Friday, September 9, 2016

THE PRAIRIE EDITOR: The End Of Polling As We Know It?

It is a very human habit to try to measure as much as we
can. Until relatively very recently, we have assumed that our
measurements were true and useful. Measuring, after all, is

Then along came the physicist Werner Karl Heisenberg, and his
now legendary “uncertainty principle” that demonstrated that
our measurements, especially of very small matters, were not
at all certain and true.

Nevertheless, even if they were not exact or certain, our
measurements remained presumably useful. After all, we used
them to construct buildings, highways, and machines; farm fields
and sell goods and services. We also devised a very special
measurement, conceded to be inexact, but which we nevertheless
assumed to be useful or accurate. That special measurement,
employed particularly in democratic capitalist societies, was
polling. Most specifically, we have seen polling as an adjunct of
the election process rise to almost an obsession.

Like a narcotic drug, political polls now addict most political
journalists, campaign operatives and heavy observers.

In its infancy, U.S. political polling made some historic errors.
There was the primitive Liberty Magazine poll in 1936 which was
subsequently promoted as an accurate prediction of incumbent
President Franklin Roosevelt’s imminent landslide defeat by
Kansas Governor Alf Landon in 1936. Roosevelt subsequently
won one of the largest landslide vistories in history. Twelve years
later, the two major national U.S. polls predicted a decisive victory
for New York Governor Thomas Dewey over President Harry
Truman. Truman then won by a huge margin in 1948.

Since that time, of course, we have been told that political polling
has become a science. There is always a conceded margin of error,
but properly done polls, we are also told, are scientifically accurate,
with the caveat that they are accurate only at the moment time they
are taken.

For several decades, these assumptions seemed to be true, with
very occasional exceptions. But then, in the past few decades, as
myriads of polls, “respectable” or not were created and conducted,
the rare “wrong” poll became more and more commonplace.
Even as the polls were more and more in error, the political class,
including the media, became more and more dependent on them ---
to the point of the present addiction.

In 2016, with a grass roots uprising I have named the “mutiny of
the masses,” I think we are approaching a critical mass of political
poll credibility. The reasons for this are both technological and
psychological. Door-to-door polling has become too expensive, and
landline telephone polling no longer reaches a meaningful group of
voters, as as well as seems too intimidating. Robo-calling and online
polling, on the other hand, are not personal enough. Heisenberg
lives again! The measurer unintentionally (or sometimes, intentionally)
alters what (or who) is being measured.

With two months to go, I contend that most polls do not accurately
or usefully measure the presidential race (especially by the
standards of past cycles). I do agree that, as we get very close to
election day, the polls will become more and more accurate, but if
the presidential contest is very close, the polls might not predict the

Ask former President Alf Landon and former President Thomas
Dewey, and their supporters, how that will feel.

Copyright (c) 2016 by Barry Casselman. All rights reserved.

No comments:

Post a Comment