Las Vegas Sun

September 16, 2019

Currently: 87° — Complete forecast

Pollsters’ difficulties result in dearth of polling ahead of Nevada caucuses

Clark County Republican Convention

Mikayla Whitmore

Delegates from the Republican caucuses attend the Clark County Convention to elect representatives to the state convention at the Rio All-Suite Hotel & Casino on Saturday, April 9, 2016..

Nevada is one of the earliest states in the presidential nominating process — the third overall behind the Iowa caucuses and the New Hampshire primary — and is seen by many as a bellwether for voting trends in the expansive American West.

Presidential preference polling hasn’t been as frequent in Nevada as in other states with early presidential caucuses or primaries. RealClearPolitics has two Nevada-specific polls listed in its aggregation since July, and the Monmouth University Polling Institute, one of the most respected polling groups in the nation, has only polled Nevada once during this primary season, in early June. Change Research has done two, and Gravis Marketing released its first caucus poll on Aug. 20.

Mark Mellman, president and CEO of the Washington, D.C.-based political research firm The Mellman Group, said that Nevada presented a lot of “unique challenges” for polling firms.

“Unless you’re really familiar with the state and familiar with the challenges, you’re going to get it wrong,” he said.

But while the state presents challenges, polls conducted near the 2016 elections were generally not too far off the mark statewide, with only a few near the election falling outside their announced margin of error.

Work hours

Frequent polling in Nevada, as opposed to other early voting states, is lacking for multiple factors, including odd work hours for Nevada voters and the relative newness of Nevada’s “first in the West” caucus, pollsters say.

Las Vegas, with its economy based around tourism, has a large population of service workers with hours outside of the traditional 9-5 workday. That makes it difficult for traditional polling, pollsters say.

“A lot of people work at night in Las Vegas,” Mellman said. “That’s very different than it is in other states where you have a much heavier workforce during the day.”

Brad Coker with Mason-Dixon Polling and Strategy, echoed Mellman. Coker said that most traditional telephone polling is done during the evening or on weekends, in order to reach the biggest range of voter demographics.

“Typically, if you call during the daytime, polls get skewed very heavily toward older women,” he said.

Coker said rural areas should be easier to poll in Nevada due to the fact that most overnight workers are in urban areas. It’s a trend that doesn’t apply in other states, he said, where urban and rural areas generally offer the same ease of access to voters.

Relative newcomer

With a little help from then-Sen. Harry Reid, Nevada’s presidential caucuses were moved to early in the calendar starting in 2008. Before that, caucus turnout in Nevada was relatively low, as the contests were late enough in the process that the ultimate party nominee would have gained a foothold in other state primaries or caucuses.

As such, there’s only been three “early” presidential caucuses in Nevada, and pollsters say that Nevadans are less aware of the process than other early states because of its newcomer status.

Mellman said that caucuses in general are harder to poll due to the lower turnout.

Patrick Murray, the director of the Monmouth Polling Institute, said caucuses tend to have lower turnout than primaries, and Nevada caucus turnout is lower than other early states like Iowa. “It’s not as an established presence on the early primary calendar as these other states are,” he said.

New Hampshire has had the first primary since 1920, and Iowa has had the first caucuses since 1972.

The caucus process, both Mellman and Murray said, is not as entrenched in people's minds in Nevada.

“People are not quite as aware about what (the caucus) will entail and how to do it,” Murray said.

Murray said it could be twice as expensive to poll Nevada as other states due to the difficulty in actually finding likely participants. Less likely caucus-goers means more phone calls, which means more money spent.

“You take that all into account, it’s probably why a lot of pollsters are looking at the other states,” he said.

Shift of polling responsibility

Coker said that polling, traditionally the domain of news organizations, has shifted to universities and other organizations since the newspaper industry has taken financial hits.

“I think the NCAA has taken over the media polling business,” he joked.

He said that there were historical connections to media in other early states that are interested in doing polling. New Hampshire, for example, is connected to the Boston media market, he said, and Iowa has the Des Moines Register poll.

“I just think there’s more tradition and more history with those two,” he said, adding that newspaper polling in Nevada has “slowed to a trickle.”

Mellman said local entities were generally the ones who picked up polling costs.

“Good polling is also expensive” he said. “Somebody’s got to be willing and able to spend the money on it to make it work.”

Polling can indeed be pricey. Coker said that, ballpark, the cheap side of statewide polling can be around $5,000-6,000 and the more expensive side can run $20,000-$25,000. Caucus polling is almost always more expensive due to lower voter turnout, which means there are fewer people pollsters can use in their sample of “likely caucus-goers.”

The number of questions can also affect price, as can the length the polling call will take. The longer the poll, the more it costs — Coker compared the process to buying a house.

“Do you want a doublewide or do you want a McMansion somewhere?” he said.

Other Nevada issues

Murray said the national narrative around which candidate is ahead in polls seems to be driving state-level opinion in “a way we haven’t seen before.”

He said this national narrative could harm Nevada’s clout going into 2020, as it could diminish the impact a Nevada decision may have, if a national sense of a frontrunner takes over before the caucuses. “It really became a national media game must quicker than it has in the past,” he said.

Murray also said that the rollout of a virtual caucus — which will be done by phone in both Iowa and Nevada this cycle — is a new development facing pollsters. He said the overall accuracy of virtual caucus polling won’t be known until after the caucus results are tallied.

2016 Nevada polling

RealClearPolitics, which aggregates poll results in elections, lists 30 polls conducted in 2016 matching up Hillary Clinton and Donald Trump.

The average it found was that Trump led Clinton by 0.8 percentage points headed into the election. Clinton won Nevada by 2.4 percentage points.

Polls had Trump and Clinton moving back and forth over the months leading to the general election.

While Clinton won Nevada by 3.2 percentage points more than predicted — going off the RealClearPolitics average — multiple polls had her up 4-7 points in the race. Trump had his share of good polling in the state, as well, with multiple polls giving him a 4-6-point lead in the race.

Interestingly, few polls conducted near the election in Nevada were substantially outside the margin of error. The margin of error is the percentage poll results could differ from the real population value.

For example, a poll that predicts Tom Steyer winning the Nevada caucus by 4 percentage points with a plus or minus 3% margin of error is saying that the total percentage points could be from 1 to 7. A larger margin of error means the poll’s results are possibly less accurate.

Ahead of the 2016 election, pollsters like the Emerson College Polling Society, 8 News Now and Gravis Marketing all had polls close to the election in which the results were well within the margin of error.

Others didn’t match up so well with the final result. CNN had Trump ahead by 6 points with “likely voters” with a 3.5% margin of error — the final result was an 8.4 percentage point difference from their prediction.

Others were less substantially outside the margin of error. Remington Research had Trump up by 1 percentage point with a 2.31% margin of error, which put their prediction at 3.4 percentage points away from the total, or a little over 1% from the margin of error.

Uncertainty is built into the polls as well with what is called a confidence level. For example, if a confidence level is at 95% for a poll, then there is a 95% chance the poll is within the margin of error.

Ultimately, most pollsters in Nevada got it at least close to right in 2016. RealClearPolitics averaged the results of the polls it tracked at .8 percentage points in favor of Trump. With Clinton’s win with 2.4 percentage points, the average was off by 3.2.