With all due respect to Floyd Ciruli, his sample was too old, too white, too Republican.
That's the assessment of Magellan Strategies CEO David Flaherty on why Ciruli's poll for the University of Denver's Crossley Center for Public Opinion Research found Trump and Clinton in a tie while Magellan Strategies' poll found Clinton ahead by six points.
Democrats were so eager to see some good news after others polls locally and nationally showed an uncomfortably close race that Magellan's site briefly crashed after the poll results were announced Thursday afternoon.
Flaherty is quick to say that turnout will determine the outcome on Tuesday, and Ciruli -- Flaherty reiterates again his respect for the prominent pollster -- could yet be proved right. Republicans have been slowly narrowing the Democrats' lead in early voting over the course of the week.
Here's a round-up of recent Colorado polls:
Note, these are all from voter surveys that occurred after FBI Director James Comey announced on Oct. 28 that the FBI was looking at emails found on Anthony Weiner's computer to see if they have any relevance to the bureau's investigation of Clinton's private email server.
- Ciruli's poll for DU found Clinton and Trump tied 39-39 when voters were polled on a four-way race that includes Libertarian Gary Johnson and Green Party candidate Jill Stein. In a two-way race, Clinton was ahead by just one point, 42 percent to Trump's 41 percent.
- Magellan Strategies found Clinton ahead of Trump by six points, 44 percent to his 38 percent, with support for third-party candidates declining and more of those voters moving to Clinton than to Trump.
- Emerson College, in a poll that found Clinton losing ground in other states, gave her a three point lead over Trump in Colorado, 44 percent to his 41 percent. A poll Emerson did in September had Trump ahead here by four points, 42 percent to her 38 percent, so this result, from voter surveys done between Oct. 28 and 31, essentially flips the state for her.
- Remington Research gave Clinton a one point lead, 45 percent to Trump's 44 percent.
So, none of these polls show Clinton losing to Trump, but several are within the margin of error with a not-small number of undecided voters.
Flaherty doesn't put much stock in either the Emerson or the Remington polls because they only use land lines. That gets you an older, whiter and more Republican-identified voter pool.
First aside: For whatever it's worth, FiveThirtyEight gives Magellan a C grade and Emerson a B, while not grading the University of Denver poll or the Remington poll.
Second aside: The Remington poll suggests 38 percent of Latino voters in Colorado and 27 percent of black voters will go for Trump, which would certainly be a surprising outcome.
Every poll uses certain methods to adjust the responses demographically because the group of people who take the survey aren't a perfect match the electorate, but you can't entirely get rid of a lean that comes from using only landlines or having over- or under-representation of certain groups.
DU used a mix of cell phones and landlines -- far more cell phones than Magellan's poll -- but the poll participants were older than those in Magellan's poll and also whiter.
"If we do have an older electorate, at the end of the day, Floyd's survey is going to be more accurate," Flaherty said.
Flaherty also expects Latino turnout to be higher than in 2012.
Magellan puts more emphasis on turnout from the previous election in weighting its poll results compared to other firms.
So who's going to win Colorado?
We'll have to wait and see. The New York Times gives Clinton an 85 percent chance of winning the state, most probably by a four-point margin, and FiveThirtyEight gives Clinton a 74.6 percent chance of winning, most likely by a five-point margin.
And that's good for Clinton because she needs Colorado as the polls get even tighter in other swing states.
FiveThirtyEight's Nate Silver highlights the weakening of Clinton's "firewall." Even just a few weeks ago, she seemed secure enough in enough swing states that there was little way for Trump to break through, but now the potential loss of New Hampshire is cause for concern.
"Couldn’t Clinton win Nevada to make up for the loss of New Hampshire? Or Florida? Or North Carolina? Well… of course she could. All those states remain highly competitive. The point, as we’ve said before, is just that Clinton’s so-called firewall is not very robust. If you’re only ahead in exactly enough states to win the Electoral College, and you’d lose if any one of them gets away, that’s less of a firewall and more of a rusting, chain-link fence."
FiveThirtyEight now gives Clinton a 66.2 percent chance of winning the election. The New York Times still has her chances at 85 percent.
Vox walks through the differences in their methodology and several other sites doing poll averages in this helpful post. Among them, FiveThirtyEight drops older polls out of its averages more quickly and uses national polls to weight data from states where there hasn't been recent polling. This means FiveThirtyEight's forecast is both more likely to pick up a key turning point in the race and more volatile in response to new developments that may or may not have a lasting impact on voter preferences.
