Most election pollsters aren't really in it for the money

Every day in this election cycle it seems like there are new
polls about the presidential race, and it seems like
somebody must be making a lot of money from all
that data collection.

But here's the thing: Most of these polls were never meant
to make money. They are just marketing and brand
placement to help sell something else.

Even the biggest firms, like Gallup, give away political
polls to promote the commercial market research for
which corporate clients will pay hundreds of millions
of dollars. Information like "how often do you shave?"
and "what kind of razors do you buy?"
helps consumer products firms hone their businesses.

Or take colleges and universities like Quinnipiac, Marist
and Monmouth. If it weren't for their polling operations,
you may not recognize those names at all. That brand
recognition attracts more applications, better students
and more prestige and tuition money for the schools.

The only real money in election polling goes to the
pollsters hired by the campaigns. And those millions
of dollars are less for the polls themselves than the
guru-like advice the pollsters provide to the candidates
to shape messaging and strategy.

"Most polls are a loss leader for marketing," said one
California-based strategist who has managed many local
and statewide campaigns there. "It's about creating a
brand that people recognize, which in turn legitimizes the
research that companies will pay for."

Survey research companies like Public Policy Polling,
YouGov and American Research Group publish dozens
of polls that get their names out in public during the
election. The exposure can pay off in private marketing
efforts later — a market that is about five times bigger
than the one for political polling.

From a campaign's point of view, all those individual polls
are "essentially promotion" rather than useful data points,
said the strategist, who declined to be named to preserve
his business relationships. He said that fundamentally,
they are a way to tell the score in a horse race, feeding
the public's need for daily data.

Like many polling companies, many colleges and
universities are also now using political polling as
a loss-leader, trading information for public exposure
and better name recognition. About a third of political
polls this year have been conducted by colleges
and universities, compared to less than 15 percent
in past years, according to a CNBC analysis of
data maintained by news site FiveThirtyEight.

Many are following the blueprint established by
Quinnipiac University, which used a prominent polling
program to aid in a rapid expansion that tripled its
student body. Today, more than 200 schools are
members of the American Association for Public
Opinion Research, according to data provided by a
spokesman.

"That's one of the benefits, and certainly something that
the administration likes," said Spencer Kimball, a pollster
and the advisor for the Emerson College Polling Society,
which uses polling to teach students and has dramatically
increased its poll output in this election cycle. "They like
to see the college names in the papers."

Emerson, which uses robo-calling, is spending about
$100,000 on polls this cycle and expects it will have
published about 80 by the time the election is over. That's
a tiny budget compared to schools that are using phone
banks, which cost $12,000 to $15,000 per poll
if students are running the phones or $18,000 to $20,000
if they're using outside phone banks, Kimball said.

Michael Traugott, a professor who studies polling at the
University of Michigan and is a former president of
AAPOR, echoed those estimates, adding that a complex
live interview poll could cost up to $40,000, but
robo-polling can be as cheap as $5,000 to $10,000
for a typical poll. That's a relatively low cost for the
visibility polls can bring.

But not all polls are equally accurate, and some in the
polling industry say that robo-polling (which, by law,
can't contact cell phones) and misuse of statistical
techniques have eroded poll quality in some cases.
One veteran pollster told CNBC that in his past working
days, certain colleges would call him up asking what
their poll results should be.

"Technology has allowed people to enter the field
without much formal training and we see that
periodically with some problems that pollsters
have," Traugott said. "You can run an operation with
really low overhead, you can pay somebody to do the
interview and you can even pay
somebody to analyze the data for you."

Some of the most prominent polls are conducted by large
media outlets, often in partnership with other companies.
Those include CNN's polls with Opinion Research
Corporation, NBC's new deal with Survey Monkey, Fox
News' surveys with Anderson Robbins Research and Shaw
and Company Research, and CBS's 40-year partnership
with the New York Times.

Those companies, as well as many smaller local outlets,
have an obvious interest in creating news through polling.
But there isn't much money to be made in running media
polls.

"Media pollsters come and go, disappear left and right,"
Jay Leve, CEO of SurveyUSA, said in an email. "You
operate on tiny margins."

Media organizations have been under a lot of pressure to
cut costs, while at the same time improving data quality,
Traugott said. Those cost considerations have some
pollsters relying on weighting algorithms that can skew the
results — for instance, the recent criticism leveled at the
methodology used by the University of Southern California
and the Los Angeles Times, which let one respondent have
undue influence on the findings.

"This is a fairly direct impact of the financial hard times in
the industry," Traugott said. "For some pollsters they cut
back on the number of polls, for others they try to cut costs
by altering the mode of data collection by going from the
phone to web or face-to-face interviews to robo-calls."