The Democratic primary election: What happened?

NOTE: Last week,
SoonerPoll sent this critical analysis of its own polling to
CapitolBeatOK. Without editorial revision or commentary, it appears
below. To access the tables and other aspects of the analysis, visit
here (

Prior to the July 27th
primary election, SoonerPoll was one of several pollsters in the state
who had Lt. Governor Jari Askins losing to Attorney General Drew
Edmondson in the Democratic Primary Race for Governor.

Since the election,
researchers at SoonerPoll have been examining voter turnout and
conducting extensive polling in order find out why pre-election polls
conducted by our house and others were so drastically different from
election results.

“We have waited until now
to address the issue because we wanted to have hard evidence as to what
happened in the election,” Bill Shapard, CEO of SoonerPoll, said. “It
would be of no use to us or the public at large if we had tried to
explain this phenomenon without empirical data from the state election
board and our own follow up polling.”

Voter Turnout

Voter turnout often plays
a significant role in an election and in this election it was one of
many factors that helped decide the outcome.

When developing a voter
turnout model in the state of Oklahoma two counties, Tulsa and Oklahoma,
carry more weight than rest because of their population size.

Prior to the election,
Tulsa County was considered to be Edmondson’s base.  Edmondson won Tulsa
County with 61.8 percent of the vote on Election Day, however Tulsa
county had the fourth lowest turnout in the state.

“Since we live in a state
so dominated by two large cities, a drastically low turnout in either
city would result in a broken turnout model,” Shapard said.

Voter turnout in this
season’s primary elections was low for Democrats across the nation.  In
fact,  a CBS special report found that the Democratic turnout in
Oklahoma was the lowest on record.  Conversely, Republican turnout is
very high nationwide and Republican turnout in Oklahoma set a record

Turnout was especially
low in Tulsa County, where only 19.9 percent of registered Democrats
voted in the election. Tulsa County had the fourth lowest turnout in the
state, well below the average county turnout of 26.4 percent.

Edmondson won Tulsa
County by a better than 3 to 2 margin, receiving 61.7 percent of the
vote in the primary. If that margin had held, Edmondson would have won
the election if voter turnout in Tulsa County had been 5.67 points
better, or 25.6 percent, just under the statewide average.

Similarly, Muskogee,
Wagoner and Washington counties were all counties considered to be part
of Edmondson’s base that had below average turnout.

“Had Edmondson been able
to ensure just an average turnout in his base counties, the election
would have gone his way,” Shapard said.

Getting out the vote

State election board data
shows counties that were considered to be Edmondson’s base had below
average turnout in the primary, while Counties considered to be Askins
base had an above average turnout.

In order to measure how
“Get Out the Vote” campaigns effected the primary, SoonerPoll
commissioned and conducted a poll of 450 registered Democrats who voted
in the July 27th primary and 401 registered Democrats who were likely to
vote in the July 27th primary but did not vote.

Of all those who voted in
the primary, 58 percent said they received no contact from the
Edmondson campaign while only 50 percent said they received no contact
from the Askins campaign

When asked about whether
they received mail from one of the candidates 36 percent said they had
received mail from Askins, compared to only 30 percent from
Edmondson. The same was true when voters were asked about recorded
calls, 25 percent said they had received a call from the Askins
campaign, while only 20 percent received a call from the Edmondson

When the probability of
voter turnout is analyzed through a binary logistic regression model,
associations between the Askins campaign contact tactics of mail pieces,
live phone calls and literature and voting was identified.

Neither contacts by the
Edmondson campaign or historic controls for voter participation — age,
income, education — were statistically significant.

“The three contacting
methods identified in the regression model were more effective in
mobilizing voter turnout than any other contacting methods used by
either Askins or Edmondson,“ Ryan Nelson, Senior Research Analyst at
SoonerPoll, said.

The Oklahoma Gazette
reported immediately after the July primary on the Askins’ “Get Out the
Vote” [GOTV] strategy. The strategy was based in part on selective
targeting and mobilization of voters in areas where Askins or her
legislative allies were deemed politically-strong, and also on an
aggressive ground game that used personal contacting, mail, phones, and
literature to mobilize select voters.

“Around the margins, it
was the Askins ground game and GOTV effort that drove the shape of the
electorate in a low-turnout primary,” Dr. Keith Gaddie, Vice-President
of SoonerPoll, observed.

President Obama

The poll conducted by
SoonerPoll following the primary election asked respondents a variety of
other questions, including whether they approve of President Barack

In the current political
environment, the president and his policies have become the focus of
political debate from the national to the local level. The president’s
approval rating, while never strong among voters in Oklahoma,
nonetheless has consequences for Democratic voter participation in

The poll results reveal
that those with strong opinions for or against the president were more
likely to vote than those with weaker or no opinions.

Primary voters who
strongly support Obama were more likely to prefer Askins over Edmondson
with few undecided, while voters who strongly oppose Obama were evenly
split with about a third in favor of Askins, about a third in favor of
Edmondson and about a third undecided.

“Obama had a polarizing
effect on the Democrat electorate, those undecided Democratic Primary
voters who oppose Obama had difficulty picking a candidate because
neither candidate represented the anti-Obama candidate,” Shapard
said.  “For obvious political reasons, neither Askins nor Edmondson
wanted to distance themselves from their party in Washington or Obama.”

When the Obama approval ratings are compared to how Democrats intend to vote in November a similar trend is revealed.

Those who strongly
approve of Obama are more likely to say they will vote for Askins,
inversely, Democrats that do not approve of Obama are more likely to
either support Mary Fallin or be undecided.

“Right now, polling shows
that Mary Fallin is receiving a large number of crossover votes from
Democrats who are unsatisfied with Obama and have nowhere else to turn,”
Shapard said.

Gaddie agrees, and sees
this as a hurdle for all Democratic competitors this November. “The
challenge for Democrats on the ballot, including Ms. Askins, is to
figure out how to invigorate and mobilize these Democrats, who need to
turnout and support her at an 85 percent clip to prevail in November,”
Gaddie said.

Impact on the general election

When the respondents who
voted in the primary election were asked who they plan to vote for in
the general election, 57 percent said Askins and 24 percent said Fallin
while 20 percent remained undecided. Of those who did not vote, 46
percent said Askins and 28 percent said Fallin while 26 percent remained

A deeper look into
secondary polling data indicates that Democrats who stayed home in the
primary election were ‘disenfranchised’ with the party.  Non-voters were
slightly more likely to be conservative, disapprove of Obama, and not
have a preference in the primary.

“The large number of
undecideds and Fallin supporters among those who did not vote, indicates
that many Democrats may have sat out of the primary election because
they anticipated voting Republican in November regardless of who the
candidate was,” Shapard said.

In the 2008 election,
49.3 percent of Democrats, 43.8 percent of Republicans, and 6.7 percent
of Independents turned out to vote. Using that turn out model and the
rule of thumb that 90 percent of Republicans will vote for the
Republican candidate and Democrats and Republicans will split the
Independent vote, Republicans need as little as 15 percent of Democratic
votes to win the election.

Historically, statewide
Republican candidates have needed a much larger percent of the
Democratic vote to win. That amount has slowly decreased overtime as
Republicans have increased in registration while Democratic turnout has

With 24 percent of
Democratic voters and 28 percent Democratic non-voters in favor of the
Republican nominee, the 15 point threshold is a daunting figure. The
most recent SoonerPoll conducted earlier this month shows 31.2 percent
of Democrats are now in favor Fallin, raising the hurdle even higher for
the Askins campaign.

“Right now the election
is Fallin’s to lose,” Shapard said. “She has a 17.3 percent crossover
advantage and has the support of both Republicans and Democrats who do
not approve of the president,” Shapard said.

The Fallin campaign has
been able to capitalize on the low approval of Obama among
Democrats. Conservative political groups unaffiliated with Fallin’s
campaign have been running commercials which link Askins to Obama. so
far the commercials have been effective and will probably continue to be
until Askins makes an effort to distance herself from the president.

“If Askins has any hope
of making up ground in this race she is going to have to focus within
her own party,” Shapard said. “Democrats like Joe Manchin from West
Virginia have been able to overcome similar situations by giving
Democrats who don’t approve of Obama a Democratic candidate to run to.”

Gaddie says that the Askins campaign needs a game-changer in order to stay competitive in this election.

“The initial debate has
telegraphed a plan to challenge Republican characterizations of the
Lieutenant Governor that tie her to unpopular national Democratic
politicians and policies,” Gaddie said. “They were able to run stealthy
in the primary in order to break the turnout model. There is no such
privilege in a high-salience general election with an energized
conservative electorate. They have to move, aggressively and in the
open, if they want to overcome the ambivalence of a significant and
important portion of the Democratic electorate.”

Following the
election SoonerPoll commissioned and conducted a poll of 450 registered
Democrats who voted in the July 27th Primary and 401 registered
Democrats who were likely to vote in the July 27th primary but did not. 
The scientific study used live interviewers by telephone from, 2010 and
has a margin of error of ± 4.62 percent among those who voted and ±
4.89 percent among those who did not vote.