NH: Rasmussen - Romney 48%, Obama 45%

[MorningInAmerica]

Party ID is 27%D/27%R/45%I

Obama's job rating is 49/51%
Romney's favs are 50/50

Obama: 45% (-3)
Romney: 48% (+5)

[big bad fab]

At last, some more suspense in this boring election year ?

After all, on NH, maybe ARG wasn't so bad

[LiberalJunkie]

At last, some more suspense in this boring election year ?

After all, on NH, maybe ARG wasn't so bad

Both ARG and Rasmussen polls are bad.

[Franzl]

Worth keeping our eyes on, I suppose. Romney has lived there for the last 6 years practically... Still doubt it though.

[Fmr. Pres. Duke]

Romney needs this state if he is going to have any chance at all.

[xavier110]

I'm.....dubious

[ElectionAtlas]

New Poll: New Hampshire President by Rasmussen on 2012-09-18

Summary: D: 45%, R: 48%, I: 4%, U: 3%

Poll Source URL: Full Poll Details

[ajc0918]

Romney needs this state if he is going to have any chance at all.

It would help, but he can win without it.

[Southern Senator North Carolina Yankee]

Worthless without OH, VA and FL. With all three though, it gets him exactly 270.

[Niemeyerite]

Junk, I guess.

[SPC]

I'm not sure what would be funnier, Romney losing the election but winning NH as a "consolation prize" for spending the last six years there, or Romney losing the election and NH.

[Snowstalker Mk. II]

I suspect that after IN and NC, NH is the most likely state to flip from 2008.

[Eraserhead]

Obama's approval is 49% but he only gets 45% of vote. Probably not, Scott.

[ucscgaldamez]

interesting fact..about the internals...Obama leads whites by 2%, yet trails Romney by 3. Blacks vote 100% for Romney and 70% of other races prefer Romney to Obama 11%.  It is likely that Obama is really leading the state by 4.

[Badger]

interesting fact..about the internals...Obama leads whites by 2%, yet trails Romney by 3. Blacks vote 100% for Romney and 70% of other races prefer Romney to Obama 11%.  It is likely that Obama is really leading the state by 4.

I think you just cracked the code here. Good eye.

[Landslide Lyndon]

Scotty keeps fighting the good fight.

[Bacon King]

I love how Rasmussen's own methodology even openly admits their polls are garbage nowadays, if you just read between the lines ever so slightly.


Here's how a Rasmussen poll is made!!!

• Poll conducted during a four-hour period of a single weekday evening
• short time means no calling back any numbers
• Robot voice asking questions and telling you to press buttons
• Questions asked of whoever picked up the phone; could easily be small child
• Calls are only made to landlines
• Instead of, you know, calling cell phones, they channel inspiration from Zogby: "To reach those who have abandoned traditional landline telephones,  Rasmussen Reports uses an online survey tool to interview randomly selected participants from  a demographically diverse panel."
  
  Let's see. At this point, they have data showing which buttons were pressed by whoever they called that evening who happened to be home, who decided to answer that phone call, and who didn't hang up when they heard a recording on the line. They add this data to internet surveys taken by people who deign to sign up to be part of the "demographically diverse" Rasmussen internet panel and trust none of these internet folks lied about any of their personal information.
  
  But wait, now the fun part!
  
  
• The sample isn't representative in the slightest, so it gets adjusted to match census data for the area!
• Still not representative enough, so it gets adjusted again to match voter registration records!
• Still not good enough, so it goes through a likely voter screen! Only the people who pushed the correct buttons when the robot voice asked them about their voting habits make it through this part!
• Almost good, but still one problem- it has to be adjusted for partisanship! This means they use the magical dynamic weighting system which uses "the state’s voting history, national trends, and recent polling in a particular state or geographic area" to figure out exactly how to BS where the numbers should have been if this was anything resembling a scientific poll.
  
  So, to recap: Have robots blanket the states with calls for a few hours while everyone's out eating dinner or watching football, record the answers from whoever does pick up the phone, mix that data with some totally legit internet surveys, arbitrarily readjust the numbers four times in a row, and presto! You have your very own Rasmussen poll!
  
  (source: http://www.rasmussenreports.com/public_content/about_us/methodology)

[minionofmidas]

Tbh, no poll is much better.

[J. J.]

interesting fact..about the internals...Obama leads whites by 2%, yet trails Romney by 3. Blacks vote 100% for Romney and 70% of other races prefer Romney to Obama 11%.  It is likely that Obama is really leading the state by 4.

The black population is about 1.1% of the state, so it's probably a really low sample for that subgroup, like about 5 people.

[Bacon King]

Tbh, no poll is much better.

True, but Rasmussen is a unique case because they include virtually every possible design flaw in their methodology.

[Craigo]

Tbh, no poll is much better.

True, but Rasmussen is a unique case because they include virtually every possible design flaw in their methodology.

I agree. An lack of cell phone voters in the IVR sample, the internet panel may have selection and response bias, the party ID weighting is methodologically suspect, and the 15% subsample is probably too small anyway - it's no shock that Rasmussen results look skewed.

This isn't just cost of doing business stuff that affects every pollster, like random error and maybe a problem with the likely voter screen.

[Likely Voter]

With the race where it is now, these periodic Romney leads in Rasmussen polls are plausible, if not likely. But what happens in October of Obama moves to a bigger national lead? Will Ras follow along, or ratchet down his LV model again (with even more whites and olds) to keep putting out periodic "good news" polls for his customers.

[Craigo]

With the race where it is now, these periodic Romney leads in Rasmussen polls are plausible, if not likely. But what happens in October of Obama moves to a bigger national lead? Will Ras follow along, or ratchet down his LV model again (with even more whites and olds) to keep putting out periodic "good news" polls for his customers.

If history is any guide, his numbers will begin to conform to the aggregate sometime in October. 

[Likely Voter]

With the race where it is now, these periodic Romney leads in Rasmussen polls are plausible, if not likely. But what happens in October of Obama moves to a bigger national lead? Will Ras follow along, or ratchet down his LV model again (with even more whites and olds) to keep putting out periodic "good news" polls for his customers.

If history is any guide, his numbers will begin to conform to the aggregate sometime in October. 

That's true and if he did push the model too far his final realignment to closer reality could end up looking like a huge Obama surge (that only he is seeing)

[Torie]

Bacon King, doesn't every poll re-weight, because white olds typically answer land lines more often?