Polling the Polls: Is Darryl Glenn 10 Points Behind Michael Bennet or 2 Ahead?
Michael Bennet and Darryl Glenn during their only debate face-off to date, during a Club 20-sponsored event in Grand Junction. Additional photos below.
This week on his Facebook page, Darryl Glenn, the Republican candidate for Colorado's U.S. Senate seat, shared the results of a September 23 Gravis poll showing him leading Democratic Senator Michael Bennet by two points.
A cause to celebrate? Perhaps — but only if Glenn and his backers ignore a September 25 CNN/ORC poll in which Bennet is said to be out in front by ten points.
The twelve-point spread between these two surveys is extreme, especially in comparison with the numbers pertaining to the presidential contest. The CNN/ORC poll has Hillary Clinton leading Donald Trump by one point among registered voters but trailing by the same margin when it comes to likely voters, while Gravis gives Trump a four-point registered-voters lead.
This deluge of digits will likely leave the average voter in a state of confusion over what's really going on, and Floyd Ciruli of Ciruli & Associates, a Colorado-based polling and consulting firm, understands. But he has some tips about how to figure out which polls are the most reliable.
Gravis, for instance, calls itself nonpartisan, but Ciruli says, "They're a Republican firm," and the latest poll was issued on behalf of Breitbart News, which he describes, accurately, as "very conservative and pretty pro-Trump. In fact, their chairman [Stephen K. Bannon] is currently running Trump's campaign."
Darryl Glenn and a friend.
On the other side of the coin is Public Policy Polling, or PPP, a prominent company that Ciruli describes as "a Democratic firm. So when you see one of their polls, it's something to keep in mind."
As this last phrase implies, Ciruli doesn't completely dismiss the work of Gravis and PPP, which he says are both "professional operations." But neither does he suggest that their findings should be taken as gospel independent of other surveys.
Take Gravis' Glenn poll, which has little in common with others conducted about the Colorado Senate race. "Through August and early September, the vast majority of polls showed Bennet's lead in double digits: eleven points, twelve points, thirteen points — and ten points now, according to CNN," Ciruli says. "So for Mr. Glenn to be ahead by two now would be a pretty extraordinary swing, particularly since he's done nothing to explain it. He's appeared in one debate that was relatively low-key" — a Club 20-sponsored gathering staged in Grand Junction — "and I don't believe he has any advertising up of significance. So either Gravis has an unusual methodology or it's probably an outlier."
Granted, there's one other possibility: Trump is doing so much better in Colorado than before that his popularity is driving additional support to Glenn, one of the few Republican candidates in these parts to openly embrace The Donald. (Glenn had a prominent speaking role at the Republican National Convention, for instance.) And Ciruli doesn't reject this prospect out of hand.
"There's a lot of volatility in the presidential race," Ciruli notes. "I was the person who, in late June, announced that Colorado was no longer a battleground state. We're not a blue-collar state in the way many are in the old rust belt, and I think we have the highest level of four-year college graduates in the country; we just passed Massachusetts. So there's the sense that we shifted a little to the blue — and after that, most national prognosticators said Colorado was either leaning Democratic or was likely Democratic. But right around Labor Day, things began to shift, and now, Colorado is back in the swing of things."
In Ciruli's view, this shift has everything to do with what he sees as "a bad three weeks for Hillary Clinton" — a period marked by her stumble at a September 11 event that preceded an admission that she was suffering from pneumonia, plus the steady drip-drip-drip of criticism related to the long-running scandal over her use of a private, Denver-based e-mail server and questions related to the Clinton Foundation charity.
These developments have exacerbated what Ciruli calls "indicators that Hillary's people may not turn out on election day and real questions about what's going to happen with third-party candidates." Indeed, both the Gravis and CNN/ORC polls show Libertarian presidential candidate Gary Johnson and Green Party hopeful Dr. Jill Stein cumulatively attracting support of twelve points or more.
Further complicating matters are the increasing number of polls being conducted in Colorado since the contest tightened, as well as a wider variety of approaches to polling. "There are a lot of new techniques," Ciruli says, "including a lot of panel surveys," defined by Pew Research as "a sample of respondents who have agreed to take part in multiple surveys over time."
This approach is less expensive that old-fashioned telephone polls, which have been complicated by the increasing use of cell phones, and plenty of major news operations are giving them a try, Ciruli allows: "The Los Angeles Times has hooked up with USC and is doing a panel survey just about every day — and they've looked overwhelmingly good for Trump for months. The New York Times has done some panel surveys, and whenever you see anything by SurveyMonkey, that's a panel survey."
Problem is, "panel surveys don't have much of a track record," Ciruli says. "And because panels attract people who are participating on a consistent basis, you worry that they may be different from the average voter you're trying to track down on the phone."
With this in mind, Ciruli feels voters who want to get the best sense of where competitions actually stand check out poll aggregators of the sort offered by RealClearPolitics. For example, the site's Glenn-Bennet page averages results from five September polls, including Gravis and CNN/ORC, to determine that Bennet is solidly in the lead, but not by as much as a month ago: 7.4 percent.
When it comes to the year's biggest race, Ciruli says it's too soon to tell whether the September 26 debate, which Clinton is widely viewed to have won, will have much of an impact on the overall numbers. He thinks it may take until the weekend to know if "September 26 was Donald Trump's worst day of the campaign so far, just like September 11 has been the worst for Hillary Clinton." In the meantime, the headline on Ciruli's recent presidential-election-related blog post puts the present circumstances succinctly: "Super Bowl of Debates and Tie Game."
Another Ciruli post, dated September 27, states that "Late September Polls Keep Bennet With Solid Lead" — and in it, he doesn't factor in the Gravis survey at all, even though the gig might have been his. He reveals that "Breitbart sent out bids to get a polling operation, and they sent me one. But I really like to stay neutral."
This quality is getting rarer and rarer these days. Here are the Gravis and CNN/ORC polls.