The Republican primary in Georgia just concluded
demonstrates one more time how increasingly unreliable
political polling has become, especially in primary
A just-before-the-primary poll in that race had Jack
Kingston five points ahead of his opponent David
Perdue, In the actual voting, Perdue won by almost two
As one of the nation’s savviest academic political pundits,
Steven Schier of Carleton College, has observed, “Primary
polls are among the most unreliable because it is very
difficult to identify those who will actually vote.”
One should always read the fine print on polls. Two of the
most important factors are the size of the poll, and whether
those polled are “registered voters” or “likely voters.”
A third factor is the so-called “margin of error.” Polls under
1000 participants obviously have the highest margins of
error, but even “margins of error,” as usually reported, are
subjective or distorted, except by the most objective and
careful polling firms. “Registered voters” as a category today
is almost meaningless in any race that is competitive. “Likely
voters” is by definition a subjective category.
I have written about this before, but polling accuracy conditions
continue to become less and less reliable.
Polls can still be useful in measuring are short-term and
intermediate changing trends in voter attitudes, but even then
the standards in the polling process should be high.
I am repeating all of this because we are about to hold an
important national mid-term election. This will be followed
almost immediately by the race for the presidential nomination
of each national party, and that will be followed by the 2016
presidential election and congressional elections. Much of the
reporting on these contests will be based on polls.
The rise of partisan polling for public consumption (in contrast
to private traditional and legitimate partisan polling for
candidates and parties for campaign use) has been significant
in recent years, and the opportunities for these polls to mislead
voters has also risen alarmingly. Private polling remains a
valuable and necessary tool for political campaigns, and as I
have previously pointed out, are among the most accurate and
realistic polls (out of necessity).
Even news organizations which publish amalgamated polls
(a combination of all or most polls) in a particular political race
face increasing distortion as the number of partisan and
less-than-professional polls are included in them.
Pollsters make their living by their polling, and are not likely
to be self-critical of their profession. Journalists obviously find
that polls make their jobs easier, and are prone to accept poll
numbers uncritically. Politician, political parties and political
consultants use favorable polls to obvious public relations
But in the barrage of all those poll numbers to come, who is
looking at them critically on behalf of the most important
consumers, i.e., readers and voters?
Caveat suffragator! Let the voter be wary!
Copyright (c) 2014 by Barry Casselman. All rights reserved.