A close reading of longitudinal Colorado data holds out the possibility Romney might beat Obama by 5% in our state. Read on for details.
Nate Silver, the FiveThirtyEight blogger, recently reported on the “house lean” of various national pollsters. Of special note for Colorado political junkies (and Denver Post writers), Silver showed Public Policy Polling as leaning 3.1% to the Democratic side. Since PPP has done almost half of the Obama vs Romney polls in Colorado since January, 2011, this could offer a decidedly jaundiced view of Romney's chances in our state. But, having already offered a correction to PPP's sampling, let's move on.
The eleven Colorado polls out there offer a lesson in the importance of turnout efforts.
A scatterplot review shows that the size of a poll's sample advantage for one party or the other is 74% correlated to the advantage either Obama or Romney have. Put simply, if a poll has lots more Democrats than Republicans in its sample, Obama wins; when Republicans outnumber Democrats in the sample, Romney makes it a neck-and-neck horse race.
One poll is excluded: a PPP poll that showed a statistically outsized Obama advantage of 13%; the R-squared diminished substantially when it was included. Two polls are at -5% turnout, about what current active voter registration shows) and a tie (Purple Strategies and Rasmussen). One pollster did not respond to an email request for partisan sample information. The three polls at the upper right are all PPP (!).
The red diamonds are actual turnout by party and election results (for Obama in '08, for all statewide candidates averaged in '10; source: Secretary of State). Both appear to show some hallmarks of a “wave” election, favoring Democrats in 2008 and Republicans in 2010. Note that even in 2008, Republicans outnumbered Democrats at the polls. We should, based on this, be suspicious of polls whose samples include more Democrats than Republicans. Pollsters, it ain't gonna happen in 2012 either. (Post reporters should also decline to report polling results as though they'd arrived by angel from heaven; tell your readers about sampling problems!)
Republicans have, legitimately, worried about the party's “broken brand.” While Democrats still had the advantage nationally, Coloradans once again give Republicans the plurality lead. Part of that brand repair may be the result of our State House Republicans' legislative program that rolled back some of the Democrats' $5.9 Billion in tax hikes.
Gallup shows the advantage Democrats still had in 2010 with Independents who leaned more heavily to them than to Republicans. That may explain why Bennet and Hickenlooper won, although each enjoyed massive campaign cash advantages that allowed them to “buy” votes. If the Republican advantage of 2011 continues to this fall's election and we match Democrat and liberal special interest spending, Obama may have an uphill trudge.
With a 5% registration advantage and Independents leaning leaning only two percent more toward Democrats for the first time in two elections, this could be a very good year for Republican candidates. In 2008, Independents in Colorado leaned 11.4% more to the Democrats than toward the Republicans. McCain was, back then, 8.9% behind Obama, so he was about 2% BTE (better than expected). IF he gets both the 2011 Republican plus leaners advantage AND does 2% better than that, Romney could win Colorado by 5% over Obama.
ONE CAVEAT: Eric McGhee and John Sides appear to have demonstrated that “the partisan composition of voters depends on campaign activity” more than on job approval or the state of the economy. This means Colorado's results in November will depend on how well both sides perform in persuading swing voters and in turning out their committed supporters. Hard work matters.
Special thanks to Purple Strategies, Project New America and Marist for providing information not part of their press releases and websites.