NH-01/02 (WMUR): Shea-Porter in a tight race; Kuster with a strong lead

[NHLiberal]

NH-01:

Guinta 46%
Shea-Porter 43%

Shea-Porter 45%
Innis 38%

NH-02:

Kuster 45%
Lambert 36%

Kuster 49%
Garcia 35%

Kuster 47%
Lawrence 35%

Favorability Ratings
Shea-Porter: 43-31
Guinta: 32-27
Innis: 11-3
Kuster: 34-28
Lambert: 8-3
Garcia: 9-6
Lawrence: 5-3

https://cola.unh.edu/sites/cola.unh.edu/files/research_publications/gsp2014_summer_congraces070814.pdf

[free my dawg]

Junk firm.

[NHLiberal]

Junk firm.

No.

[free my dawg]

Junk firm.

No.

Uhh... It's commonly known throughout NH politics that WMUR isn't taken seriously. There's this, and the general fact that even Hassan's inner circle didn't take their poll they had that had her up by 10 seriously in 2012.

Something is gravely wrong with their methodology.

[Attorney General, LGC Speaker, and Former PPT Dwarven Dragon]

Junk firm.

No.

Uhh... It's commonly known throughout NH politics that WMUR isn't taken seriously. There's this, and the general fact that even Hassan's inner circle didn't take their poll they had that had her up by 10 seriously in 2012.

Something is gravely wrong with their methodology.

Perhaps Hassan didn't take it seriously simply because other pollsters were showing the race much closer. Hassan ended up winning by 12 in 2012, WMUR had her up 11 in their final poll, while PPP, Rasmussen, New England College, and NBC had the race within 5 points in their final polls. WMUR essentially got the new hampshire governor's race right in 2012, when everyone else was significantly off on the margin.

[free my dawg]

Junk firm.

No.

Uhh... It's commonly known throughout NH politics that WMUR isn't taken seriously. There's this, and the general fact that even Hassan's inner circle didn't take their poll they had that had her up by 10 seriously in 2012.

Something is gravely wrong with their methodology.

Perhaps Hassan didn't take it seriously simply because other pollsters were showing the race much closer. Hassan ended up winning by 12 in 2012, WMUR had her up 11 in their final poll, while PPP, Rasmussen, New England College, and NBC had the race within 5 points in their final polls. WMUR essentially got the new hampshire governor's race right in 2012, when everyone else was significantly off on the margin.

Results don't mean anything when methodology is bad. EPIC-MRA was also right, but early on they were as bad as the rest of the Michigan polls.

[Attorney General, LGC Speaker, and Former PPT Dwarven Dragon]

Junk firm.

No.

Uhh... It's commonly known throughout NH politics that WMUR isn't taken seriously. There's this, and the general fact that even Hassan's inner circle didn't take their poll they had that had her up by 10 seriously in 2012.

Something is gravely wrong with their methodology.

Perhaps Hassan didn't take it seriously simply because other pollsters were showing the race much closer. Hassan ended up winning by 12 in 2012, WMUR had her up 11 in their final poll, while PPP, Rasmussen, New England College, and NBC had the race within 5 points in their final polls. WMUR essentially got the new hampshire governor's race right in 2012, when everyone else was significantly off on the margin.

Results don't mean anything when methodology is bad. EPIC-MRA was also right, but early on they were as bad as the rest of the Michigan polls.

And what is the best judge of whether methodology is good or bad? Not how much things swing from poll to poll, as that can simply be due to a major development in a campaign, an error among a small number of a company's polls (in which the significant majority of polls do not have such error), major changes in the national enviornment, or simply a natural statewide tendency to have significant swings. No. The best judge of whether methodology is good is whether the poll,  1) is close to the actual election day result, and, if appropriate, 2) closely lines up with several other non-internal polls. Any methodology can be good as long as the poll ends up being close to the actual election day result and, if appropriate, lines up with other non-internal polls. So, let's go poll by poll here (2012 data used since we obviously don't know 2014 results yet):

2012 NH President:
Correct Winner? Yes
Points off actual margin (in final poll)? 3
Lined up with other polls? Yes

2012 NH House:
Correct Winner? Yes for both districts
Points off actual margin (in final poll)? 1 for district 1, 5 for district 2
lined up with other polls? No, but that was the fault of the other pollsters (district 1); No (district 2)

2012 NH Governor:
Correct Winner? Yes
Points off actual margin (in final poll)? 1
lined up with other polls? No, but that was the fault of the other pollsters

So, in 2012, WMUR did very well in NH Governor and House District 1. President wasn't too bad either. House District 2 leaves something to be desired, but at least they had the correct winner. Overall, not really that bad of a pollster and definitely not junk.

[Oldiesfreak1854]

Junk firm.

I don't think so.  NH-02 is a close district, so I wouldn't at all be surprised if she lost, especially if it turns out to be a strong GOP year.

[free my dawg]

Junk firm.

No.

Uhh... It's commonly known throughout NH politics that WMUR isn't taken seriously. There's this, and the general fact that even Hassan's inner circle didn't take their poll they had that had her up by 10 seriously in 2012.

Something is gravely wrong with their methodology.

Perhaps Hassan didn't take it seriously simply because other pollsters were showing the race much closer. Hassan ended up winning by 12 in 2012, WMUR had her up 11 in their final poll, while PPP, Rasmussen, New England College, and NBC had the race within 5 points in their final polls. WMUR essentially got the new hampshire governor's race right in 2012, when everyone else was significantly off on the margin.

Results don't mean anything when methodology is bad. EPIC-MRA was also right, but early on they were as bad as the rest of the Michigan polls.

And what is the best judge of whether methodology is good or bad? Not how much things swing from poll to poll, as that can simply be due to a major development in a campaign, an error among a small number of a company's polls (in which the significant majority of polls do not have such error), major changes in the national enviornment, or simply a natural statewide tendency to have significant swings.

Having lived under Shea-Porter for a good portion of her tenure, I can say that she's not the best campaigner, and while she's relatively close to me ideologically, someone like me will have a tough time getting elected to Congress in NH-1. Frank Guinta is a corrupt Tea Partier, but that's been well-known for a while and cost him his Senate seat. They're both bad, but they aren't Akin/Mourdock tier. There's no remote explanation for each candidate to gain the support of that much of the electorate in a span of three months.

No. The best judge of whether methodology is good is whether the poll,  1) is close to the actual election day result, and, if appropriate, 2) closely lines up with several other non-internal polls. Any methodology can be good as long as the poll ends up being close to the actual election day result and, if appropriate, lines up with other non-internal polls.

How you get to the result is different from getting the right result. If you literally make numbers out of thin air and they happen to be right, that's a junk firm. If you make a poll, model the electorate exactly after 2010 or 2012, and say "F(inks) it, close enough", but get it right, that's a junk firm. And if your poll firm shifts from candidate to candidate like Johnny Manziel shifts from woman to woman, then that's a junk firm.

Junk firm.

I don't think so.  NH-02 is a close district, so I wouldn't at all be surprised if she lost, especially if it turns out to be a strong GOP year.

Garcia and Lambert are too conservative for the district. Despite being elected in a D+4 district, Lambert was the second most conservative member of the Senate in his tenure, and Garcia is Bill O'Brien's most famous and one of his staunchest allies. If Odell, Forrester, or Hemingway ran, they could make a run at it, but I don't see Kuster losing, especially considering she kept it close in an open 2010 race.

[Attorney General, LGC Speaker, and Former PPT Dwarven Dragon]

Junk firm.

No.

Uhh... It's commonly known throughout NH politics that WMUR isn't taken seriously. There's this, and the general fact that even Hassan's inner circle didn't take their poll they had that had her up by 10 seriously in 2012.

Something is gravely wrong with their methodology.

Perhaps Hassan didn't take it seriously simply because other pollsters were showing the race much closer. Hassan ended up winning by 12 in 2012, WMUR had her up 11 in their final poll, while PPP, Rasmussen, New England College, and NBC had the race within 5 points in their final polls. WMUR essentially got the new hampshire governor's race right in 2012, when everyone else was significantly off on the margin.

Results don't mean anything when methodology is bad. EPIC-MRA was also right, but early on they were as bad as the rest of the Michigan polls.

And what is the best judge of whether methodology is good or bad? Not how much things swing from poll to poll, as that can simply be due to a major development in a campaign, an error among a small number of a company's polls (in which the significant majority of polls do not have such error), major changes in the national enviornment, or simply a natural statewide tendency to have significant swings.

Having lived under Shea-Porter for a good portion of her tenure, I can say that she's not the best campaigner, and while she's relatively close to me ideologically, someone like me will have a tough time getting elected to Congress in NH-1. Frank Guinta is a corrupt Tea Partier, but that's been well-known for a while and cost him his Senate seat. They're both bad, but they aren't Akin/Mourdock tier. There's no remote explanation for each candidate to gain the support of that much of the electorate in a span of three months.

No. The best judge of whether methodology is good is whether the poll,  1) is close to the actual election day result, and, if appropriate, 2) closely lines up with several other non-internal polls. Any methodology can be good as long as the poll ends up being close to the actual election day result and, if appropriate, lines up with other non-internal polls.

How you get to the result is different from getting the right result. If you literally make numbers out of thin air and they happen to be right, that's a junk firm. If you make a poll, model the electorate exactly after 2010 or 2012, and say "F(inks) it, close enough", but get it right, that's a junk firm. And if your poll firm shifts from candidate to candidate like Johnny Manziel shifts from woman to woman, then that's a junk firm.

This isn't a math class in which the teacher picks their favorite method of doing a problem and then makes everyone do it that way on the test to get the points for a specific question.  No, this is polling, in which it doesn't matter whether the methodology used by a polling company is your favorite, favorite way to concoct a poll or not, as long as it gets the right result. As long as you get close to the actual election day result and show the correct winner, I see no reason why it should matter whether a polling company is perceived to be D-Leaning or R-Leaning or whether it has my favorite methodology or not. To be honest, it's not as if you're the only person judging polls in this manner - Nate Silver, of all people, sees PPP as unreliable because he doesn't like its methodology, despite PPP's excellent track record of being close to the actual results. But that doesn't change the fact that saying "hmmm, I don't like this methodology! Goodbye!" is a very poor way to judge polls. Obviously, if a pollster literally just makes up numbers out of thin air without actually surveying anyone and gets it right, then they are a junk firm who got lucky. But if a pollster can get the results right with a methodology that seems dubious (to some), as WMUR and PPP can, that makes them, while not necessarily the best pollster in the world, at least more of a good pollster than a bad pollster. When I see a poll, I ask myself three questions:

Is it an internal?
Does it line up with other polls?
How has it performed in the past in terms of correctly predicting results?

As long as a poll is 'no' on the first question, and 'yes' on the other two, I include it in my average that I use to help predict the midterms (Senate, Governor races(not really doing the house)). Depending on the situation, I may also include polls that answer 'no' to one of the questions as long as it is not an internal. And that's literally all I consider. I don't care how a poll gets to an answer, as long as it has the right answer, and wild swings seems no reason to think otherwise, as it certainly does not always mean that a pollster is bad (as I explained previously), and NH is naturally prone to wild swings anyways. If the 2014 results come in and WMUR does terrible, then we have every right to disregard it in 2016. But WMUR's 2012 record, as I detailed previously, does not provide significant evidence to declare it a bad pollster.

[free my dawg]

No. The best judge of whether methodology is good is whether the poll,  1) is close to the actual election day result, and, if appropriate, 2) closely lines up with several other non-internal polls. Any methodology can be good as long as the poll ends up being close to the actual election day result and, if appropriate, lines up with other non-internal polls.

How you get to the result is different from getting the right result. If you literally make numbers out of thin air and they happen to be right, that's a junk firm. If you make a poll, model the electorate exactly after 2010 or 2012, and say "F(inks) it, close enough", but get it right, that's a junk firm. And if your poll firm shifts from candidate to candidate like Johnny Manziel shifts from woman to woman, then that's a junk firm.

This isn't a math class in which the teacher picks their favorite method of doing a problem and then makes everyone do it that way on the test to get the points for a specific question.  No, this is polling, in which it doesn't matter whether the methodology used by a polling company is your favorite, favorite way to concoct a poll or not, as long as it gets the right result.

Polling involves statistics. Opinion surveys have something to do with mathematics, and UNH's methodology problems have been well-documented for years.

To be honest, it's not as if you're the only person judging polls in this manner - Nate Silver, of all people, sees PPP as unreliable because he doesn't like its methodology, despite PPP's excellent track record of being close to the actual results. But that doesn't change the fact that saying "hmmm, I don't like this methodology! Goodbye!" is a very poor way to judge polls.

That's a false analogy. Silver said PPP's methodology of not releasing an unfavorable poll to the public (according to them they didn't feel it was accurate) was bad. I'm saying how WMUR's methodology is bad because it has wild swings, most likely because they have a minuscule sample size of about 250-300 voters.