Election Thoughts... The home straight...

So the debates are over.  I thought last night that, like the last one, Obama completely destroyed Romney.  Apparently that means Romney won in the post-truth universe of the GOP, but never mind.  I was edumacated by Romney...  for a start I didn't know that Iran was landlocked...

Anyway.  There's some interesting patterns emerging to this embedded observer.  Firstly the 'horse race' narrative.

This has been a common meme of the media, even the 'lame stream' media.  When it looked like Obama was walking this, they brought it back to the horserace.  And now, it's all about the neck and neck horserace and whether or not Romney's 'momentum' can take him over the top.

Secondly, there's the concept of 'momentum' - in that Romney has it.  This isn't all that clear from the data.  The race was always going to tighten.  I'm not convinced that even if Obama had been more forthright in the first debate, the polls wouldn't be far from where they are.  The reason is simple.  Republicans came home to back the candidate and, frankly, they were always going to.  The real damage was if Democrats couldn't be bothered.  I think that's been solved, and looking at the early polling numbers, there is proof in that pudding.

Finally, there's the nature of the polls.  The polls are currently, charitably, all over the place.  However, there's an interesting thing emerging which, I suspect will have a profound impact on the nature of polling moving forward.  I'll look at the major types of polls and the weird one.

Weird first: Gallup.  There's no polite way to phrase this but Gallup is a mess right now.  They've got Romney up by 6 with 'likely voters'.  He's not.  They have the race roughly tied when they look at Registered Voters.  That smacks to me of a problem with their screen for likely voter screen, because there are other polls which have similar deltas.  Verdict: Outlier, the race isn't that wide.  Interestingly though, +6 is how much tilted towards the Republicans their results were in the 2010 mid-terms...

Telephone RoboPolls: Rassmussen, PPP, Gravis, Purple - these are polls that automatically dial landlines.  These mostly have the race bumping around a tie, with a Romney edge (Ras puts R+4 today, but more on that later) That's the first problem right there.  The proportion of the population with landlines is reducing all the time, plus even with landlines participation is hovering around 10%.  I think this will be the last election where polls like this are even looked at.  Rassmussen muddies the water even more by applying a weighting to the results based on what they think the electorate looks like.  It's typically accepted that this gives their polls a +2/+3 lean towards GOP.

Internet Polls: YouGov, TPP, Ipsos, Google - these are all showing good numbers for Obama +2/+3 nationally.  These are currently discounted by most of the poll trackers as they're suspicious of the technology.  I suspect that we'll see that fade in another 4 years as more polling moves to online.

Live Calls including cellphones: Marist, Quinnipac, the major network polls etc... - these are typically running tie to O+2 nationally and give Obama his best numbers in the swing state polls.  These are probably the most accurate as they capture the widest input.  The problem is, for the poll hungry media, these are also the most expensive and there are less of them.

The core problem for Romney remains that while the race has got tighter, Obama has a slight edge, 2 weeks out, in the main swing states, and, so far, ignoring the robopolls, he's holding that.  It's coming down to 1 state - if Obama wins Ohio, he wins the White House.  If he doesn't...  he might still win the White House but it depends on Colorado or New Hampshire(*)

(*) I'll caveat that while things look dodgy for Obama in Virginia and Florida, he's tied in all the 'real' polls there too.

Finally: My prediction from a few days ago holds.  Obama nationally by about 1%, with 288(ish) EV.