One of the happy results of Tuesday's Colorado caucus was that it reminds us of the difference between polls and election results. Denver Post polls conducted Jan. 21-23 showed the Democratic race in a statistical tie, while Mitt Romney had a 19-point lead among Republicans. Yet Obama ended up winning by 35 points, while Romney won by 41 points.
The American Research Group conducted three polls in Colorado last year, in March, July and September. ARG reported Hillary Clinton with a stable 11- to 17-point lead. The Republican field was more variable, with Giuliani always at or near the top, being challenged first by McCain and then by Thompson.
Mitt Romney, who ended up winning Colorado with 60 percent, never broke into double digits in any of the 2007 ARG polls.
The miserable predictive power of the Colorado polls could be seen as reinforcing the argument of my counterpart Jason Salzman, who argues that newspapers shouldn't spend money on polls. Indeed, he argues that newspapers should not even report on polls conducted long before the election.
But early polls - if conducted rigorously - can at least provide information about where a race stands at a particular moment in time. And sometimes the early polls do provide insight into how the election will turn out. In the 2006 gubernatorial race, the early polls showed Bob Beauprez trailing Bill Ritter, and Beauprez was never able to close the gap.
Mickey Kaus, a columnist for Slate.com, also argues that polling is more difficult these days because voters are dividing into two camps. On the one hand, thanks to the Internet, voters who are highly engaged in public issues have far more information than ever before. As an avid follower of politics ever since the 1968 presidential election, I can assure you that today, there is literally a hundred times more political information available, for free, than in the days when no one except political professionals got their political news from more than a couple local papers, three national television networks, and perhaps a magazine or two.
On the other hand, argues Kaus, a very large segment of the voting public is now more disengaged from the news than it was in the past. That segment doesn't make up its minds until the last 72 hours before an election. Pollsters have a lot of trouble gauging this capricious and large group of voters.
Pollsters also get confused when people behave differently than in the past. "Turnout models" try to pick which respondents are really going to vote, or to attend a caucus. In a normal year, a precinct caucus is lucky to have a few dozen people. But on Tuesday, 191 people showed up at my caucus.
So if you like politics, enjoy media stories about polls the same way that baseball fans enjoy a spring training game. They know that the teams that come in first in the Grapefruit League or the Cactus League may finish last during the regular season.
And just as the papers assign reporters to cover the Rockies' exhibition games in Arizona, preseason political polls are worthy of newspaper attention. Discerning readers just need to remember that the only poll that counts is the one on Election Day.
Kudos to the Post's Politics West Web site for its Super-Duper Tuesday coverage. Politics West provided a U.S. map by which readers could see the Democratic and Republican votes in each state.
Especially useful were the delegate counts from each state. The delegate counts were often incomplete, but the Post outpaced many of its national competitors by providing timely delegate data. And delegates are, after all, what winning the nomination is really all about.
Unfortunately, the Politics West map provided no data about what percentage of the vote had been counted in a given state. On this score (as well as visual appeal), the Rocky's political map that evening was superior.
One way that almost all the Colorado media fell short in their civic duty in the past few weeks was failing to explain the caucus process itself.
Even in years when few people caucus, a thorough explanation of how caucuses work should appear at least once in every one of the state's medium and large daily newspapers. In a year when caucus turnouts set records, the media's failure to inform was especially unfortunate.
It would have been better for the process of self-government if many more caucus participants had known in advance that the caucuses would be choosing delegates to a county convention (for president) and choosing separate delegates to a county assembly (for other offices), and that at a caucus, any person could offer himself as a delegate to the convention, the assembly, or both.
A letter to the editor of the Colorado Statesman, the weekly political newspaper, provided some of the information which should have been provided in almost every serious daily newspaper.