve always found the reasons PPP gives for not weighting by Party ID stupid.

    When I said they should take off their rose colored glasses and reexamine, I meant that they should at least state in their write up that they think Lynch’s position is somewhat less of a total rout and more like just a simple landslide.  

  9. mean that a bunch of people in NH will vote for Obama anyway despite their reservations because they dislike the GOP field even more.

  10. they were asking people who they voted for in 2008, so if the margin in the poll reflects the actual margin in 2008, that suggests that it’s realistic.

  11. the point was to show that the numbers don’t give any sort of ludicrously unrealistic lead.

    But anyway, in 2008, the breakdown was 29/27/45. With 84/13/57 against Sununu, this means that he’d have 53.52 percent of the vote, which assuming PPP rounded is the same.  

  12. If a candidate drops in popularity, voters will tell pollsters that they didn’t vote for him even when they did.  I’m not sure if this is simply people misunderstanding the question, or people actually suppress the memory of a vote they regret, but either way it is a phenomena.

    In regards to the NH poll, I think it’s perhaps slightly skewed to the Dems, but not very dramatically.  Partisan ID is flexible.  Many voters will tell a pollster a different party than their actual registration if they feel more aligned with that position.  Correcting for party weights causes you to miss shifts in opinion.  (This, BTW, is why Zogby was overly favorable to Kerry in 2004 and McCain in 2008, they adjusted their samples based on the last election’s turnout.)

  13. his 2008 numbers? Voters lie about (or, as some would generously put it, “misremember”) who they voted for when the politician in question has slipped in popularity. Obviously Obama is less popular now than he was in November, 2008 nationally, but his approvals have also taken a beating in New Hampshire in particular. So for him to have maintained the same level of support, either there is a pro-Obama trend suddenly sweeping NH that I’m unaware of, or the poll includes too many Democrats. Once again, I’m not a polling expert, but doesn’t that make some sense?

  14. I mean really, by what standard does one weight by party id? This isn’t the case where you have a clearly objective method to weight it by, as is the case with demographic weighting, so you have to start guessing.

    Plus, you aren’t even factoring in that it’s possible that the sample has “too many” Democrats, but that’s because many left-leaning Independents decided to identify themselves as Democrats. Which means if you try to re-weight by what you think is the proper party-id, you’ve just introduced bias which makes the poll more Republican-friendly than it should have been.

  15. This is something several of us have pointed out many times before, andyroo regularly providing calculations to prove it, that if you play around with the turnout model ID and re-do the numbers using the crosstabs, it barely budges the toplines.

    So go ahead and reverse the Dem and GOP percentages of respondents, and see what you get:  not that different.

    The reality of the math is that a change in the party ID that small, and yes moving them around in the mid-single digits is small, moves the job approvals and other numbers only a little bit.

  16. I a lot of republicans, who still vote republican, don’t consider themselves republican anymore, because they don’t want to be associated with a party that is doing so many childish things.  SO, if you weighed by party ID, you would get an invariably more conservative electorate, because more republicans would be saying they are independents.

    Not weighing by party ID has problems, because your results purely hinge on how random the sample was, and it’s possible some groups are over sampled.  While that has problems though, weighing by party ID has more problems.

  17. In the same poll he leads Mitt Romney 47-46. He got 54% in 2008. He gets less than that against every Republican polled except Sarah Palin.

  18. these toplines aren’t ridiculous (since Lynch won 3-to-1 in both 2006 and 2008), and the Obama-McCain breakdown squares with reality. Splitting hairs with crosstabs was false hope for our side, and it’s probably going to be false hope for your side too.

  19. 1 of 20 polls should, by definition, be an outlier.

    538 and DCCyclone have both mentioned this often.

    i think this is an outlier, i’d say move the vote 5 points towards the gop from here.

    but a few outliers is not only normal, it’s good.

  20. A little early in the cycle for that, don’t you think?

    As conspiracy points out, PPP’s results are consistent with other polls where comparable polls exist.

    There’s nothing wrong with PPP, they’re plenty credible.

    Just face it, a Presidential electorate is going to be like 2008.  Even if voter interest moved away from Obama a little, that’s offset by people of color naturally continuing to rise as a percentage of the electorate every 4 years.  Only in midterms will you continue to see a turnout model more familiar and comfortable to your party over the past several decades, but even that will move against the GOP over time.

    Yes I’m bragging on this one, but this is all exactly as I predicted last fall before the midterms.  I argued here that swing voters would get their anger out of their system on election day and then calm down.  That’s exactly what’s happened.

  21. It’s just saying who the respondents voted for in 2008, and the 9-point Obama advantage means the voter interest is about the same as in 2008.

    And that makes sense in this state more than most battlegrounds.  This is a swing voter-dominated state, not a state Obama could “adjust the electorate” as David Pluoffe likes to put it.  There are college students, but the colleges all are very small, and when you do the math I’m sure any heightened turnout for Obama was a very tiny bump in the state’s turnout.  Otherwise there are few minorities or other Democratic-favoring demographics to gin up.

  22. Well, yea, I do expect outliers – it was actually the GA poll from a few days ago that got me thinking they were not working with a great model, but time will tell.  Polling it hard work, I’m sure I wouldn’t be able to do it very well.

Comments are closed.