Is a "double bell curve" Clinton swing possible?

[Adam Griffin]

The national polls - more so as of late, but to a degree it is has been this way for months - seem out of line with battleground state polls. It seems to be a bigger phenomenon than in past elections. Most people assume one set of polls or the other must be off. Is it possible that both are correct?

Basically, is it possible that Clinton is going to overperform both in safe D states and safe R states, and that she will improve by much less compared to 2012 in battleground states (or even see Trump gain ground in battleground states as a whole)? This could lead to a stronger PV performance by Clinton than by Obama in 2012, but could result in the Electoral College being much closer than it was then.

Image to illustrate the notion; just a crude sketch to outline the broader principle. Don't hold the precise numbers in it as necessarily being indicative of my prediction or the reality.

[angus]

That looks more like a sinusoidally-varying function than a bimodal Gaussian distribution (aka "double bell curve").
 

[afleitch]

If you say, falling further 'behind' in states the Democrats used to be competitive in at a presidential level (WV, KY etc), states that her husband won, but make significant strides in say Texas, Utah and the Dakota's, where Bill never one, then yes I think it's possible.

[Camaro33]

I commend a fellow math-minded person for creating this thread. I can absolutely see this happening.

On the other hand, I can also envision a far more evenly distributed state-by state result, where many R states swing D and many D states swing R. I am pretty convinced that some safe R states will swing D by a large amount (Utah etc), but some safe D New England states my swing a handful of points right. I will not be surprised by any bit if Trump outperforms Romney in New England by several points.

[RaphaelDLG]

The logic doesn't make any sense, though.  Especially when you factor in that Clinton will have a huge cash advantage and will be POURING money into these battleground states.

There would have to be some demographic characteristic about the battlegrounds that makes them favorable for Trump, and I don't see anything in particular immediately apparent.  The battlegrounds aren't uniformly disproportionately white and blue collar, are they?

Seems more likely to me that one of the two (battleground averages or PV averages) is just simply less accurate at this point in the year because of the way the samples are conducted.  I remember NS saying something about that, I can't remember which one he said is historically more accurate at this point in the year.

[IceSpear]

The logic doesn't make any sense, though.  Especially when you factor in that Clinton will have a huge cash advantage and will be POURING money into these battleground states.

There would have to be some demographic characteristic about the battlegrounds that makes them favorable for Trump, and I don't see anything in particular immediately apparent.  The battlegrounds aren't uniformly disproportionately white and blue collar, are they?

Seems more likely to me that one of the two (battleground averages or PV averages) is just simply less accurate at this point in the year because of the way the samples are conducted.  I remember NS saying something about that, I can't remember which one he said is historically more accurate at this point in the year.

Yes, Occam's razor suggests that one or both of the state/national polls are wrong.

[Mr. Morden]

To what extent are the national and state polls actually disagreeing with each other though?  How much is just the fact that each given swing state is only polled intermittently, while the national polls come out every day?  So there's a time lag between them because for a given state, you might still be looking at a poll from three weeks ago as the "current" poll, while the national poll is from today?

[Seriously?]

To what extent are the national and state polls actually disagreeing with each other though?  How much is just the fact that each given swing state is only polled intermittently, while the national polls come out every day?  So there's a time lag between them because for a given state, you might still be looking at a poll from three weeks ago as the "current" poll, while the national poll is from today?

National polls are mainly RV at this point in time.
State polls are mainly LV. That's part of the difference.

Outside of Seltzer/Bloomberg (always not great nationally) and Reuters (Internet basically respondent poll), the margin for the rest of the polls suggest a Hillary! 3-7 point lead. Taking 2012 into account, the numbers are pretty much where they should be on the state level within your standard MOE and not a ton of data points. Obama won by 4+ percentage points in 2012.

Way too much is being made of sporadic polling before both the conventions and the consolidation of each base. Come November, we'd be looking at these June polls and laughing.

[Wisconsin+17]

TBH, I've been predicting Hillary +8 for awhile now.

[catographer]

Right now, I'm expecting this:

Clinton/Hickenlooper 325 EV 49.2%
Trump/Brown 213 EV 44.1%
Other 6.7%

I predict almost the exact same map. Also, note that the state polling was correct while the national polling was wrong in 2012.