So, it's now exactly four weeks before the 2014 mid-terms, and time to compare to last week's polling aggregate snapshots to today.
Here were the aggregates from last Thursday (October 2, 2014):
2014 battle for control of the US Senate Page 8 US Message Board - Political Discussion Forum
And now, here are the aggregates as of today, October 7, 2014:
DEMOCRATS LEADING:
Aggregate, MI:
August 11, 2014: Peters +4.0
September 16, 2014: Peters +5.2
September 22, 2014: Peters +5.4
October 2, 2014:
Peters +6.0
October 7, 2014:
Peters +6.7
The needle has moved +0.7 in Peters' direction over the last week. Of the competitive Senate races, this one is probably the safest for the Democrats at this time. Peters is heading into the safe zone.
Aggregate, NH:
August 11, 2014: Shaheen +10.4
September 16, 2014: Shaheen +3.5
September 22, 2014: Shaheen +5.0
October 2, 2014: Shaheen +4.0
October 7, 2014: Shaheen +4.6
Last Thursday, I wrote the following:
As of today, the needle has moved 1 point in Scott Brown's (R) direction, but the disparity in polling is just nothing less than amazing. To have a tie poll and a Shaheen +10 poll conducted within pretty much the same time frame are too things that cannot exist in the same universe at the same time. And ARG has a conservative mathematical bias, not a liberal one. Conversely, New England College has a slight liberal mathematical bias, not a conservative bias. As has been discussed by me VERY OFTEN, Rasmussen has a verifiable mathematical bias of 4 points to the right almost all of the time, so a Ras +6 for Shaheen (though the poll is now 3 weeks old) jives with an ARG +10 for Shaheen.
If we eliminate polls older than two weeks old from this aggregate would be Shaheen +5.6. That is the more accurate statistic right now. Right now, she is doing better than the RCP aggregate indicates, because of "cold coffee" polls that should have been cycled out by now.
Aggregate, NC:
August 11, 2014:
Tillis +1.3
September 16, 2014:
Hagan +3.7
September 22,2014:
Hagan +5.0
October 2, 2014:
Hagan +4.2
October 7, 2014:
Hagan +3.7
Technically, this has moved 0.5 points away from Kay Hagan and toward Thom Tillis, only, the most right leaning of all the pollsters, Civitas (R),is showing Hagan with +7 and also hitting the 50% mark. If you reduce the time frame down to the last two weeks (14 days), then the bottom three polls would drop out and the average of the top four polls would be
Hagan +3.75, so we are seeing a lean, but stable lead for incumbent Hagan (D) in the Tarheel state. That being said, in a wave, a +4 aggregate can be overcome. Kay Hagan is in no way in safe waters.
REPUBLICANS leading:
Aggregate, CO:
August 11, 2014: Udall +3.7
September 16, 2014: Udall +1.5
September 22, 2014: Udall +0.6
October 2, 2014:
Gardner +1.5
October 7, 2014:
Gardner +0.6
Since last Thursday, the aggregate needle has moved 0.9 toward incumbent Udall (D). This is, on paper, a true dogfight, but please remember that one week ago, I pointed to the RCP end-average from 2010 for Colorado. Here it is again:
And one week ago, I wrote the following:
This is enlightening. In 2010, RCP had as it's aggregate Buck +3. Bennett won on election night by +1. This means that the RCP aggregate was off by 4 points to the right. Even PPP (D) was off to the Right. Afterwards, most analysts wrote that the Latino vote in Colorado, Nevada and New Mexico was grossly underestimated and calculated.
For this very reason, regardless whether the needle shows Udall +1.5 or Gardner +1.5, I would say that if there is a state where the aggregate could be way off, it would be a state like Colorado.
In other words, it's still very much a dogfight in Colorado. No one has put this race away.
And that testimony from one week ago still stands. We now have historical evidence from 2008, 2010 and 2012 that polling in the western states where there are sizeable Latino communities has been unreliable. And the polling statisticians from both sides, I am sure, are keenly aware of this historical fact.
Just to underscore this, here was the RCP average for the Reid (D-inc) / Angle (R) senatorial from 2010:
No, your eyes are not fooling you. The end aggregate for NV (which has a sizeable Latino population, similar to CO and NM, also CA) showed Angle +2.7, but Reid won by +5.6, which means that the aggregate was off by 8.3 points. That is WAY off. And notice that even PPP (D) showed an Angle win. All polling firms in 2010 miscalculated the Latino impact on the overall vote in the SW of the USA.
Just remember these historical facts when looking at the aggregates for CO and NM this time around.
Aggregate, IA:
August 11, 2014: Ernst +0.8
September 16, 2014: Braley +1.4
September 22: Braley +0.1
October 2, 2014: Ernst +2.8
October 2, 2014: Ernst +1.9
Since October 2, 2014, the needle has moved 0.9 toward Braley.
Just to show how volatile this race is, one week ago, I wrote the following:
The needle has moved 2.9 points in Jodi Ernst's (R) direction and the most damning piece of evidence is the Des Moines Register poll, which is one of the two real GOLD STANDARD polls in Iowa. Ernst is taking the lead in this state, but as you can see from the four aggregate values above, this pendulum could swing again.
If these numbers hold, then it appears that CO and IA may be the two cliffhanger races on election night, 2014. This is every bit as much a dogfight at Colorado at this moment.
Aggregate, AK:
August 11, 2014: -no aggregate was possible-
September 16, 2014: Sullivan +1.3
September 22, 2014: Sullivan +1.3
October 2, 2014:
Sullivan +4.7
October 7, 2014: Sullivan +4.7
Despite new polls, Sullivan is maintaining a +4.7, because the CBS poll from August that showed Sullivan +6 has been replaced by the newer CBS poll showing exactly the same margin.
Now, Democrats can take the tack that polling in Alaska is notoriously unreliable, but a +5 is pretty hard to get around. Sullivan is definitely in the lead.
Aggregate, GA:
August 11, 2014: Perdue +3.2
September 16, 2014: Perdue +3.0
September 22, 2014: Perdue +3.3
October 2, 2014: Perdue +3.4
October 7, 2014:
Perdue +3.1
Just as Kay Hagan has demonstrated a lean but resilient lead in NC, in Georgia, Republican Perdue is demonstrating a lean but resilient circa +3 to +3.5 lead. The needle has moved only +0.3 toward Nunn (D), which could just as well be nothing more than so-called "statistical noise". All of these polls are essentially within the two week time frame. Now, last week, I leveled some well-founded criticism of InsiderAdvantage, a firm notorious for inflated R-margins and also a firm that hides it's data behind a paywall and even then, does not release critical internals. That being said, two other firms are also showing the same margin for Perdue. I would really like to see a PPP (D) result from this state for comparison and will also remind that for most of the South, SUSA has somewhat become the gold standard. SUSA only shows +1 for Perdue, but that poll will soon fall out of the two-week window. However, no matter how you slice it, Perdue is still in the lead. But as is the case with LA, there is a third party candidate on the ballot (which most pollsters are ignoring, at their own peril) and like LA, GA has a 50% hurdle to overcome, otherwise, there is a runoff election in December.
Speaking of Louisiana:
Double aggregate, LA:
Jungle Primary, October 2, 2014:
Landrieu +1.2
Jungle Primary, October 7, 2014:
Landrieu +2.7
Two-"man" race, possible runoff, October 2, 2014:
Cassidy +4.6
Two-"man" race, possible runoff, October 2, 2014:
Cassidy +5.6
We are now getting some solid data-points with which to work. In the Jungle primary, the needle has move 1.5 points to Landrieu. But, and this is the kicker: Rob Maness (R running as a third party candidate) has an aggregate of 8 points. And in the two-way polling, the needle has moved 1 point toward Cassidy. Crazy, what? Now, watch the math closely:
100 - 8 = 92.
92 / 2 = 46
46% (without the Maness 8%) is temporarily the new "50 yard line", if you will.
a Landrieu aggregate +2.7 means that the margin is spead on the top and the bottom of 46%, by 1.35. (2.7 / 2 = 1.35)
46 + 1.35 extrapolates Landrieu at 47.35%.
46 - 1.35 extrapolates Cassidy at 44.65%
Manness 8.00%
Total: 100.00%
Margin: Landrieu +2.7
That is ASSUMING that the 8% for Maness is reality, for 3rd party candidates tend to do worse on election night than aggregate polling often shows.
But with Manness at 8%, if these figures hold, then there is no conceivable way for Landrieu to get to 50%, which means there will be a runoff, and if the runoff "two way" numbers hold, then all the Jungle Primary does is to stave-off the inevitable: a Cassidy +5 to +6 win. In other words, what hurts Cassidy in the Jungle Primary may help him come December.
Last week, I wrote:
And indeed, we now have some fresh polling DNA, which helps. In the Jungle Primary, Landrieu is definitely ahead: Landrieu +1.2, and that stat is being held down by one CBS poll that is now 4 weeks old and was already flawed in that is was conducted over a 2-week time span, which is ridiculous. Without that poll, the aggregate would be: Landrieu +2.3.
That's the good news for Landrieu. The bad news is that in direct two-man polling, Cassidy is clearly ahead with an aggregate of +4.6 (it is likely lower, since the FOX +13 is quite obviously an outlier) and since it is highly unlikely that 3rd party candidate Maness (who is polling about 9-10%) will not win the jungle primary, it is highly likely that this thing goes into overtime and there will be a runoff election. Especially if the fate of the Senate would be decided in LA, this race could become especially interesting.
Aggregate, AR:
September 16, 2014: Cotton +2.5
September 22, 2014: Cotton +2.5
October 2, 2014: Cotton +3.6
October 7, 2014:
Cotton +3.7
As is the case with Hagan (D) in NC and Perdue (R) in GA, Tom Cotton (R) is maintaining a lean but resilient lead in AR against Democratic incumbent Pryor. Two cold-coffee polls have now been cycled out and one new poll has been cycled in, and the aggregate result is practically identical to last week.
Two important data-points here: Rasmussen has a proven mathematical bias of around +4 to the Right and right now, the +7 for Cotton from RAS is 3.3 points to the Right of the aggregate. None of this surprises me. What DOES surprise me is the quirky result from Suffolk, which, in 2012 ended up with alot of egg on it's face by going on TV on FOX one full month before the Presidential election and declaring that it would do no more polling in battlegrounds VA, NC and FL, that it had already "painted those states red" for the GE. Of course, when Obama won both VA and FL, this made Suffolk look pretty silly. Here is a link to prove what I just wrote:
Suffolk Poll Romney To Win Virginia Florida N.C. - Business Insider
So, with a proven Conservative bias in the states that mattered in 2012, it is indeed strange to see a +2 for Pryor in a state that Romney swept by +24 just two years ago.
Another damning piece of evidence is the PPP (D) poll, which also shows a Cotton +6. Conservatives love to say that PPP is biased to the left, but actually, in 2010, it has a very slight aggregate bias to the RIGHT.
No matter how you slice it, Tom Cotton is currently leading in Arkansas.
Aggregate, KY:
August 11, 2014: McConnell +2.5
September 16, 2014: McConnell +5.2
September 22, 2014: McConnell +5.2
October 2, 2014: McConnell +5.3
October 7, 2014: McConnell +4.2
Technically, after weeks of unchanged numbers in the Bluegrass state, the needle has moved 1.1 points toward Democrat Alison Grimes, but on another thread, I did a quick analysis of the "Bluegrass poll"s track record over the last 6 years and discovered a strong mathematical bias to the LEFT, a very strong one at that. You can read the findings here:
Mcconnell Falls Behind In Kentucky US Message Board - Political Discussion Forum
Incumbent McConnell is still ahead here. He already prevailed in a close election in 2008, and like Harry Reid (D-NV), he knows how to fight a close race. Remember, I am writing this praise as an avowed Democrat.
INDEPENDENT leading:
Aggregate, KS:
September 22, 2014: Orman +1.2
October 2, 2014: Orman +5.3
October 2, 2014:
Orman +5.3
The needle has moved 4.0 points toward Orman since September 22, 2014.
The most damning piece of evidence that Roberts (R) is probably going to lose is the big +10 for Orman from Marist, and I will note again that is statistically impossible for a +10 for one candidate and a mathematical tie for that same candidate to appear at the same time in two polls. One of those two polls is definitely off.
Either way, Orman is ahead, demonstrably.
BTW, I made a thread about a very good, concise assessment of Mr. Orman, here:
Good Concise Assessment Of Greg Orman i-ks Sen US Message Board - Political Discussion Forum
I suspect that in the next two weeks, Mr. Orman will hit the 50 mark in some polls.
----------------------------------------------------------------
FACIT: movement in both directions since last week, but the likely GOP pickups have not changed and the two states that are real dog-fights have also not changed. The GOP is still on target to pick up at least 6 seats net and therefore get to 51 (majority) in the US Senate.
That being said, in the generic aggregate, the GOP lead has actually shrunk since last week:
Generic aggregate, October 2, 2014: GOP +2.9
Generic aggregate, October 2, 2014: GOP +2.1
Still no statistical signs of a wave, but plenty of evidence that the GOP is winning where it needs to, which is exactly in line with 160 years of US mid-term election history:
Congressional Elections compared to Presidential Terms 1855-present US Message Board - Political Discussion Forum
