06 December 2012

Preliminary comparision II: CT, MI and IA compared to the end polling averages

Comparison I, with emphasis on Ohio and Oregon, is HERE.
You can also see the current partisan rankings at that link.

Today: Connecticut, Michigan and Iowa - all of which have certified their results.

In the state of Connecticut, the end polling looked like this:

Nr. CONNECTICUT Date Sample MoE Obama Romney Und Mar.

AVERAGE N/A N/A N/A 53,00 41,50 5,50 11,50


















22 YouGov 04.11.12 895 LV +/-3.6 54 39 7 15
21 PPP (D) – FINAL POLL 02.11.12 1220 LV +/-2.8 55 42 3 13
20 Mason-Dixon 24.10.12 625 LV +/-4.0 49 42 9 7
19 Quinnipiac 24.10.12 1412 LV +/-2.6 55 41 4 14
18 Rasmussen 23.10.12 500 LV +/-4.5 52 45 3 7
17 SUSA 22.10.12 575 LV +/-4.2 53 40 7 13
16 PPP (D) LCV 18.10.12 1015 LV +/-3.1 53 44 3 9
15 U Conn / Courant 18.10.12 574 LV +/-4.0 51 37 12 14
14 Siena 17.10.12 552 LV +/-4.2 53 38 9 15
13 YouGov 16.10.12 434 LV +/-5.5 53 39 8 14
12 Rasmussen 09.10.12 500 LV +/-4.5 51 45 4 6
11 Quinnipiac 04.10.12 1696 LV +/-2.4 54 42 4 12

The polls shaded in yellow were the ones in the final average mix,which was Obama +11.50%.

Actual results?


DEM GOP IND Total DEM % GOP % IND % Margin Mar %
Connecticut 2012 *?? 905083 634892 18139 1558114 58,09% 40,75% 1,16% 270191 17,34%
Connecticut 2008 997773 629428 19592 1646793 60,59% 38,22% 1,19% 368345 22,37%
Diff: -92690 5464 -1453 -88679 -2,50% 2,53% -0,03% -98154 -5,03%

This means that the President outperformed his polling by by almost 6 points, way outside the margin of error.

YouGov, which showed Obama +15, came the closest, but both Mason-Dixon and Rasmussen, both showing Obama +7, were more than 10 points off on margin, which is actually borderline criminal, if you ask me. Even worse was Mason-Dixon, which had Obama at under 50%. Fun times. This means that Mason-Dixon was also off by 9 points on the topline as well. That is an absolutely terrible track record for any pollster. The poll was taken after the 2nd debate.

Oh, and in context: Obama's 2012 margin in Connecticut was not as high as in 2012, but still higher than any other previous margin going back to 1972. This means that even in 2012, Obama did better in Connecticut than Kerry, Gore, Clinton, Bush 41, Reagan and Carter.

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

In Michigan, the end polling looked like this:

Nr. MICHIGAN Date Sample MoE Obama Romney Und. Mar.

AVERAGE (one week): N/A N/A N/A 49,77 44,98 5,24 4,79

AVERAGE (final 3 days):


50,45 45,97 3,58 4,48










Most recent (one week, no repeaters):















56 Mitchell Research 06.11.12 1305 LV +/-2.71 51 46 3 5
55 Angus Reid 05.11.12 502 LV +/-4.4 52 47 1 5
54 YouGov 04.11.12 1091 LV +/-3.3 51 44 5 7
53 Foster-McCollum /Baydoun (R) 04.11.12 1913 LV +/-2.11 46,24 46,86 6,90 -0,62
52 PPP (D) – FINAL POLL 03.11.12 700 LV +/-3.7 52 46 9 6
50 Rasmussen 02.11.12 750 LV +/-4.5 52 47 1 5
49 Grove (D) / PNA / USAction 02.11.12 500 LV +/-4.4 48 41 11 7
48 EPIC / MRA - Detroit Free Press 31.10.12 600 LV +/-4.0 48 42 10 6
47 Glengariff Group 31.10.12 600 LV +/-4.0 47,7 45 7,3 2,7










There was an 8 point disparity in polling in Michigan, betwwen Grove/YouGov and Foster-McCollum-Baydoun.

Actual results?


DEM GOP IND Total DEM % GOP % IND % Margin Mar %
Michigan 2012 * 2564569 2115256 51136 4730961 54,21% 44,71% 1,08% 449313 9,50%
Michigan 2008 2872579 2048639 88976 5010194 57,33% 40,89% 1,78% 823940 16,45%
Diff: -308010 66617 -37840 -279233 -3,13% 3,82% -0,70% -374627 -6,95%

Yepp, the President won the Wolverine State by a near landslide +9.50%. This means that the President outperformed his polling average by about 5 points.

Grove (D) and YouGov came the closest to the actual results.

Rasmussen, which showed Obama +5, missed the margin by 4.5 points, as is Rasmussen's mathematically proven ca. 4 point bias to the Right.

But Foster-McCollum-Baydoun (which was paid for by FOX NEWS) really made an ass out of itself. At Romney +0.62, FMB missed the margin by over 10 points. This is a pathetic, pathetic, pathetic statistic and proves that a firm like FMB is not trustworthy. Not only that, FMB also missed the topline percentage by 8. Just pathetic. No one should ever want to trust this firm again. It put out shit results wherever it polled, obviously towing the party line for FOX. Spit.

And the Michigan margin? Though it was lower than 2008, it was still the fourth largest margin of the last ten presidential cycles.

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

In Iowa, which is now a 5 for 6 Democratic state from 1992-2012, here is how the polling average looked on 11/06:

Nr. IOWA Date Sample MoE Obama Romney Und. Mar.

AVERAGE (one week): N/A N/A N/A 48,10 45,96 5,94 2,14

AVERAGE (FINAL 3 DAYS):


47,83 45,67 6,50 2,17










Most recent (one-week, no repeaters):















39 ARG – Final Poll 05.11.12 600 LV +/-4.0 49 49 2 0
38 PPP (D) – FINAL POLL 04.11.12 1122 LV +/-2.9 50 48 2 2
37 YouGov 04.11.12 1040 LV +/-3.5 48 47 5 1
36 Selzer / Register 03.11.12 800 LV +/-3.5 47 42 11 5
35 Grove Insight (D) / PNA / USAction 03.11.12 500 LV +/-4.4 47 44 9 3
34 Mellman (D) / AUC 03.11.02 600 LV +/-4.0 46 44 10 2
33 Gravis (R) 02.11.12 594 LV +/-4.0 49 45 6 4
32 CallFire / Faith Horizon 01.11.12 891 LV +/-3.4 50 47 3 3
31 Rasmussen 01.11.12 750 LV +/-4.5 48 49 3 -1
30 NBC / WSJ/ Marist 31.10.12 1142 LV +/-2.9 50 44 6 6
29 University of Iowa 31.10.12 320 A +/-5.6 44,4 45,2 10,4 -0,8
27 WAA (R) 31.10.12 1174 LV +/-3.0 48,8 47,3 3,9 1,5









Between NBC/MARIST and Rasmussen, we saw a polling variance of 7 points, way outside the margin of error.

Actual results?


DEM GOP IND Total DEM % GOP % IND % Margin Mar %
Iowa 2012 * 822544 730617 29019 1582180 51,99% 46,18% 1,83% 91927 5,81%
Iowa 2008 828940 682379 25804 1537123 53,93% 44,39% 1,68% 146561 9,53%
Diff: -6396 48238 3215 45057 -1,94% 1,78% 0,16% -54634 -3,72%

So, this means that with +5.81% as the final result, NBC/MARIST nailed it. Selzer, which showed +5, and is usually considered the Gold Standard for Iowa, was also close. PPP missed it by 4 - to the right. So, the next time you hear people complain that PPP always swings to the left, remember Iowa and these results.

It also means that Obama outperformed his end polling average in IOWA by about 3.6 points, just outside the overall margin of error.

So, Rasmussen and the U of Iowa, which showed Romney +1 and +0.8, respectively, did the worst. This means that Rasmussen was off by 7 points in Iowa! When you read how the U of Iowa poll was conducted (by University students over a much too long time frame), you can forgive them for crappy results. But Rasmussen, which can do overnight robo-polling? No way.

In historical context, Obama's 2012 margin is of course less than his 2008 margin, but it is the 6th highest of the last ten cycles. It was a solid win, but not a landslide. +6 is really not that close a race, folks...

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

Anyone notice a common denominator in all three of these states?

You guessed it: Rasmussen was either the worst or among the worst pollsters in all three cases.
Rasmussen missed the margin in CT by 10, in MI by 4.5 and in IA by 7 points, respectively. It also missed Ohio by 3.

That is a piss-poor track record. And you will see the name Rasmussen more and more often as I compare groups of states to their actual results, and almost always, Rasmussen can be found at the bottom of the barrel when it comes to accuracy. Were I a Rasmussen customer, I would demand my money back!

I am now very glad that I kept track of all of these polling numbers, which are now public domain. For those crappy pollsters who really did do shoddy work cannot run and hide under a rock. The disinfectanct known as FACTS will not let them get away with it.

No comments:

Post a Comment

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