US PRESIDENTIAL
ELECTION Mystery polls meet demographic
shift Dinesh Sharma
On
the eve of the 2012 US election, there seems to be
magic and mystery surrounding the scientific
polls. Nobody seems quite certain about divining
the future, with results inching back and forth -
Mitt Romney ahead last week, Barack Obama catching
up this week - all within the margin of error.
In what has turned into a "cliff-hanger"
of an election, a skirmish seems to have broken
out too between different polling experts. Not
surprisingly, the lines drawn are along the party lines;
although, both the science and art of
statistics suggests everyone who claims to be an
"expert" can retrofit their predictions to the
minds of the voters as long as they're within the
margin of error; or at least have the professional
wiggle room (+/- 3%) to back out if they end up
off the mark.
The Rasmussen and Gallup
polls have favored Republican candidate Mitt
Romney, while all other tracking polls suggest
that President Obama will win by a hair. Gallup's
daily tracking poll has had Romney ahead by a
significant margin of 6% points in a random sample
of likely voters, till they had to abruptly shut
down data collection due to the super storm Sandy.
After the storm, the gap seems to have narrowed
somewhat, but Gallup still gives Romney the edge.
The Rasmussen poll is led by Scott
Rasmussen, a committed Republican, who has shown a
lead for Romney for many months in the
national-level general election. That the
difference between Rasmussen and Gallup versus the
rest of the polls has been significant - outside
the margin of error - led Andrew Sullivan of the
Daily Beast to conclude:
So you either believe that Romney
has held the national lead 100% of the time
since September 1; or you believe that Obama has
had the lead for 86% of the time since September
1. Obviously, the two models cannot both be
true.
The differences may be rooted in
political identification and the changing
demographics of America. Gallup's sampling
composition reports 36% Republicans, 35%
Democrats, and 29% Independents. "The poll of 705
other polls shows party identification as 29%
Republican; 36% Democrat; and 31% Independent.
What has happened since the summer is a sharp drop
in the "Independent" category - giving gains to
Democrats and Republicans pretty evenly, with the
Democrats gaining a tiny bit more," according to
Sullivan.
What Sullivan does not mention,
but seems fully aware of, is that party
affiliation in this election is deeply nested with
ethnicity and race, where 96% of African
Americans, who largely tend to be Democrats, are
going to vote for the president. Similarly, a
large majority of Hispanic Americans are going to
vote for the president due to immigration
policies. The growing segment of Asian American
voters seems to be backing the Democrats as well,
based on a recent survey.
Geraldo Rivera,
an investigative journalist and now a morning talk
show host, said while he is endorsing Romney-Ryan
due for their economic policies, he is going to
vote for Obama-Biden because the president's
immigration policy is so close to his heart.
In other words, a slight upsurge of one
party or a demographic group at the polls can
register a major change at the ballot box,
especially, with 24/7 news cycle, cable news
coverage and an electorate that is wired to the
Internet and i-Phones. Different segments of the
population are connected to the electoral process
to different degrees, with the unconnected being
those without access to cell phones and newer
means of social media.
The latest
non-partisan Pew poll suggests that the president
has a 3% point edge among likely voters, Obama 50%
and Romney 47%. Interestingly, the Pew sample has
proportionally less Republicans and more Democrats
than compared to the Gallup sample.
The
disparity in results may also be due to different
methods of weighting of samples. Conservatives
keep claiming that polls have a liberal bias; they
have more registered Democrats than Republicans. A
newly constructed site, UnSkewed Polls, suggests
when the polls are weighted correctly, Romney has
a 3.5% edge over Obama (51.6% to 47.8%).
UnSkewed Polls seems closer to Rasmussen
in terms of the sample composition with more
Republican likely voters. But what are they
weighting the sample to - the US Census data of
the racial and ethnic makeup of the country which
is rapidly changing or some electoral database
from the previous elections?
One way out
of the confusion may be to create a meta-poll of
all the polls, which is what the RCP national
average does, under the guidance of the
RealClearPolitics editorial team. It "averages
credible, relevant election polls, to give a more
consistent and accurate read of where the election
is headed". Since it creates an overall average of
all the relevant polls, by design it minimizes
differences; hence the election is even much more
difficult to call as per the RCP average (Obama
48.8%, Romney 48.1%).
At the state level,
where the electoral votes are gathered, the
president seems to have an edge according to a
large number of pollsters. The president seems to
be consistently leading with a small margin in
swing states of Ohio, Wisconsin, Iowa, Virginia,
New Hampshire, and Nevada, while Romney seems
ahead in Florida, North Carolina, and Colorado.
"President Barack Obama and Republican
challenger Mitt Romney are essentially tied on the
eve of Election Day, but the Democrat has a slight
edge in some of the pivotal states where the
election will be decided," according to
Reuters/Ipsos polling released on Monday.
Predictably, the futures trading sites,
such as, InTrade and Iowa Electronic Market (IEM)
have also given the president an edge. InTrade has
been predicting that Obama has 67% likelihood of
winning, while Romney 33%. Likewise, IEM has been
predicting for almost a year that Obama will
snatch a narrow victory (51% vs. 49%) in a very
close election.
Whoever predicts this
election accurately, it will be because they
grasped the changing demographic reality of
America. As the Republican Senator Lindsey Graham
said today, "If we lose this election, there is
only one explanation - demographics."
Head
Office: Unit B, 16/F, Li Dong Building, No. 9 Li Yuen Street East,
Central, Hong Kong Thailand Bureau:
11/13 Petchkasem Road, Hua Hin, Prachuab Kirikhan, Thailand 77110
