07 October 2012

Battleground Report 10/06/2012 - T-minus 31 days, plus National Polling


Saturday, October 6, 2012, was exactly 31 days before the General Election on November 6, 2012.

There was no BATTLEGROUND REPORT from 31 days before the GE 200, but rather Poll Convergence No. 9, and the PC 9 "Distillation" (main report), both from 10/04/2008.

Complete polling results for all states for 2012 to-date are here as an EXCEL table for 10/06/2012 in GOOGLE DOCS. All polls are hyperlinked.

The BATTLEGROUND REPORTS run also with the national numbers. Here is the NATIONAL POLLING OUTLOOK thread, with links to every day going back to September 13, 2012. Want to see the numbers for a specific day? Go to that link.

The LEGEND for the following EXCEL table is HERE. In fact, you need to read it in order to develop an "eye" for the table:


A B C D E E G H I J K L M
State EV % of EC Polls 06.10.12 05.10.12 Shift 2008 2004 SWING (H-I) End polling 2008 Compare E – K Compare E – GE008
GA 16 2,97% 0 / 2 13,50 13,50 0,00 5,20 16,60 11,40 3,85 9,65 8,30
NE 5 0,93% 0 / 10 11,00 11,00 0,00 14,93 33,22 18,29 19,00 -8,00 -3,93
IN 11 2,04% 0 / 3 8,00 8,00 0,00 -1,03 20,68 21,71 1,18 6,82 9,03
AZ 11 2,04% 0 / 5 7,20 7,20 0,00 8,48 10,47 1,99 4,43 2,77 -1,28
TN 3 0,56% 0 / 1 7,00 7,00 0,00 15,06 14,27 0,79 14,00 -7,00 -8,06
MT 3 0,56% 1 / 2 6,50 7,00 -0,50 2,38 20,50 18,12 1,64 4,86 4,12
MO 10 1,86% 0 / 6 5,70 5,70 0,00 0,13 7,20 7,07 0,46 5,24 5,57
-- --
-- -- -- -- -- -- -- -- -- --
NE-02 1 0,19% 0 / 1 0,00 0,00 0,00 1,21 21,69 20,48 4,00 -4,00 -1,21
NC 15 2,79% 0 / 9 0,44 0,44 0,00 0,33 12,43 12,76 0,62 -0,18 0,11
VA 13 2,42% 0 / 7 2,29 2,29 0,00 6,30 8,20 14,50 4,93 -2,64 -4,01
FL 29 5,39% 0 / 11 2,33 2,33 0,00 2,81 5,01 7,82 1,79 0,54 -0,48
CO 9 1,67% 0 / 5 2,70 2,70 0,00 8,95 4,67 13,62 6,62 -3,92 -6,25
-- --
-- -- -- -- -- -- -- -- -- --
USA 538 100,00% 5 / 18 2,98 3,54 -0,56 7,26 2,46 9,72 7,54 -4,56 -4,28
-- --
-- -- -- -- -- -- -- -- -- --
IA 6 1,12% 0 / 8 3,98 3,98 0,00 9,53 0,67 10,20 12,75 -8,77 -5,55
NV 6 1,12% 0 / 6 4,93 4,93 0,00 12,49 2,59 15,08 6,83 -1,90 -7,56
-- --
-- -- -- -- -- -- -- -- -- --
OH 18 3,35% 0 / 10 5,09 5,09 0,00 4,58 2,11 6,69 2,30 2,79 0,51
NH 4 0,74% 0 / 7 6,50 6,50 0,00 9,61 1,37 8,24 10,43 -3,93 -3,11
MN 10 1,86% 0 / 3 8,00 8,00 0,00 10,24 3,48 6,76 11,34 -3,34 -2,24
WI 10 1,86% 1 / 4 8,65 7,09 1,56 13,90 0,38 13,52 11,34 -2,69 -5,25
PA 20 3,72% 0 / 10 8,79 8,79 0,00 10,31 2,50 7,81 11,34 -2,55 -1,52
OR 7 1,30% 0 / 1 9,00 9,00 0,00 16,35 4,16 12,19 15,57 -6,57 -7,35
MI 16 2,97% 0 / 9 9,26 9,26 0,00 16,44 3,42 13,02 14,44 -5,18 -7,18
NM 5 0,93% 0 / 5 9,86 9,86 0,00 15,13 0,79 15,92 9,26 0,60 -5,27

10/06 was a light polling day, as expected.

A poll from Montana came in very late on 10/05 and I have recorded it for 10/06. It shows Romney +4. But  an older poll that just fell out of the statistic showed Romney +5, so the statistical margin average moved just ever so slightly down, to Romney +6.50.

One poll came in for Wisconsin, from PPP (D), showing the race tightening greatly, down to Obama +2. However, the statistical average actually went UP. This can happen, as four polls from 09/20 that were still in the statistic as of 10/03, all fell out, and two of those polls also had Obama with very narrow leads (+2 and +3, respectively). So, four polls fell out of the statistics, 1 came in, and the polling gene poll fell from 7 to 4. And as a result, this actually moved Obama's average UP to +8.65.

I am going to illustrate the statistics from both 10/06 AND 10/03 (which was static through 10/04-10/05), so that people can see that no trickery or sleight of hand is at work here. Here is the current statistic, from 10/06:

The polls that are now in the "gene pool" are highlighted in yellow. The polls that WERE in the gene poll at last count are in grey. I also included two more polls in white, already out of the statistic before 10/03, just for comparison purposes:

Nr. WISCONSIN Date Sample MoE Obama Romney Und Mar.

AVERAGE N/A N/A N/A 50,85 42,20 6,95 8,65


















41 PPP (D) 06.10.12 917 LV +/-3.1 49 47 4 2
40 Marquette 03.10.12 1003 RV +/-3.3 53 42 5 11
39 Zogby / JZanalytics 03.10.12 414 LV +/-4.9 48,9 38,8 12,3 10,1
38 WAA 24.09.12 1238 LV +/-2.8 52,50 41,00 6,50 11,50
37 NBC / WSJ / Marist 20.09.12 968 LV +/-3.1 50 45 5 5
36 Rasmussen 20.09.12 500 LV +/-4.5 49 46 5 3
35 YouGov 20.09.12 804 RV +/-4.5 48 46 6 2
34 PPP (D) 20.09.12 842 LV +/-3.4 52 45 3 7
33 Marquette 19.09.12 705 RV +/-4.1 54 40 6 14
32 Quinnipiac / NYT / CBS 19.09.12 1485 LV +/-3.0 51 45 4 6

So, a PPP (D) Obama +2 is very, very narrow, but it essentially replaced a YouGov Obama +2 that had been in the mix since 09/20.

Remember my rules for averages - until three weeks before the election: a two-week period, exactly fourteen days, backwards from the date of the most recent poll, no repeaters.

Here is how Wisconsin looked on 10/03:

Nr. WISCONSIN Date Sample MoE Obama Romney Und Mar.

AVERAGE N/A N/A N/A 50,49 43,40 6,11 7,09


















40 Marquette 03.10.12 1003 RV +/-3.3 53 42 5 11
39 Zogby / JZanalytics 03.10.12 414 LV +/-4.9 48,9 38,8 12,3 10,1
38 WAA 24.09.12 1238 LV +/-2.8 52,50 41,00 6,50 11,50
37 NBC / WSJ / Marist 20.09.12 968 LV +/-3.1 50 45 5 5
36 Rasmussen 20.09.12 500 LV +/-4.5 49 46 5 3
35 YouGov 20.09.12 804 RV +/-4.5 48 46 6 2
34 PPP (D) 20.09.12 842 LV +/-3.4 52 45 3 7
33 Marquette 19.09.12 705 RV +/-4.1 54 40 6 14
32 Quinnipiac / NYT / CBS 19.09.12 1485 LV +/-3.0 51 45 4 6

Some may call this a statistical quirk, but it is not. This is actually the advantage of a two-week polling window (which will be reduced down to one week for the battlegrounds starting 21 days before the election) - stability in the averages, but sometimes unexpected statitistical bumps happen. Of the four polls that fell out of the statistic between 10/03 and 10/06, three had margins under the average for 10/03. When they were gone, and there were no less polls in the mix, the average sprung up. Now, some may want to criticize this, but if a slew of polls come in in the next days confirming the PPP Obama +2, then this average will drop considerably fast anyway as new polls with lower values enter the statistics.

And finally, it is important to remember, as with all polls, not just for battleground states, that one poll does not paint the entire picture. If Romney is indeed surging, then the vast majority of the next polls will show it, and then THOSE polls which reflect his surge will be in the statistic for a while as well, and so on, and so on.

So, no trickery, no sleight of hand. Just pure math. And this can happen when you get such a dump of polls all on one day for one state as we had in Wisconsin on 09/20: heavy data-load in, heavy data-load out.

-----------------------------------------------------------

The national numbers:

Nr. NATIONAL POLLING Date Sample MoE
Obama Romney Und/Oth Mar.

AVERAGE: N/A N/A N/A
48,54 45,56 5,90 2,98











RECENT:

















540 Clarus (R) - post-debate (combined summary poll) 06.10.12 590 LV +/-4.3
46 47 7 -1
539 Clarus (R) - pre-debate (comined summary poll) 06.10.12 590 LV +/-4.3
49 45 6 4
538 Ipsos/Reuters 06.10.12 1492 LV +/-2.9
47 45 8 2
537 Gallup * 06.10.12 3050 RV +/-2.0
49 46 5 3
536 Rasmussen 06.10.12 1500 LV +/-3.0
47 49 4 -2
532 McLaughlin (R ) /Conservative.org 04.10.12 1000 LV +/-3.1
49 45 6 4
530 YouGov 04.10.12 1000 A +/-4.7
49 44 7 5
533 UPI / CVOTER 03.10.12 1000 RV +/-3.5
49 46 5 3
530 NPR / DemCorps (D) / Resurgent Republic (R) 03.10.12 800 LV +/-3.5
51 44 5 7
527 United Technologies / National Journal 02.10.12
+/-3.7
47 47 6 0
526 NBC / WSJ 02.10.12 832 LV +/-3.4
49 46 5 3
525 PPP (KOS / SEIU) 02.10.12 1100 RV +/-2.8
49 45 6 4
521 Merriman Group 01.10.12 981 RV +/-3.1
46 43 11 3
520 ARG 01.10.12 800 LV +/-3.6
49 46 5 3
519 ABC / WaPo 01.10.12 813 LV +/-4.0
49 47 4 2
517 CNN / ORC 01.10.12 783 LV +/-3.5
50 47 3 3
515 Zogby (online) 01.10.12 800 LV +/-3.5
49,7 41,1 9,2 8,6
514 Politico / GWU / Battleground 01.10.12 1000 RV +/-3.1
49 47 4 2

Obama's national margin has dropped decisively, and in just one day. Romney is indeed surging. But we have a problem with Clarus (R). I know, my rules say "no repeaters", but Clarus itself released both of these polls, one taken on 10/02, the other taken on 10/04, all on the same day. Alone, each of these polls, which were meant to go together, do not have a big enough sample group to be considered viable as national polls: just 590 LV in a sample in a nation of 311,000,000 people just doesn't cut it, and Claris knows this as well.  On it's website, Clarus gave the following information:

"October 5, 2012
Obama lead slips post-debate -- New state polls show tighter races -- McCrory, Nixon ahead in NC, MO

YOUR DAILY POLLING UPDATE

OBAMA APPROVAL: CLARUS AVERAGE 50%
(up 1 from yesterday; up 1 from a week ago)

PRESIDENTIAL NATIONWIDE: CLARUS AVERAGE*

Obama over Romney +2 


Note: Most of the interviews used for this average were conducted before Wednesday's debate.

CLARUS POLL: BEFORE vs. AFTER DEBATE**
(Nationwide survey of likely voters, MOE +/- 4.03%)

BEFORE (conducted Tuesday): Obama over Romney +4
AFTER (conducted Thursday): Romney over Obama +1 "

So, as you can see, Clarus itself intended to publish the results together and indicates simply a +2 overall for Obama. To just include one of these two polls in the average would therefore not be in line with Clarus' own Summary, and so I am, as an exception, recording two poll results for one pollster in the average, because they were intended to go together.

It should also be noted that this is the first Clarus national poll of the entire year. They have decided to jump in at practically the last minute, which is problematic for Clarus, for now they have absolutely no baseline over time to work with, unlike Gallup, Rasmussen, PPP and to some degree, Ipsos/Reuters.

I have problems with the Clarus poll for three reasons:

1.) It is unprofessional to release polling results for two separate occasions at the same time. They could have released the 10/02 poll on 10/03, 10/04 at the latest. They didn't do this because they KNOW that their polling sample for just one poll is way too small to be viable. So, they are combining the two sample groups, over 1,100 LV, to make it viable.

2.) On their summary, they only list the average of both: Obama +2, but they do not present the actual toplines. For them, you need to see the HuffPo data research for both polls, HERE and HERE

3.) Clarus is not releasing internals. Pollsters who have nothing to hide release their internals.

This is no way to do professional polling.


Here is the polling outlook for 2012-10-006: Obama +2.98% (-0.56 shift)

Here was the polling outlook for 2012-10-005Obama +3.54% (-0.20 shift)
Here was the polling outlook for 2012-10-004Obama +3.74% (-0.05 shift)
Here was the polling outlook for 2012-10-003Obama +3.79% (+0.22 shift)
Here was the polling outlook for 2012-10-002Obama +3.57% (-0.56 shift)
Here was the polling outlook for 2012-10-001
Obama +4.11% (-0.52 shift)
Here was the polling outlook for 2012-09-030
Obama +4.63% (-0.10 shift)

Here was the polling outlook for 2012-09-029Obama +4.73% (+0.20 shift)
Here was the polling outlook for 2012-09-028
Obama +4.53% (-0.21 shift)

Here was the polling outlook for 2012-09-027Obama +4.74% (+0.36 shift)
Here was the polling outlook for 2012-09-026Obama +4.38% (+0.19 shift)
Here was the polling outlook for 2012-09-025Obama +4.19% (+0.23 shift)
Here was the polling outlook for 2012-09-024Obama +3.96% (+0.21 shift)
Here was the polling outlook for 2012-09-023Obama +3.75% (+0.06 shift)
Here was the polling outlook
 for 2012-09-022Obama +3.69% (+0.18 shift)
Here was the polling outlook
 for 2012-09-021Obama +3.87% (+0.49 shift)
Here was the polling outlook
 for 2012-09-020Obama +3.38% (-0.22 shift)

Here was the polling outlook for 2012-09-019Obama +3.60% (-0.40 shift)
Here was the polling outlook for
 2012-09-018: Obama +4.00% (+0.07 shift)

Here was the polling outlook for 2012-09-017Obama +3.93% (+0.12 shift)
Here was the polling outlook for 2012-09-016Obama +3.81% (0.00 shift)

Here was the polling outlook for 2012-09-015Obama +3.81% (0.00 shift)
Here was the polling outlook for 2012-09-014Obama +3.81% 

Net shift since 09/14Obama -0.83%


This number (Average shift since 09/14Obama -0.83) shows now appreciable drop in margin for the President following the debate on 10/03. Until that point in time, the numbers were extremely stabile.

October 6th, 2012 marked exactly 31 days until the General Election.



No comments:

Post a Comment

Constructive comments and critique are always welcome. Please keep it polite and respectful.