Monday, February 27, 2012

Too Much Polling Too Early: Caveat Lector!

I notice a chronic difficulty for many political analysts this
election cycle that is becoming particularly misleading about
probable outcomes in November. As in the 2010 cycle, political
writers and commentators are relying too much on a plethora
of polls, many of them defective in the way they pose their
questions, or their size, or most critically, the increasing
arbitrariness with which they weight their raw data before
publishing their results. The consequence is that discussions
which anticipate future results are becoming more and more
poll-oriented, and I suggest, as happened in 2010, more plainly

The volatility of the 2012 Republican presidential nomination
contest with its historically unprecedented swings in poll numbers
should be a warning not only to political analysts, but to all who
read about this cycle's political outcomes.

"What is the alternative?" some might ask. There is an alternative,
and it is more substantive examination of the candidates,
demographics, economic and political trends, and most importantly,
facts on the ground.

Two cases in point: 1. The recent poll bubble for Rick Santorum in
the GOP contest; and 2. The current discussion about control of
the U.S. senate after 2012.

A cursory look at Mr. Santorum's record as a congressman and
senator, a simple review of his controversial views on social issues,
and an examination of his performance in the 2011-12 pre-primary
campaign should have led to a great deal of skepticism about his
ability to succeed this year, even after his inconclusive three-state
primary/caucus victories recently. But this did not happen.
Instead, there was a mostly uncritical wave accepting him as a serious
candidate to win the GOP nomination.

The inescapable facts are that there are about twice as many
Democratic U.S. senate seats than Republican seats up for
re-election this year. Of the Republican-held seats only two could
be described as vulnerable, but of the Democratic seats, about a
dozen are vulnerable, many of them VERY vulnerable. Not only
that, the Republicans have recruited in most cases, strong
candidates for Democratic seats held by incumbents (in contrast
to 2010 when they did pick up six seats, but also nominated
several weak candidates who could have also won). Finally,
nothing has happened substantially which has altered the fact
that the Democratic Obama administration and the Democratic-
controlled U.S. senate are on the defensive on most hot-button
issues facing the nation, including health care, taxes, federal
spending, the over-reaching of the judiciary, government
intrusion on the private sector, conduct of foreign policy in the
Middle East, Asia, and South America, and the weakening of
the military. Of course, no one knows precisely how many net
seats the Republicans will pick up. I think it could be as few as
five or as many as eleven, but even if it were only four, that would
give the GOP control. Yet many commentators are suggesting,
based on current and very early polling, that Republican control
is in doubt.

It goes without saying that we don't know about all the events
and circumstances that might appear between now and November.
Historically, there are usually one or more major surprises, any of
which can affect the outcome of a presidential election or of some
congressional elections. In 2008, it was the mortgage banking
crisis which doomed Mr. McCain's bid for the presidency. No doubt
there will be one or more this year, affecting either Democratic or
Republican candidates. It was mathematically possible for Mr.
Santorum to win his party's nomination (although it is now
meaningfully very unlikely). It is mathematically possible for
Democrats to retain control of the U.S. senate, but barring some
extraordinary circumstance, it won't even be close.

Polling and well-run polls can be useful, especially close to the
actual voting when they are often reasonably accurate. But
excessive speculation based on a grab-bag plethora of polls is
not responsible discussion and journalism.

Caveat lector!

Copyright (c) 2012 by Barry Casselman
All rights reserved.

1 comment:

  1. In today's "immediate gratification" media world, the concept of a "long term view" of the political landscape rarely extends beyond hours much less days, weeks or months. Polling is just one manifestation of snapshots in time that become short-term self-fullfilling prophecies- until the next poll. A broader analysis of candidates or what they have said and done over time hardly seems relevant anymore. If the Prairie Editor can help us all take a deep breath and gather ourselves once in a while, some of the peaks and valleys of public opinion will be smoothed out just a little.