With Obama ahead in the polls for the last week, and the election looming, the mainstream media is treating us to more chin-scratching about the so-called Bradley Effect, or the supposed overstatement in polls by white voters of their willingness to vote for black political candidates. (Never mind its absence from the recent Harold Ford Jr. Senate race, and the presence of a greater reverse Bradley effect in Democratic primaries, as Obama well overperformed polls in an arc of states from Louisiana through Virginia.) The real story of this election is not Bradley Effect; it is the Obama Effect, which will later be understood as Obama’s ability to move certain states enormous distances from where they stood four years ago – for example, putting North Dakota, a 27 point loss, nearly within reach, and moving three different types of states significant, even huge distances from Republican Presidential voting toward Democratic Presidential voting. This post concerns all three types where there is a big Obama Effect. This is how Obama is remaking the electoral map.
The three clusters of states in which Obama moves the electorate greatly from 2004 are: (1) almost entirely caucasian states of the West; (2) Midwestern states that start with I; and (3) the a crescent of seven states within the South that have large African-American populations. In all three, it is not simply the case that Obama is performing much better than Kerry. After all, Obama is running significantly better than Kerry right now, so the amount by which Obama improves over Kerry is the baseline for seeking any Obama Effect. The degree of the Obama Effect is measured by the additional amount over that improvement from 2004 to 2008 – the further amount above trend – Obama performs. For purposes of modeling this, I use for 2008 projections Nate Silver’s model at fivethirtyeight.com. According to Nate’s model, Obama currently would win by 2.6% of the popular vote, a 5.0% improvement over John Kerry’s 2.4% loss. Nate’s 2008 projections of state by state results are consistent when aggregated with his 2008 national prediction. So the where’s Waldo of this post is to identify and analyze those instances where Obama deviates strikingly from the anticipated 5% pickup.
Group One: Overwhelmingly to Heavily Caucasian States of the West (Alaska, Colorado, Idaho, Montana, Utah, Wyoming)
There is a pronounced Obama Effect in most of Coors America. Start with Sarah Palin’s Alaska: John Kerry lost by 25 points. Even post-Palin, it models to a 15.7% loss, or an Obama Effect of 4.3%. But remember, before Palin, one poll showed Obama ahead, and the average deficit was clearly under 10 points – an OE of ten or more. Montana, a 20 point Kerry loss, models to a 3.3% loss currently, an OE of 11.7%. Colorado swings from a 5 point loss to a 3.6% victory prediction, which would be an 3.6% OE. Utah moves from a 46 point loss (yes, that happened) to a 30 point loss, for an OE of 11%. Idaho moves from 39 to 28 down, for an OE of 6%. Wyoming improves from a 40 to a 25 point loss, for an OE of 10%. North Dakota, though out of reach, moved from a crushing 27 point loss to a modeled loss of 8.3%, an OE of 13.7%.
This group could swing the election if Obama picks up the 9 electoral votes of Colorado. Montana’s 3 are within reach and could help. And Alaska would have been in play but for the Palin choice. Why the Obama Effect here? For one thing, the western populist, independent streak that turned Colorado purple, spawned Tester and Schweitzer in Montana, and Berkowitz and Begich in Alaska. For another, and most simply, Obama has committed resources to these states, scarce visiting time, but also field organizing time, which Democrats contemptuous of flyover country failed to do in ’00 and ’04. The Howard Dean 50 State strategy is paying off, and Obama is building, at least in these areas, a more national Democratic Party.
Group Two: The Midwestern States That Start With I.
Indiana, Iowa, Illinois. No, it’s not a basketball tournament with lousy mascots. It’s three very different, though contiguous, states in which Obama has spent a ton of time and resources. Indiana is the shocker: from 2004’s 21 point butt-kicking, the state models +2.4% for McCain, for an OE of a whopping 13.6%. This is get out the vote, this is McCain refusing to put his more limited resources in the state (ceding an entire cycle of voter registration to the Democrats), this is proximity to Chicago in Lake County. Flipping a state Kerry lost by 21 would be an epic achievement. Probably won’t happen, but it might. Amazing. Iowa is another triumph of retail politicking, and it has moved from a one point Bush win to a modeled 9.7% triumph, putting it entirely out of play: OE of 5.7%. Illinois shows a mild OE, moving from an 11 point Kerry win to a modeled Obama win of 16.8%, OE of 1.8%. The I’s have it, the it being a big case of Obama Effect, the big lesson here being retail politics.
Group Three: The Southern Arc (Louisiana, Mississippi, Alabama, Georgia, the Carolinas, Virginia)
Virginia is the most important and obvious case of the Obama Effect in the South. From an 8 point Kerry loss, fivethirtyeight models it as a 1 point Obama win – a nine-point move, for an OE of 4%. Virginia was part of a highly compressed primary with a few days of barnstorming rather than the exhausting ground game Obama put into Iowa and, to a lesser extent, Indiana. Here the OE comes from some Republican crossovers, and from overwhelming support from the state’s African American voters. Likewise, North Carolina moves from a 12 point loss in 2004 to a modeled 4 point loss, for an OE of 3%. However, two current polls show North Carolina tied, suggesting a higher OE. Obama’s overperformance in the Southern Arc states in the primaries suggests that polling is understating African-American affinity for Obama, making these states potential tipping points that may reveal themselves as having massive OEs of 7-10%. We will soon know. The rest of these states, though not really in play, feature at least modest OEs (Mississippi falls from a 20 to a 12.4% loss, OE of 2.6%, Georgia falls from 17 to 8 point loss, OE of 4%, Alabama 26 to 20, for an OE of 1%, South Carolina 17 to 9.4, for an OE of 2.6%.)
Conclusion: Haters of David Axelrod’s new electoral math may not like it, but Obama is on track to win red Colorado (9 EV) and Virginia (13 EV), is close to winning super-red North Carolina (15 EV), Montana (3 EV), and Indiana (11 EV), and flipping back barely-red Iowa (7 EV) and New Mexico (5 EV; OE 3.5%), and by margins Al Gore and John Kerry would envy. If the election results are equal to or better than those currently modeled (and of note, the fivethirtyeight.com model presupposes a smaller popular vote win than is indicated in the current average of tracking polls), we’ll be talking about an Obama Effect after this election, not a Bradley Effect.
