Las Vegas Sun

February 17, 2019

Currently: 51° — Complete forecast

Jon Ralston:

Latest poll on Nevada reveals little – on the surface

A couple of days before the 2010 election, after I had predicted Senate Majority Leader Harry Reid would survive, I received an email from a smart GOP operative telling me I was “very bold” but likely wrong.

“We’re crushing Harry with independents,” the Republican told me of Sharron Angle. “(The) latest (polls) are our best of the campaign.”

I relate this story not simply to fondly recall that I was correct — although that never fails to make the pundit’s chest swell — but to remind everyone in the wake of the latest Nevada numbers that polls often are wrong and at the very least can be misleading. More to the point, polls taken more than five months before an election reveal very little on the surface that we don’t already know, the proverbial snapshot in time, a selected universe, a state of play.

So when an NBC/Marist poll released this week says the presidential and U.S. Senate races in Nevada are toss-ups – 48-46 for President Barack Obama and 46-44, Sen. Dean Heller — my visceral reply is: No kidding.

But being a poll junkie, I dived into the crosstabs, which can tell you something about the survey’s validity and also set the contours that will help determine the outcomes in November. As the silly season intensifies, with polls likely to be released by independent sources such as NBC/Marist to set baselines and not-so-independent sources with a tendentious purpose (usually fundraising), it’s worth examining this poll to remember what to look for.

Back in 2010, many polls erroneously created the impression that Angle was likely to defeat Reid. But there were problems with almost every one of those surveys, easily discovered by exploring the internals. And Reid’s pollster, Mark Mellman, turned out to have the only consistently correct numbers for one reason: His model of what the turnout was going to look like most closely approximated what it actually was.

The NBC/Marist poll makes some assumptions that might be instructive for November, but also might skew the current results. Or … they might not.

What the survey certainly does is highlight the key factors that will determine the outcomes for the top-of-the-ticket races — and maybe others, too. Three keys to watch:

• Regional turnout: Nevada is three states – Southern, Northern and Rural. The greater the percentage of the vote in Clark County, the better for Democrats, who have a huge registration edge. The NBC/Marist poll turnout model indicates Clark County will make up 72 percent of the vote in November. That’s high by about 5 percentage points based on the past two elections, which shortchanges super-conservative rural Nevada by almost that much. Washoe County, the swing county, was closer. But if this is what the turnout looks like in November, which is unlikely, President Obama is likely to win the state and Rep. Shelley Berkley could be a U.S. senator. On the other hand, the NBC/Marist poll showed Berkley winning by 6 points in Clark and losing by 12 in Washoe — I am willing to bet Mitt Romney’s $10,000 that if those are the results in November, based on what the turnout likely will be, she will lose by a substantial margin. Similarly, Obama is shown only winning Clark by 5 points — if that happens, he will lose the state.

• Hispanic turnout: The NBC/Marist poll has Hispanics as 19 percent of the electorate, which is something that has never happened in the state’s history. The highest was the past two cycles, in which 15 percent of voters were Latinos. If 19 percent does occur — it’s not impossible — I find it hard to believe either Heller or Mitt Romney could win. One caveat: The survey indicates that of the Hispanics surveyed, Obama is up 61-33 and Berkley is up 57-39 — not nearly so large as Obama in 2008 or Reid in 2010. If those percentages are correct, both Berkley and Obama are in trouble.

• Gender turnout: Females made up 50 percent of the NBC/Marist universe, which seems a little low — it’s usually about 52 percent. There is a gender gap — Obama leads among women by 54-40 and Berkley by 50-40. So they are slighted in this survey.

There are other factors — how many seniors turn out, what the base loyalty is and what the partisan enthusiasm levels are. But regional, Hispanic and gender turnout will tell us most of what we need to know, so those numbers are worth tracking in all polls.

Another reason I hearkened back to that 2010 moment was how Republicans here and in D.C. chafed Thursday at the NBC/Marist results because they thought they skewed Democratic. Maybe they are.

But let me make another very bold prediction that only a thoughtful, nearly omniscient pundit can make: Maybe they aren’t.

Join the Discussion:

Check this out for a full explanation of our conversion to the LiveFyre commenting system and instructions on how to sign up for an account.

Full comments policy