Archive for October, 2008

i think . . . and hope obama wins

October 30, 2008

I think. . . and hope that Obama wins.  I still stand by my initial assessment of a number of polls, which to me indicated that the race was tightening as early as mid-October.  So far every newspaper report I’ve read confirms what I wrote ten days ago on October 20.

We must still be on guard.  I’m casting a wary eye toward Pennsylvania, by the way, because I think that state could go McCain for reasons I wrote about.

But why I think Obama will win really comes down to this: there seems to be a “jump the shark” quality to McCain’s campaign.  So many easily avoidable silly things.  This is not a lament, by the way.  I’m glad he did those things . . . like pick Palin . . . absolutely no upside to this decision whatsoever . . . skip past Romney . . . whew . . . Romney exudes a certain steadiness, and that combined with his resume certainly would have made him a formidable veep . . . like i said, “whew” . . . $150,000 on clothes . . . Joe the Plumber . . . silly things like calling Obama a “socialist”.

I think people sense this “jump the shark ” quality, a certain ickiness that I think will persuade enough middle-of-the-road undecided moderates and independents to go with the one who seems the most steady, Barak Obama.

Funny . .  but I think that’s what our democracy eventually comes down to: what qualities do candidates project, not what are their policies.

In any event, this’ll be a historic vote: I for one am going to take my digital camera into the booth to photo my ballot before I turn it in.

Advertisements

missouri, new mexico, florida, pennsylvania: uh, oh for obama?

October 20, 2008

Barak Obama is at grave risk of losing if rule-of-thumbs about modern presidential election are correct.

In projecting winners, pundits see Missouri and New Mexico as critical bell-weather states.  Candidates who win at least one of these states win overall.  Other pundits say winning candidates typically win two out of three races in Florida, Ohio and Pennsylvania.

Meta-analysis of polling data indicates Obama will lose Missouri, New Mexico, and Florida.  The Tom Bradley effect may be underway in Pennsylvania, suggesting that even as he exhibits a commanding lead Obama should critically evaluate polls from the Keystone State before drawing any conclusions.

MISSOURI: McCAIN: SOLID BASE OF SUPPORT

Obama comfortably leads McCain in the latest Missouri poll from Rasmussen, 52 to 46.  Obama’s margin dwindles when numbers are averaged since September 22 – when Congress and the White House began tackling the financial crisis in earnest.  Since then, Obama has posted 48.7 average to McCain’s 46.7.

But scratch beneath the surface of the polling data and something interesting emerges.  The same data that shows Obama leading shows McCain’s base of support has hardened while Obama’s base remains somewhat tepid.

Statistically-speaking, a “hardened base” of support is expressed in terms of “standard deviation,” a rudimentary statistical tool measuring volatility around an average. Candidates want poll figures that are high and stable, i.e. not jumping up and down.  Candidates without a hardened base are susceptible to losing voters who only tepidly support them, as well as undecideds who may be swayed to the other side.

McCain’s standard deviation is 1.9 for the period stretching from September 22 to now. In other words, McCain’s polls ranged between 44.8 and 48.6, all the while averaging 46.7. Obama’s standard deviation is 2.1, meaning since September 22 his polls ranged between 46.6 and 50.8, all the while averaging 48.7.

What distinguishes McCain is that the Arizonan’s standard deviation is below the threshold separating hard from soft support, or 2.0.  We have found repeatedly that candidates with standard deviations below 2.0 that also decreased over time have come out winners, and this applies to candidates whose polling average is close enough to opponents they trail (see far below: Bush v. Kerry, New Mexico 2004).  Candidates are not required to have standard deviations lower than 2.0 to win, but indications are that those exhibiting this trait have a greater likelihood of winning.

To be sure, Obama’s standard deviation at 2.1 is close to the threshold, so this race is competitive. But candidates’ respective standard deviations suggest McCain is less likely to lose voters while Obama is vulnerable. Data below come from Daviswissing.com, Real Clear Politics, and Pollster.com.

Missouri  Prediction: John McCain.

POLLS DATE McCain Obama
Rasmussen Reports 10/15-15 46 52
CNN/Time/Opinion Research 10/11-14 49 48
Zogby 10/09-13 44 50
Rasmussen Reports 10/12-12 47 50
SurveyUSA 10/11-12 43 51
PPP 10/11-12 46 48
American Research Group 10/04-06 49 46
Rasmussen Reports 10/05-05 47 50
CNN/Time/ORh 09/28-30 48 49
SurveyUSA 09/23-24 48 46
Post-Dispatch/R2000 09/22-24 47 46
Research 2000 09/15-18 49 45
American Research Group 09/13-16 50 45
Rasmussen Reports 09/11-11 51 45
CNN/Time/OR 09/07-09 50 45
Rasmussen Reports 08/07-07 48 41
       
Long-Term Average 08/01-10/15 47.6 47.3
Short-Term Average 09/22-10/15 46.7 48.7
       
Long-Term Standard Dev 08/01-10/15 2.2 2.9
Short-Term Standard Dev 09/22-10/15 1.9 2.1

NEW MEXICO: REPEAT of BUSH-KERRY, 2004

Judging by the latest poll from Rasmussen and the poll average since September 22, Obama is blowing out McCain in New Mexico.  But McCain’s low standard deviation of 1.8 for September 22-October 13 period means that he has a strong base of support, which hardened even more since the mid-August-to-October 13 period, when the volatility index was hovering at 2.4.

Obama volatility index is too high at 3.4, different but not much different than his long-term index of 3.7  That’s a red flag.  Support for Obama is tepid — more so here than in Missouri — even if on average he’s polled higher than McCain.

We believe McCain will pull a GW Bush 2004: Bush trailed Kerry but had a tremendous low standard deviation, which decreased even more over time.  In calling New Mexico for GW three weeks before the election, I argued that when support for a candidate is tepid (i.e. Kerry), in a competitive race between two candidates, undecideds and tepid supporters will switch to the SAFE candidate.  In Bush, you had an incumbent; in McCain, you have a neighbor of Mew Mexico.

New Mexico Prediction: John McCain.

POLL DATES McCAIN OBAMA
Rasmussen 10/13-13 42 55
SurveyUSA 10/12-13 45 52
Zogby 10/09-13 44 51
Albuquerque Journal 09/29-02 40 45
Rasmussen 10/01-01 44 49
SurveyUSA 09/29-30 44 52
PPP (D) 09/17-19 42 53
American Research Group 09/14-16 44 51
Survey USA 09/14-16 44 52
Allstate/National Journal 09/11-15 42 49
Rasmussen Reports 09/08-08 49 47
CNN/Time/OR 08/24-26 40 53
Rasmussen Reports 08/20-20 41 47
Mason-Dixon 08/13-15 45 41
       
Long-Term Average 08/13-10/13 43.3 49.8
Short-Term Average 09/29-10/13 43.2 50.7
       
Long-Term Standard Dev 08/13-10/13 2.4 3.7
Short-Term Standard Dev 09/29-10/13 1.8 3.4

FLORIDA: McCAIN STRENGTHENS BASE, OBAMA HASN’T

McCain trails Obama when comparing poll averages for both candidates since September 22, 45.6 to 48.5.  With the latest Survey USA poll having McCain in the lead at 49 to 47, there is a strong hint that McCain might be catching a wind beneath his wings.  Combine this latest poll with, again, McCain’s low short-term and threshold-breaching standard deviation of 1.9 to Obama’s 2.7, there’s evidence that McCain has garnered a strong base and is positioned to cut into Obama’s tepid supporters, as well as sway undecideds.

Florida Prediction: leaning toward McCain.

POLL DATES McCain Obama
SurveyUSA 10/16-16 49 47
Research 2000 10/13-15 45 49
Hamilton 10/10-15 43 47
CNN/Time 10/11-14 46 51
InAdv/PollPosition 10/13-13 44 48
Datamar 10/12-13 42 47
Rasmussen Reports 10/12-12 46 51
Zogby 10/09-13 47 48
Florida Chamber (R) 09/30/01 45 42
Rasmussen Reports 10/08-08 47 50
Strategic Vision 10/06-08 44 52
FOX News/Rasmussen 10/05-05 45 52
Rasmussen Reports 10/05-05 45 52
Insider Advantage 09/30-30 46 49
CNN/Time/OR 09/28-30 47 51
Suffolk/WSVN 09/27-30 42 46
Quinnipiac 09/27-29 43 51
Rasmussen Reports 09/28-28 47 47
Survey USA 09/27-28 48 47
Public Policy Polling 09/27-28 46 49
American Research Group 09/23-25 46 47
Rasmussen Reports 09/24-24 48 47
Strategic Vision 09/20-22 48 45
Rasmussen Reports 09/21-21 51 46
Mason-Dixon 09/16-18 45 47
Research 2000 09/15-18 46 45
Survey USA 09/16-17 51 45
AMerican Research Group 09/14-17 46 46
SEA Polling 09/14-17 47 45
CNN/Time/OR 09/14-16 48 48
Allstate/National Journal 09/11-15 44 44
Rasmussen Reports 09/14-14 49 44
AMerican Research Group 09/14-14 46 46
Insider Advantage 09/10-10 50 42
Quinnipiac University 09/05-09 50 43
Rasmussen Reports 09/07-07 48 48
Public Policy Polling 09/06-07 50 45
Mason-Dixon 08/25-26 44 45
Strategic Vision 08/22-24 49 42
Quinnipiac University 08/17-24 47 43
Kitchens Group 08/18-21 42 39
American Research Group 08/18-20 47 46
Rasmussen Reports 08/18-18 46 43
Insider Advantage 08/11-11 48 44
Survey USA 08/01-03 50 44
       
       
Long-Term Average   46.6 46.4
Short-Term Average   45.6 48.5
       
Long-Term Standard Dev   2.4 3.1
Short-Term Standard Dev   1.9 2.7

PENNSYLVANIA: TOM BRADLEY EFFECT UNDERWAY? BUT WILL IT MATTER?

Imagine you are on Southwest Airlines.  As the jet takes off and ascends toward 10,000 feet, you experience air turbulence.  This is natural.  At 10,000 feet, you can turn on your electronic items and begin to move about, for at this height there’s every expectation of smooth sailing.  This pretty much describes Obama’s experience in Pennsylvania – where he’s scored at or above 50 in many polls – except in one regard. In the upper stratosphere, Obama is still experiencing turbulence.

While his short-term average hovers at a winning 51.2 – a decisive lead over McCain’s 40.9 average – curiously, Obama’s standard deviation is still relatively high at 2.9.  Obama’s polling average is in the part of the stratosphere where one would expect smooth sailing.

That he’s experiencing turbulence suggests fickle support among the very same voters who oxymoronically put Obama consistently above 50 points.  What’s happening?  Voters in Pennsylvania are oscillating between Obama and “undecided,” not between Obama and McCain, and why this is happening **may** be related to the Tom Bradley effect.  Our conjecture is that people for whom race is a consideration are **perhaps** telling pollsters they support Obama while in other polls they are changing their minds in **greater than expected** frequencies, resulting in Obama’s oscillation even as he hits at or above 50 points.  It is worth noting that earlier Pennsylvania was perhaps one of the most racialized Democratic primary contests (Geraldine Ferraro’s comments, the Rev. Jeremiah Wright, etc.), so to think that this dynamic has disappeared is a matter that is open to discussion.

By contrast, in another state where Obama is similarly beating McCain convincingly poll after poll — California — Obama has not only consistently scored above 50 (55 short-term average) but also managed to harden his base in scoring a below-threshold standard deviation of 1.4.

While the Tom Bradley effect **may** be underway in Pennsylvania, it remains to be seen if the magnitude of potential worst-case impacts for Obama is in the end enough to help McCain, who is substantially behind Obama.  As it is, even McCain has exhibited a volatility index above the 2.0 threshold, meaning that Pennsylvania voters are also fickle toward him.

Pennsylvania prediction: TOSS-UP, but slightly leans Obama.

POLL Dates McCain Obama
Muhlenberg 10/15-19 41 53
Morning Call Tracking 10/13-17 39 52
Muhlenberg 10/10-14 38 52
Survey USA 10/11-13 40 55
Zogby 10/09-13 40 52
Morning Call Tracking 10/09-13 38 51
Muhlenberg 10/05-09 39 50
Marist 10/05-08 41 53
Strategic Vision (R) 10/05-07 40 54
Morning Call Tracking 10/03-07 38 50
Rasmussen 10/06-06 41 54
SurveyUSA 10/05-06 40 55
Muhlenberg 09/30-04 40 50
Quinnipiac University 09/27-29 39 54
Rasmussen Reports 09/28-28 42 50
Rasmussen Reports 09/24-24 45 49
Survey USA 09/23-24 44 50
American Research Group 09/20-22 46 50
Strategic Vision 09/20-22 46 47
Allstate/National Journal 09/18-22 41 43
Mason-Dixon 09/16-18 44 46
Univerity of Wisconsin 09/14-17 45 45
Marist College 09/11-15 44 49
Rasmussen Reports 09/14-14 47 47
Quinnipiac University 09/05-09 45 48
Rasmussen Reports 09/07-07 45 47
Strategic Vision 09/05-07 45 47
CNN/Time/OR 08/24-26 43 48
Quinnipiac University 08/17-24 42 49
Rasmussen Reports 08/19-19 40 45
Susquehanna Polling 08/11-14 41 46
Franklin&Marshall College 08/04-10 41 46
       
Long-Term Average   41.9 49.6
Short-Term Average   40.9 51.2
       
Long-Term Standard Dev   2.7 3.2
Short-Term Standard Dev   2.5 2.9

SUMMARY: OBAMA LEADS BUT HIGH STANDARD DEVIATION IS TROUBLING

In the key battleground states discussed above, Obama leads in many polls, on a latest or on an average poll basis.  But he exhibits above-the-threshold standard deviations that are high enough to raise deep concerns.  This continuing phenomenon is the mathematical equivalent of what many say about Obama’s campaign thus far, particularly on Barak’s inability to “seal the deal.”

To be sure, our analysis of other states shows that Obama will win Nevada, North Carolina, and Ohio.  In these states Obama’s standard deviation is well below the 2.0 threshold, and in comparing long- to short-term standard deviations, his volatility index actually declined, meaning he has a firm base of support, in addition to a commanding **and** growing average polls.

Our analysis also shows while Obama is close to winning Virginia (though his volatility index is higher than desired), McCain’s poll numbers place him in what we refer to as the “dead-zone”, i.e. high and increasing standard deviation, low and declining poll average.

Recommendations: Obama should hold one of his 30-minute television events this week, as soon as he comes back from Hawaii.  He needs to deploy Bill and Hillary Clinton, both of whom drive media attention like no other.  Above all: Obama needs to suck as much oxygen out of the room as soon as possible – NOW – to smother so to speak any possibility of McCain building on the solid base described above.  That’s McCain’s m.o. after all: he’s the Republican come-back kid and, I believe, the stars are aligning to allow that narrative to roar back.

On a final note, Colorado will be the key, hotly, hotly contested race akin to Florida 2000 or Ohio 2004.

Below are two examples from 2004 which show the importance of the 2.0 standard deviation threshold.

Bush v. Kerry: New Mexico, 2004: Winner: Bush

POLLS Bush Kerry
09/15 – 09/16: Alburquerque Journal 43 46
09/15 – 09/16: Mason-Dixon 47 43
09/13 – 09/16: ARG 44 49
08/27 – 09/01: Alburquerque Journal 45 42
8/17 – 08/19: ARG 42 49
07/06 – 07/08: ARG 42 49
03/30 – 04/01: ? 46 45
     
Long-Term AVG 44.1 46.1
Last Four AVG 44.8 45.0
     
Long-Term ST DEV 2.0 3.0
Last Four ST DEV 1.7 3.2

Bush v. Kerry: Pennsylvaniua, 2004: Winner: Kerry

POLLS Bush Kerry
10/04 – 10/06: ARG 46 48
10/03 – 10/05: Survey USA 47 49
10/01 – 10/04: Westchester Univ 43 50
09/25 – 10/01: Rasmussen 47 47
09/27 – 09/29: Strat Vision 48 45
09/27 – 09/28: Mason Dixon 44 45
09/25 – 09/28: CNN/USAT/Gallup 49 46
09/22 – 09/26: Quinnipiac 46 49
09/17 – 09/24: Philly Inquirer 47 49
09/21 – 09/22: FOX 45 48
09/14 – 09/15: Mason-Dixon 44 45
08/26 – 08/28: STRAT VISION 48 45
08/16 – 08/22: Strategic Vision 42 48
08/11 – 08/16: Quinnipiac 42 47
07/17 – 07/21: LATimes 38 48
05/22 – 05/23: Fox TV 46 41
05/21 – 05/22: Quinnipiac 43 44
     
Long-Term AVG 45.0 46.7
Last Four AVG 45.8 48.5
     
Long-Term ST DEV 2.8 2.3
Last Four ST DEV 1.9 1.3