Disaggregating the polls

All the poll aggregators are telling the same story. Republicans have a very small edge in enough Senate seats to win a majority. However, this edge is so narrow that a Democratic ground-game advantage  may be enough to overcome the small advantage and allow Democrats to maintain  their majority with a 50-50 split (depending on which party Greg Orman caucuses with).

Over the next couple of days we will take a look at the 2014 Senate polls a fews different ways with the goal of breaking free of the aggregation morass (but no deskewing!).

First up  is a look at how the individual pollsters see the election.

Survey USA

CO – Gardner 45, Udall 43
GA – Perdue 48, Nunn 45
KS – Orman 47, Roberts 42
KY – McConnell 44, Lundergan Grimes 43
NC – Hagan 44, Tillis 44
NH – Shaheen 48, Brown 46
SD – Rounds 35, Pressler 32, Weiland 28

Rasmussen (GOP +5/7)

AK – Sullivan 48, Begich 45
AR – Cotton 47, Pryor 44
CO – Gardner 51, Udall 45
GA – Perdue 48, Nunn 48
IA – Ernst 48, Braley 45
KS – Orman 49, Roberts 44
KY – McConnell 52, Grimes 44
LA – Cassidy 52, Landrieu 43
MT – Daines 55, Curtis 35
NC – Hagan 48, Tillis 46
NH – Shaheen 48, Brown 42
SD – Rounds 44, Weilend 29, Pressler 18
WV – Capito 50, Tennant 39

CBS/NYT/YouGov (GOP +6/7)

AK – Sullivan 48, Begich 44
AR – Cotton 47, Pryor 42
CO – Udall 47, Gardner 46
GA – Perdue 47, Nunn 44
IA – Braley 44, Ernst 44
KS – Roberts 42, Orman 38
LA – Cassidy 46, Landrieu 42
NC – Hagan 44, Tillis 41
NH – Shaheen 46, Brown 41

CNN/ORC (GOP +5/7)

AK – Sullivan 50, Begich 44
AR – Cotton 49, Pryor 47
CO – Gardner 50, Udall 46
GA – Nunn 47, Perdue 44
IA – Braley 49, Ernst 48
KS – Roberts 49, Orman 48
LA – Cassidy 50, Landrieu 47
NC – Hagan 46, Tillis 43
NH – Shaheen 49, Brown 47

PPP(D) (GOP +5/7)

AK – Sullivan 44, Begich 42
AR – Cotton 44, Pryor 39
CO – Gardner 47, Udall 44
GA – Perdue 47, Nunn 47
IA – Braley 48, Ernst 47
KS – Orman 45, Roberts 42
LA – Cassidy 48, Landrieu 45
NC – Hagan 46, Tillis 43
NH – Shaheen 49, Brown 45

Fox News – Projected GOP win

AK – Sullivan 44, Begich 40
AR – Cotton 46, Pryor 39
KS – Roberts 44, Orman 39
LA – Cassidy 51, Landrieu 38

NBC/Marist -Projected GOP win

AR – Cotton 45, Pryor 43
CO – Gardner 46, Udall 45
IA – Ernst 49, Braley 46
KS – Orman 45, Roberts 44
NC – Hagan 44, Tillis 44

USA Today Suffolk – Projected GOP win

AR – Pryor 45, Cotton 43
CO – Gardner 46, Udall 39
IA – Ernst 47, Braley 43
KS – Orman 46, Roberts 41
LA – Cassidy 48, Landrieu 41
NC – Hagan 47, Tillis 45
NH – Shaheen 49, Brown 46

Gravis – Project GOP win

AR – Cotton 47, Pryor 43
IA – Ernst 49, Braley 43
KS – Orman 47, Roberts 45
LA – Landrieu 45, Cassidy 45
NC – Tillis 48, Hagan 43


CO – Gardner 47, Udall 43
IA – Ernst 49, Braley 45


CO – Gardner 47, Udall 46
IA – Ernst 47, Braley 46
KS – Orman 46, Roberts 46
NC – Hagan 48, Tillis 46

Survey USA points to Democratic upset in Senate

While poll aggregators are all the rage – and are the most accurate tool for predicting elections – it is worth looking at some of the individuals pollsters (especially some of the more reputable ones) to see what they predict. Lost in the all of the polls is the prediction by Survey USA that the Democratic Senate majority might actually survive Election Night 2014.

According to Survey USA, Republicans trail in their currently held seats in Kansas and Georgia. They call Kentucky a dead heat with Mitch McConnell leading by  a single point after trailing his Democratic opponent Alison Lundergan Grimes by two in their previous survey. South Dakota is also very much in play, according to Survey USA as Democratic candidate Rick Weiland only trails Republican front-runner Mike Rounds by three.

The one race they see breaking for the GOP is Colorado where Cory Gardner clings to a two-point lead over Democratic incumbent Mark Udall. Democratic incumbents also cling to narrow leads in New Hampshire and North Carolina.

It is not possible to predict the complete Senate picture from Survey USA’s polling because they have not conducted surveys in Alaska, Arkansas, Iowa, Louisiana, Montana, and West Virginia, which are all states where Republicans are hoping to pick up seats. That being said, it is reasonable to suggest that if Republicans are losing in Georgia and Kansas, while fighting to stay alive in Kentucky and South Dakota, Democrats would be favored to keep the Senate.



Republicans projected to win Senate

Senate 2014 10-15

Projection –  Republicans 51, Democrats 49

Current Senate make-upDemocrat 55 (including 2 independents)Republican 45.

Seats not up for re-election – Democrat 34, Republican 30.

Safe Democrat (6)-  Delaware, Hawaii*, Illinois*, Massachusetts*,  New Jersey*, Rhode Island*

Lean Democrat (4)-  Minnesota*,  Oregon*, New Mexico*, Virginia*,

Slight Lean Democrat (3)- Michigan, New Hampshire*, North Carolina*

Toss-up (3)Colorado*, Iowa, Kansas*

Slight Lean Republican (5) – Alaska*, Arkansas*, Louisiana*, Georgia, Kentucky*, 

Lean Republican (3) - Montana, South Dakota, West Virginia

Safe Republican (12)- Alabama*, Idaho*, Maine, Mississippi*, Nebraska, Oklahoma (2), South Carolina (2)*, Tennessee*, Texas*, Wyoming*

*Denotes incumbent is running

Democratic trends:

Georgia – Michelle Nunn has gained on David Perdue in just about every poll taken over the past month. If the Election were held today Perdue would have a hard time reaching 50% and avoiding a run-off. The problem for Nunn is that even if she beats Perdue on Election Day she will still likely fall short of 50%

South Dakota – Two polls show Republican favorite Mike Rounds leading by only three points. The only problem for Democrats is that each of the polls showed a different candidate running second. Mike Rounds is still a solid favorite.

Republican trends:

Alaska – Dan Sullivan is now leading Mark Begich by nearly five points in the Real Clear Politics average. Given Alaska’s history of under-polling Republicans, this race is looking very good for the GOP.

Colorado – Cory Gardner has solidified his position as the polling favorite with two polls showing him leading by six points.
New Hampshire – Scott Brown posted his first lead, but still faces an uphill battle.

New Hampshire – Scott Brown took a lead in one poll, but still faces an uphill battle.

North Carolina – Thom Tillis finally halted his slide and has pulled to within two points.

Races most likely to change parties:

1. Montana (95%) Polls, candidate, environment, and terrain favor GOP. 
2. West Virginia (90%) Polls, candidate, environment, and terrain favor GOP.
3. South Dakota (80%) Polls, environment, and terrain favor GOP.
4. Alaska (70%) Terrain and environment favor GOP. Solid Democratic incumbent. 
5. Louisiana (60%) Terrain and environment favor GOP. Solid Democratic incumbent. Looks like a December run-off.
6. Arkansas (55%) Terrain and environment favor GOP. Strong Democratic incumbent.
7. Iowa (55%) Republicans got their candidate. Terrain slightly favors Democrats, while environment slightly favors Republicans.
8. Colorado (55%) Terrain slightly favors Democrats. Environment favors GOP. GOP may have better candidate.
9. Kansas (45%) Terrain favors GOP. State environment favors the Independent. Candidates both have vulnerabilities
10. North Carolina (45%) Terrain and environment slightly favor GOP. Mediocre Democratic incumbent, fractured GOP. Republican candidate has very high unfavorables.
11. New Hampshire (40%) Terrain and candidates favors Democrats.
12. Georgia (40%) Terrain and environment favor GOP. Solid Democratic candidate performing well in polls, but unlikely to win run-off.
13. Kentucky (35%) Terrain and environment favor GOP. 
14. Michigan (20%) Terrain favors Democrats. Environment slightly favors Republicans. Democrat is starting to build a small but consistent lead.
15. Minnesota (5%) Terrain and candidates favors Democrats.         
16. Virginia (5%) Terrain and candidates favors Democrats.

Michelle Nunn is closing the gap in Georgia

Michelle_Nunn_2012Democrats believe Greg Orman in Kansas is their best chance to take a seat from Republicans, but Michelle Nunn is slowly gaining ground on David Perdue in Georgia. Nunn saw her chances rise during the spring of 2014 before falling after the GOP united behind Perdue. With the election less than a month away, Nunn is on the rise again.

According to Real Clear Politics, Perdue still leads by 3.2 points, but Nunn has gained in each of the four polls used in the average. Three of the four only have minor changes, but combined the four represent a noteworthy trend.

CBS/NYT/YouGov – Nunn +2
Rasmussen – Nunn +1
Insider Advantage – Nunn +6
Survey USA – Nunn +2

The four polls do not include a survey by Landmark, which has been the most favorable to Nunn, so the 3.2% average lead for Perdue may be overstated. Also keep in mind that Georgia has a 50% threshold, so at the very least, there is a fair likelihood of the race going to a run-off.

Given both candidate’s need to reach 50% and the presence of a Libertarian candidate, the recent surge by Nunn is a real problem for Republicans. The matter is further complicated by the January date of the run-off, which means neither candidate may be seated by the start of the next Congress. While Republicans are advancing on other fronts, Michelle Nunn presents a real chance for Democrats to at least put off the final decision as to who will control the Senate until January.

GOP may be cruising to victory in Alaska

DanielSSullivanRepublican Dan Sullivan has passed Democratic Incumbent Mark Begich in the polls. The latest PPP(D) poll moved seven points in Sullivan’s direction while the most recent Rasmussen poll added three points to his lead. Overall, Sullivan leads by nearly five points in the Real Clear Politics average.

Polling in Alaska has historically been abysmal. However, although Alaskan polling has been innacurate, it has been reliable in one way – it has significantly underestimated Republican support.

In 2006 when Sarah Palin was first elected to office, the final pre-election Rasmussen poll showed her with a mere one point lead in the week before the election. Palin went on to win by seven points.

During the three-way race in 2010, the October polls showed Republican nominee Joe Miller and Republican write-in candidate Lisa Murkowski combining for just under 70 percent of the vote while Democratic candidate Scott McAdams polled at 26 percent. Murkowski and Miller ending up getting 75% of the vote while McAdams only won 23%. The Republicans over-performed the polls by about six points, while the Democrat underperformed by nearly three points for a net nine point difference.

The biggest polling debacle in Alaska came in 2008 when the RCP average predicted Begich would defeat longtime incumbent Ted Stevens by double digits. Begich went on to win by only two points. Polling that same year predicted Republican United States Representative Don Young would go down to defeat by five points while Young ended up actually winning re-election by five. Finally, the RCP average also underestimated McCain’s margin of victory by seven points.

There was no polling in Alaska in 2012 due to Mitt Romney’s anticipated easy victory and the general difficulty polling Alaska’s remote areas. Nevertheless, the trend since 2006 is clear. Pollsters, across the spectrum significantly underestimate Republican strength.

The theories for the problematic polling have revolved around the transient population and the lack of landline phones – or in some instances the lack of any phone – which has traditionally served as the primary method of polling. The lack of quality independent pollsters may also contribute to the problem. What has yet to be explained is why these polls consistently underestimate Republican candidates.

Republicans have already established large leads in South Dakota, Montana, and West Virginia, and consistent leads in Arkansas and Louisiana. Alaska would be seat number fifty-one if the GOP can avoid losing one of their own in Kansas. Unless polling has found a way to correct itself, Sullivan may very well have a lead close to double digits and be on his way to an easy win.

Thom Tillis’ Trouble

Thom Tillis has come up on the short end in the last seven polls against Democratic incumbent Kay Hagan. Particularly,  troubling for Tillis is the absolute disdain voters have for him. According to PPP(D), just 34% of voters have a favorable opinion of Tillis while 48% have an unfavorable opinion. For a challenger those numbers will be fatal if they hold through Election Day.

Of particular concern for Tillis are the findings in the most recent High Point University poll. Tillis still trails in this poll despite a very friendly electorate. According to High Point, Obama’s approval is 38%, Kay Hagan’s approval is a dismal 39%, and Democrats have a net (-4) favorability rating while Republicans have a net (+6) favorability rating. Fifty percent of votes identify as conservative while less than 20% identify as liberal.

The bottom line for Tillis is that despite getting a very friendly sample  – likelier a friendlier one than he will get on Election Day – he still trails. The silver lining is that the voters he needs to win over should be more malleable as they tend to be more conservative and don’t like Hagan either.

Battle of the Pollsters

Voter_pollIf CBS News and the New York Times are to be believed, Republicans are cruising towards an Election Night victory. The latest CBS/NYT poll shows Republicans leading by seven points in the generic ballot. Considering that the GOP lost the total Congressional vote by a point in 2012 and still retained their majority by a reasonable margin, a seven-point lead indicates a measurable bump in seats for Republicans.

When CBS/NYT polled the Senate races they found Republicans leading by at least four points in five seats currently held by Democrats. Altogether, the GOP led in eight races with Democratic controlled seats. Should their numbers hold up on Election Day, Republicans would have 53 Senate seats in the next Congress.

The CBS/NYT polls are not alone in their predictions of a GOP wave. The generic ballot poll conducted for Fox News also shows Republicans leading by seven points. Quinnipiac found Republican Cory Gardner leading Democratic incumbent Mark Udall in Colorado by eight points and Republican Joni Ernst leading her Democratic opponent Bruce Braley by six points in Iowa. If Gardner and Ernst manage to win by half those margins it will be a very good evening for Republicans.

Despite these polls, Democrats believe they have a chance to keep their Senate majority. A Republican pollster shows Kay Hagan cruising to re-election with a 10-point lead in North Carolina. Two polls in Georgia show Democrat Michelle Nunn leading Republican David Perdue. A Nunn win would virtually end any chance the GOP has of retiring Harry Reid as majority leader.

Rasmussen Reports, once every Republican’s favorite pollster, has been particularly bullish on Democratic chances this election season. Rasmussen gives Democrats good news from the top down. President Obama is staying over 45% and occasionally approaches 50% job approval among likely voters while Democrats have a three-point lead in the generic ballot. According to Rasmussen, Republicans trail by one point in Arkansas (he is the only pollster showing Democrats with a lead) and trail by six in North Carolina. The GOP’s chances of winning a majority are a coin toss at best according to his polls.

The polling situation is similar to 2012 with the roles reversed. According to the 2012 polls, Mitt Romney’s best-case scenario was a split decision while Obama had the potential for a comfortable victory. Ultimately, the pollsters who favored Obama were correct because they accurately predicted the very favorable Democratic electorate.

This year Republicans have the potential to win as many as 10 seats if everything goes their way while Democrats are hoping to hold the GOP gains to four or five seats as a best case scenario. Should the electorate split the difference the GOP will win a Senate majority.

Elections are ultimately decided by the make-up of the electorate on Election Day. It is the job of the pollsters to tell us ahead of time what the electorate will look like. This year it looks like there will be clear winners and losers in the polling industry no matter what the final results are.


Helping You Predict Political Winners and Losers