Campaign Myth 2: Thirty Percent Will Flip the South
Earlier this week, the Associated Press released an analysis suggesting that if Barack Obama lives up to his pledge to boost African American turnout by 30 percent, he would score big wins across the south. But other data suggest Obama faces even longer odds in his quest to pick up electoral votes from "red states."
The AP used the average GOP vote margins from the past four presidential elections and data from the Census Bureau's Current Population Survey to show that if black turnout jumped 30 percent (with no corresponding change in the number of white voters), Obama could flip a swath of southern states, including Georgia, Arkansas, Louisiana, Tennessee, Virginia and Florida. (All but Florida were easy GOP wins in 2000 and 2004.)
It turns out that assumptions about baseline turnout are crucial.
Bill Clinton, governor of a rural southern state, won or was competitive in all of these states in 1992 and 1996, aided by a popular third party candidate in Ross Perot, so including those elections in the equation for the baseline GOP vote sets a lower hurdle for Obama than looking only at the 2000 and 2004 elections. Also, using the CPS data - based on a post-election phone survey - may overstate the numbers of African Americans who voted in those contests, therefore exaggerating the impact of a 30 percent increase in black turnout.
An alternative approach yields very different results.
Using the 2004 election as a baseline - this is the election against which final analysis will gauge Obama's ability to "turn" states "blue" - and exit poll estimates of the proportion and preferences of black voters, suggests black turnout would have to spike considerably higher - 50 percent or more - for Obama to win many of these states.
Taking Georgia as an example: George W. Bush beat John Kerry by 17 points in 2004, a massive margin, and better than his 12-point victory in 2000. Average GOP advantage: 425,796 votes. But add in 1996 (when Bob Dole beat Clinton by a single point) and 1992 (a narrow Clinton win), and the average drops to 216,218 votes, a much lower threshold. Using the CPS data further confounds the issue. The 2000 CPS estimate for black turnout in Georgia exceeds the total number of African American registered voters in the Georgia Secretary of State's database by more than 27,000.
Substituting the 2000-2004 average for the 1992-2004 average and using estimates of black voter turnout from the state government, shows that black turnout would have to go up by 81 percent to put Obama over the top; again assuming all else remained the same. Compared with 2004 alone, black turnout would have to about double (increase 96 percent) to give Obama the state's 15 Electoral College votes.
(Another difference between this analysis and the AP's is that the new black voters were allocated according to vote preferences from 2004; the AP assumed 100 percent would vote for Obama. Raising the percentage of black voters supporting Obama would marginally change the new estimates, but not much; 88 percent of black voters opted for Kerry in 2004.)
The comments to this entry are closed.