Friday, July 20, 2012

The Electoral College Map (7/20/12)

This week has been full of surveys from mostly blue states -- at least as measured by FHQ graduated weighted average -- and the handful of polls released on Friday was no exception.

But before I get into what those polls mean -- if anything -- I did want to revisit, as promised, the polls from a day ago. As FHQ mentioned on Thursday, none of the polls did anything to shake up anything on any of the graphics we publish every day. However, I did want to make a comment or two about how some -- well, especially the Virginia survey released by Quinnipiac -- were discussed upon release. [These operate more as rules of thumb for how to follow along here than as comments specific to the Q-poll of the Old Dominion.]
1. One of the emerging problems, or what can, perhaps, be called the greatest emerging pet peeve of mine in 2012 poll watching, is the tendency of some commentators/pundits/whoever to flashback to the last poll conducted by the same polling outfit. Now, to be honest, there is something responsible in that. If one wants to find the next best comparison for one poll, an argument can be advanced that earlier polls by the same company are a better place to look than more recent polls by other firms. The problem is that those comparisons often come with little or no context and certainly is rarely in-depth enough to mention polling variability.  
Take that Virginia poll:
Some folks wanted to track the decreasing lead from March in a Q-poll conducted then; an 8 point drop overall. [Set aside for the moment the fact that Politico transposed the 2 (placing it in a fictitious 52 for Obama instead of 42 for Romney) and 0 (inserting it in a 40 for Romney instead of a 50 for Obama in March) and increased the decline from 8 to 12. ...and put it in a headline!] Again, this is fine on some level. It is definitely true that the Obama lead shrank -- or Obama's share shrank while Romney's grew is the more accurate description maybe -- but the context is missing. Yes, the survey was in the field a week after the Virginia primary, but it was also being conducted in the midst of and after the Alabama and Mississippi primaries that Romney lost. Now, FHQ made the case at the time that Romney was still going to be the nominee, but those losses did up the uncertainty in the race, shifting the perceived outcome from "Romney's probably going to win the nomination" to "Romney's still probably going to win the nomination, but how long is this going to take and how much damage will it do to him by the time it is over". Look, I'm not saying that this is necessarily what is driving the change, but I would argue that it matters. the extent that such things imperil the interpretability of a mid-March, state-level horserace poll anyway. John Sides wasn't tweeting about this poll, but his comments that there is too much attention being paid to single polls right now and that that's what polling averages are for is the right way to think about this. And hey, let's be honest, FHQ has a pretty high bar that is likely never to be achieved in terms of how much context makes the cut in any given analyses. Not everyone is as mind-numbingly interested in all of this. But context matters in every poll comparison.
2. Let me start this next comment by saying that I may be the only person in higher education who references Chappelle's Show in class. That is a function of the fact that 1) no one who wants to keep their job does this unless they do so very carefully and 2) the show has been off the air for a hundred years. Nevertheless, as some were attempting to make a mountain out of a molehill with this one Virginia poll, the same Chappelle's Show scene kept popping into my head. If you watched the show there was this one sketch where Dave riffs off the idea that he is perhaps irrationally skeptical of any evidence that is part of the trial of a black celebrity. In one scene he envisions himself as a prospective juror  in the OJ Simpson trial and when confronted with the evidence from the trial has one simple reaction: "Sir, I'm not impressed." 
Now sure, I just sucked all of the humor out of that -- and if you've seen that sketch just chuckled softly for a moment and read on -- but the line is an important one. I'm not impressed. My point is, don't write a "this poll is evidence of a big change in this race" sort of story/analysis unless you are going to actually demonstrate a change in the presidential horserace.  
The Virginia Q-poll? Sir, I'm not impressed.   
Again, context. [AGAIN!?!] First, the presidential race continues to be remarkably stable; the sort of stable in both national and probably to a lesser extent in the state level surveys (due to fewer polls) that is antithetical to an 8 point swing. Secondly, does anyone think that Virginia is +8 Obama? If we follow the political science pendulum swing model (see 7/17/12 post), then a +7 Obama state in 2008 is probably not +7 in 2012 if the pendulum has swung away from the president and his party. 
The battle of this election will be over North Carolina (maybe) and the handful of current light blue states unless there is sustained evidence that the Romney has been successful in stretching the battlefield into the Lean Obama states. At this point we don't have that kind of evidence and until such time that such evidence is revealed, I'm not impressed. Virginia or anywhere else. 

On to today's polls:

New State Polls (7/20/12)
Margin of Error
Poll Margin
FHQ Margin
+/- 3.9%
647 likely voters
North Carolina
+/- 4%
600 registered voters
+/- 4.5%
500 likely voters
+/- 4%
630 registered voters

Poll Quick Hits:
Outlier status for the time being. Not even the Priorties USA poll out last week had Obama with this large of a lead in the Sunshine state.

North Carolina:
Not an outlier. There are an awful lot of polls that have shown the Tarheel state to be within 1-3 points either way. But a slight majority of them have tilted toward Mitt Romney.

The Keystone state continues to be solidly lodged in Lean Obama territory. That said, this Rasmussen poll has Romney at his highest share in any survey since a February survey the same company conducted. With few exceptions Obama has been in the 45-49% range all along in Pennsylvania. This poll does not deviate from that pattern. Typically narrowing? No, it's a little early to break out the Jim Campbell references.

The margin is in line with other polls, but the two candidates shares in this Survey USA poll are a little off from what they have been in the Evergreen state. This represents Obama's lowest share in any of the eight Washington state surveys conducted thus far this year. Romney's share had climbed into the low 40s in the three most recent polls stretching back into May but dips here into the upper 30s where it had been in polls released during the first third of the year.

The map remains unchanged above as does the Electoral College Spectrum below. Yet, I don't want to totally skip over either. Mark Blumenthal at HuffPo Pollster had a great read on the tradeoffs of using registered voter screens in polls versus likely voter screens at this point in the race. The piece was prompted by the argument elsewhere that a switch from the former to the latter would shift things toward Romney. How much? Well, Blumenthal understandably left that somewhat ambiguous. It is in inexact science to attempt to accurately project how much things might shift. So it was left at a shift of a "few" percentage points.

Fine, but what would that mean? A two point shift uniformly applied across all states would push just Florida from Obama to Romney in FHQ's averages; moving 29 electoral votes to the former Massachusetts governor. The tally would then be 303-235 electoral votes in the president's favor. Now, if that "few" translated into a four percentage point shift to Romney, Iowa, Virginia, Colorado and Ohio would all join Florida in moving over to the Romney side of the ledger. That's enough to inch Romney over the 270 electoral vote barrier, but only just barely with no states to spare. Ohio is, after all, the tipping point state in the order at the moment.

The Electoral College Spectrum1
1 Follow the link for a detailed explanation on how to read the Electoral College Spectrum.

2 The numbers in the parentheses refer to the number of electoral votes a candidate would have if he won all the states ranked prior to that state. If, for example, Romney won all the states up to and including Ohio (all Obama's toss up states plus Ohio), he would have 272 electoral votes. Romney's numbers are only totaled through the states he would need in order to get to 270. In those cases, Obama's number is on the left and Romney's is on the right in italics.

3 Ohio
 is the state where Obama crosses the 270 electoral vote threshold to win the presidential election. That line is referred to as the victory line.

And the Watch List? Washington on, Florida off. Keep in mind, however, that it was an outlier poll in the Sunshine state that drove Florida off the list.

The Watch List1
from Strong Romney
to Lean Romney
from Toss Up Obama
to Lean Obama
from Toss Up Romney
to Lean Romney
from Lean Obama
to Toss Up Obama
New Hampshire
from Toss Up Obama
to Lean Obama
New Mexico
from Strong Obama
to Lean Obama
North Carolina
from Toss Up Romney
to Toss Up Obama
from Strong Obama
to Lean Obama
West Virginia
from Strong Romney
to Lean Romney
1 Weighted Average within a fraction of a point of changing categories.

Are you following FHQ on TwitterGoogle+ and Facebook? Click on the links to join in.

No comments: