Friday, May 28, 2010

More evidence Dems aren't losing the center

The main reason Scott Brown was able to pull off a shocking upset in the Massachusetts Senate race earlier this year was that he beat Martha Coakley 55-41 with moderate voters. Just how rare of a feat is that for a Republican Senate candidate? There isn't a single other one in a competitive Senate race we've polled this year leading with moderates- in fact with the exception of Florida the Democratic candidate is up by at least 8 points with centrist voters in every race.

It's just another data point showing that Democratic troubles this year are not the result of them losing the center, but of conservative voters being more motivated to turn out. In places like Kentucky and Missouri, where the GOP will be slightly favored to win this fall, Robin Carnahan and Jack Conway lead their respective Republican opponents by 40 points with moderates. Democratic candidates also have 20+ point leads with moderates in places like North Carolina, Colorado, and Ohio.

I think we may have all (myself included) made too much of the impending doom the Massachusetts results portended for Democrats this year. The combination of an exceptionally strong Republican candidate in Brown and an exceptionally weak Democratic candidate in Coakley created a formula that made it possible for the GOP to win moderate voters. But the Republicans haven't shown the ability to replicate that formula in any other key Senate contest yet, and as a result Democrats are decidedly winning the center.

Here's out Senate data on moderates:

-Jack Conway leads Rand Paul 60-20
-Robin Carnahan leads Roy Blunt 61-21
-Elaine Marshall leads Richard Burr 59-24, Cal Cunningham leads Burr 54-26
-Michael Bennet leads Jane Norton 54-31
-Lee Fisher leads Rob Portman 47-27
-Joe Sestak leads Pat Toomey 45-29
-Alexi Giannoulias leads Mark Kirk 36-23
-Harry Reid leads Sue Lowden 51-41
-Blanche Lincoln leads John Boozman 49-40, Bill Halter leads Boozman 45-36
-Paul Hodes leads Kelly Ayotte 47-39
-Charlie Crist has 34% to 32% for Kendrick Meek and 19% for Marco Rubio


Chris said...

Forget moderates, what is the spread with unaffiliated voters?

sunny.texas said...

Be careful, I can see November from my house.

Anonymous said...

So, how come Republicans are in the lead in all these races? Obvious answer is that for most self-described "moderates", it means liberal-lite and they always vote for Democrats, some of these GOPers have huge leads even though they lose moderates by big numbers, answer is that these moderates aren't center, they're leftist Democrats. Nice nickname though.

Unknown said...

No matter how moderate their views are most Republicans consider themselves conservatives. Democrats, on the other hand, think liberal is a bad word, and feel much more comfortable thinking of themselves as being in the middle.

Look at all your polls. Conservatives significantly often outnumber liberals in each one. In South Carolina it's 4 to 1. Does anyone believe that conservatives outnumber moderates in California 34% to 23%?

Republicans need to do well with moderates to win, but they don't need to do well with self-described moderates. Independents, regardless of ideology, are the key.

OSUPhantom said...

It's clear that the moderates are not the ones to win in these elections. It's independent who are going for Republicans by big numbers right now.

Dustin Ingalls said...

"So, how come Republicans are in the lead in all these races?"

They're not. In the latest polling, Fisher is up on Portman, Bennet on Norton, and Sestak on Toomey, albeit by very narrow margins. The only Republicans up by double digits are Boozman and Ayotte.

You guys are right that self-identified conservatives (and Republicans) have grown this cycle relative to Democrats and liberals. The self-identified numbers always fluctuate with the public mood; in 2008, liberals were more prevalent than they are now (though conservatives still outnumbered them thanks to Republicans making "liberal" a dirty word), but Democrats had the self-ID'd advantage in some places where they don't now. Moderates this year are probably more left of center than usual, and independents more right of center than usual. As is often the case, most self-identified "independents" aren't really independent; most Americans tend to vote one way or another, and this year these independents seem to just be angry at everyone, including Republicans, but are tending to lean toward Republicans. You can find data on independents by going back and looking through all our polling on these races. In general, Republicans are leading with them, but in some races, Democrats are.

Christian Liberty said...

When 40% of Americans are conservative and only 20% are leftists, you are redefining the "center" as from the 20 yard line (far left) to the opposite 40 yard line (center-right). That is NOT the center. Your "center" skews left from sampling bias.

Christian Liberty said...

Democrats are clearly losing the nation. Democrats' share of the generic congressional ballot is near 50-year historic lows.

Dustin Ingalls said...

"Your "center" skews left from sampling bias."

There's no sampling bias. People are what they say they are, and we don't weight for party or ideology.

Christian Liberty said...

Dustin, are you falling for your own spin? Or are you trying to pull a statistical trick? As you can tell from the comments above, I'm not the only one that's NOT BUYING IT.

The trick goes like this:
Conservatives are twice as numerous as liberals (confirmed by gallup over multiple years). Conservatives are roughly 40%, moderates 40%, "liberal" 20%. This means that the group that self-identifies as "moderates" DOES SKEW LEFT.

Do I need to draw you a picture? Suppose you have a number line from 0 (far left) to 100 (far right). By selecting self-identified "moderates", you are selecting the portion of the population that lines up between 20 and 60, and then you claim that it is significant that they don't average out to 50 but rather SKEW LEFT.

Are you buying your own spin on your statistical tricks, or are you doing this deliberately to deceive people into believing that polls are coming out more pro-Democrat?

It would be far more honest to use Independents (which are more factually placed in the center) rather than to use moderates (which SKEW LEFT) and try to pass off a left-leaning sample as "the center".

As Democrats and Republicans have similar proportions nationally, independents are a far more center-based group than left-leaning "moderates".

Christian Liberty said...

You can claim that self-identified moderates lean Democrat, but you cannot claim that self-identified moderates are in "the center" of the population ideologically. When 40% of Americans are conservatives and only 20% are liberal, the remaining moderates do NOT straddle the center but skew to the left. With all the evidence that conservatives outnumber liberals, passing off moderates as "the center" is dishonest; such a claim is contradicted by multiple years' worth of solid evidence:

Anonymous said...

Dustin Ingalls said: "People are what they say they are, and we don't weight for party or ideology."

If in fact what you are saying is true and, as such, the word "moderate" is, at best subject to individual interpretation and application, why then would it be used AT ALL
in your polls, analysis or commentary. It would seem for all practical purposes to be pointless in doing so. The very fact that it is being used only lends credence to those questioning the reasons for doing so in the first place. I would think that if PPP wants to be or remain credible as a polling service, it would do all that it can to avoid perceived bias in it's polling methodology or through conveyance of elicited data, relative to those polls.

Christian Liberty said...

Democrats ARE indeed losing the center

Independents, not moderates, are more representative of the median (center) position of ideological thought in America.

Most recent public PPP data,
among independents:

Paul leads Conway 44-29

Blunt leads Carnahan 47-35

Burr leads Cunningham 41-32
Burr leads Marshall 40-38

Bennet leads Norton 42-40
Buck leads Bennet 40-39
Bennet leads Wiens 39-35
Norton leads Romanoff 40-35
Buck leads Romanoff 36-34
Romanoff leads Wiens 35-33

Portman leads Fisher 37-25
Portman also led Brunner 37-25

Sestak leads Toomey 35-34
Toomey led Specter 41-38

Kirk leads Giannoulias 33-30

Lowden leads Reid 62-27
Tarkanian leads Reid 55-30
Lowden leads Berkley 55-24
Tarkanian leads Berkley 53-27
Lowden leads Goodman 48-30
Tarkanian leads Goodman 50-30
Lowden leads Miller 51-19
Tarkanian leads Miller 50-18

Boozman leads Lincoln 66-20
Boozman leads Halter 56-18

Ayotte leads Hodes 49-34
Binnie leads Hodes 46-36
Binder leads Hodes 39-37
Lamontagne ties Hodes 38-38

Crist 35 - Rubio 24 - Meek 22

Christian Liberty said...

Democrats are losing independents (the real center) not just in the senate races, but in the (generic) congressional races, says Gallup.

Independents prefer to vote for a Republican congressional candidate by a margin of 47-34. Registered voters in general prefer Republicans by a margin of 49-43.

This represents a historically high margin, a 48-year high in national pro-Republican (or anti-Democrat) sentiment. Never before in Gallup polling history have the American people preferred Republicans by this high of a margin. Even in 2002 and 1994, the highs were only +5. This year's +6 margin is just more evidence that 2010 may yet be the most pro-Republican year in at least 48 years.

Christian Liberty said...

Among independents (PPP):

Campbell 43 .Boxer 36
Devore 42 ...Boxer 35
Fiorina 42 ..Boxer 32

McCain 44 ....Glassman 28
Glassman 40 ..Hayworth 38

Democrats are indeed losing the median (center position) voters, the independents, the unaffiliated voters.

Christian Liberty said...

Democrats are indeed losing the center.

Independents prefer a Republican congressional candidate, 33%-30%

Independents prefer a Republican for congress, 40%-33%. Obama's approval %s among independents is only 44%.

Independents prefer to vote for a Republican congressional candidate by a margin of 47-34.

Fox/Opinion Dynamics:
Independents prefer a Republican candidate 23%-18% among voters ready to make a decision. When undecided voters are asked which why they would lean, the total numbers are 33%-23% in favor of a Republican representative for their district.

"the plurality (45%) of voters not affiliated with either major party now prefer the Republican candidate, while 23% like the Democrat. These findings have remained fairly consistent for months now."

Christian Liberty said...

My previous replies to this post refuted the premise that moderates (rather than independents) were the reliable indicator of the center. I also showed voluminous evidence that Democrats were indeed losing independents, thus losing the real center.

Now, in addition to showing that Democrats are losing independents, I shall grant your premise that two-party vote-share among moderates is a relevant indicator of what would occur in November, and follow this premise where it leads. If we grant the premise that performance among moderates is a significant indicator, the next question that follows is whether Democrats winning a majority of moderates would be sufficient. The answer is clearly no; winning a majority of moderates is wholly insufficient to lead a Democrat to victory in November.

Using rough national numbers, 40% of voters are conservative, 36% are moderate, and 21% are liberal.

In fact, not only are conservatives more numerous than liberals nationally (by margins of almost 2 to 1), but conservatives also outnumber liberals in each and every state of all 50 states.

Now, if Democrats win 80% of liberals and Republicans win 80% of conservatives, Democrats would actually need 70% of moderates to win November. (20x0.40 + 70x0.36 + 80x0.21) = 50%

Democrats require a supermajority of moderates in November to overcome the disadvantage of playing to their small left-wing base. If the electorate is even more partisan, Democrats would need an even larger supermajority among moderates to make up for their loss of conservative voters.

Among moderates, 70% becomes the estimated threshold, not 50%.

This means that Conway (KY), Carnahan (MO), Marshall (NC), Cunningham (NC), Bennet (CO), Fisher (OH), Sestak (Pa), Giannoulias (IL), Reid (NV), Lincoln (AR), Halter (AR), and Hodes (NH) are ALL underperforming relative to the (estimated) 70% supermajority that they would require among moderates. (Florida’s 3-way is more complicated.)

By this metric, every one of these Democrats is in electoral trouble come November.

So, in conclusion, Democrats are not only losing the center (independents), they are also underperforming among the center-left (moderates).

In the next post, we can test this hypothesis by looking at state polls on specific senate races.

Christian Liberty said...


Liberals (16% of voters): Fisher 78-9 (90-10)
Conservatives (42% of voters): Portman 67-9 (88-12)
Modertes (41% of voters): Fisher 47-27 (63.5-36.5)

If Fisher wins 90% of liberals and 12% of conservatives, Fisher would need 74.5% of moderates. Fisher only gets 63.5% of moderates, underperforming by 10 points. Likely Portman.

Liberals (19% of voters): Sestak 65-16 (80-20)
Conservatives (40% of voters): Toomey 68-12 (85-15)
Moderates (41% of voters): Sestak 45-29 (60-40)

If Sestak gets 80% of liberals and 15% of conservatives, he would need 73% of moderates. But Sestak only gets 60% of moderates in this poll, underperforming by more than 10 points. Likely Toomey.


Liberal ... 20%
Moderate.... 44%
Conservative.... 36%

Liberals: Giannoulias 81-7 (92-8)
Conservatives: Kirk 69-9 (89-11)
Moderates: Giannoulias 36-23 (61-39)

If Giannoulias gets 92% of liberals and 11% of conservatives, he would need 63% of moderates. Giannoulias gets 61% of moderates who were ready to make a decision, below expectations but within the margin of error. Tossup.

Liberal ...... 17%
Moderate....... 45%
Conservative.... 38%

Liberals: Reid 79-12 (87-13)
Moderates: Reid 51-41 (55-45)
Conservatives: Lowden 82-12 (87-13)

If Reid gets 87% of liberals and 13% of conservatives, he would need 67.3% of moderates. But Reid only gets 55% of moderates, greatly UNDERPERFORMING. Likely Republican.

Liberal .......... 9%
Moderate......... 46%
Conservative..... 45%

Liberals: Lincoln 69-10 (87-13), Halter 70-11 (86-14)
Moderates: Lincoln 49-40 (55-45), Halter 45-36 (55-45)
Conservatives: Boozman 82-9 vs Lincoln (90-10) Boozman 78-8 (91-9) vs Halter

If Lincoln gets 87% of liberals and 10% of conservatives, she would need a staggering 82% among moderates. At only 55% among moderates, she is nowhere close. And Halter was doing even worse, judging by the most recent public PPP poll (released Feb 2). Halter seems to be doing better since then, but no public PPP data is available. Based on the most recent PPP poll, it would be a nearly certain Republican pickup.

New Hampshire:
Liberal ... 18%
Moderate....... 47%
Conservative..... 35%

Liberals: Hodes 81-8 (91-9)
Moderates: Hodes 47-39 (55-45)
Conservatives: Ayotte 80-7 (92-8)

If Hodes gets 91% of liberals and 8% of conservatives, he would need 65% of moderates, but Hodes only gets 55% of moderates. Hodes underperforms by 10 points with moderates. Likely Ayotte.

So to sum up, Democrats UNDERPERFORM among moderates in KY, MO, NC (Cunningham or Marshall), Ohio, Pennsylvania, Nevada, Arkansas, and New Hampshire. In Illinois, Giannoulias is within the margin of error. Only in Colorado, does any Democrat beat expectations among moderates.

We also found (based on most recent PPP polling) that Democrats would need a supermajority of moderates to win in November. Democrats would require anywhere from 58% to 85% supermajorities among moderates to overcome their numerical disadvantages (more conservatives than liberals and often even more conservatives than moderates) and Democrats' underperformance with conservatives.

So, even if we look at Democrat performances among moderates (the center-LEFT), Democrats still underperform. In many of these races, Democrats underperform among moderates by 10 points or more.

This post confirms (by state-specific polls) the statements of the previous post. The only addendum (correction) I would need to make is that Bennet is in a stronger position in Colorado because of a high proportion of liberals (20%) and higher proportion of moderates (44%) and lower proportion of conservatives (37%). Every other state poll just confirms that Democrats are in trouble from coast to coast.

Christian Liberty said...

Now to test the assumption from the previous post with state-specific PPP polls. The assumption was that Democrats would win 80% of liberals and Republicans would win 80% of conservatives… and that 40% of voters would identify as conservatives, 40% as moderates, and 20% as liberals. For this reason, we will look at specific state polls for ideology and candidate preference crosstabs… and focus on those who have a preference, leaving out the undecideds. The conclusion from the previous post was that these assumptions would require a 65% supermajority threshold among moderates for Democrats to prevail in November. Now we will look at actual data and adjust that threshold if needed, and see if any of the Democrats actually meet that state-specific threshold among (left-leaning) moderates.

All raw data from PPP:

Conway leads 69-17 among liberals (13% of voters). Paul leads 64-11 among conservatives (49% of voters). If we throw out the undecideds and only look at those who had a preference, Conway leads 80-20 among liberals who have decided and Paul does even better, 85-15, among conservatives (which are almost four times as numerous).

To win November, Conway needs moderates (38% of voters) to support him 85%-15%. (15x0.49 + 85x0.38 + 80x0.13) = 50.05. But Conway only gets 60% of moderates to Paul’s 20% (Conway gets 75% of those decided and Paul gets 25%). Conway is actually UNDERPERFORMING among moderates and conservatives.
Advantage Rand Paul.

Another interesting note: these calculations show that if Conway only gets 15% of conservatives and 80% of liberals, he would need an even greater 85% vote-share among moderates. More evidence that Democrats in swing states must reach out to the center and to the right, not the left.

Carnahan 87-12 among liberals (15% of voters)
Blunt 79-9 among conservatives (44% of voters)

Carnahan gets 88% of liberals to Blunt 12%
Blunt gets 90% of conservatives to Carnahan 10%

Given these splits, Carnahan would need 80% of moderates to win in November. (10x0.44 + 80x0.41 + 90x0.15) = 50.7%.... Carnahan’s 61-21 split (74-26 among decideds) is UNDERPERFORMING. Advantage Blunt.

North Carolina:

Liberals (16% of voters):
Cunningham leads 16-74 (82-18 among decideds)
Marshall leads 80-10 (89-11)
Conservatives (43% of voters):
Cunningham behind 11-72 (13-87)
Marshall behind 12-74 (14-86)

Moderates (41% of voters):
Cunningham 54-26 (67.5-32.5)
Marshall 59-24 (71-29)

If Cunningham gets 13% of conservatives and 82% of liberals, Cunningham would need 76.3% of moderates… but he is only getting 67.5% of those who have decided. Given the above splits with conservatives and liberals, Cunningham is UNDERPERFORMING among moderates.

If Marshall gets 14% of conservatives and 89% of liberals, Marshall would need 72.5% of moderates, but she only gets 71%. She is also underperforming among moderates, but it is a closer race.

Both races lean toward Burr because the Democrats continue to underperform among moderates and/or conservatives.

Colorado (Bennet/Norton):

Liberals (20% of voters):Bennet 78-8 (90-10)
Conservatives (37% of voters): Norton 71-15 (82.5-17.5)
Moderates (44% of voters): Bennet 54-31 (63.5-36.5)

If Bennet wins 90% of liberals and 17.5% of conservatives, he would need 58% of moderates. Bennet now gets 63% of moderates – enough for a small lead overall. As of this poll: advantage Bennet, within the margin of error.


Christian Liberty said...

In January, after Scott Brown's victory, Tom wrote: "Even the day before they went down in flames Creigh Deeds and Jon Corzine were winning moderates in their races by margins of 56-42 and 47-37 respectively"

The leads of these candidates, who went down in flames, are statistically idental to the polling leads of Lincoln (AR), Hodes (NH), Reid (NV), Giannoulias (IL). It's quite reasonable to expect these Democrats to also go "down in flames."

Web Statistics