Les Jones

Politics > Winner of the Election Prediction

Days before the election, I began tabulating blogger's predictions for the presidential race. In the end I collected over 60 responses, from Atrios to Wizbang.

The final results are in, and the electoral vote tally is Bush 286, Kerry 252. Three bloggers predicted the results: Election Projection, Outside the Beltway, and VodkaPundit.

The tiebreaker is the date of the prediction. Outside the Beltway made its prediction on November 1. Stephen Green of VodkaPundit made his prediction, based on TradeSports numbers, on October 31. Scott Elliott of Election Projection, based on a composite poll analysis, made his projection no later than October 30.

UPDATE: I originally called it for Scott Elliott, but Outside the Beltway notes that Elliott later changed his prediction slightly three days after I recorded it. That throws a monkey wrench into the results. To avoid any more bake-offs, I'm calling it a three-way tie. All three blogs did a great job of covering the election. Congrats to Election Projection, Outside the Beltway, and VodkaPundit.

Checking the Pollsters on the Popular Vote

PoliPundit and RealClearPolitics are checking the pollsters for accuracy. RCP finds that Battleground/Tarrance, Pew Research, and CBS/NY Times polls were the most accurate, while Marist College, Battleground/Lake, and FOX/Opn Dyn were the least accurate.

In a November 2 article in The Hill, Kerry chief pollster Mark Mellman wrote that the Kerry camp's internal polls indicated Bush would win 51.6% of the popular vote. Bush did in fact win 51% of the popular vote.

Virginia Postrel notes an unusual, and unusually accurate prediction made by 7-11 convenience stores:

7-Eleven's coffee cup poll--coffee buyers could pick between Bush and Kerry cups--proved remarkably accurate: 51.08 percent for Bush, 48.92 percent for Kerry.

"Our popular vote was absolutely right on," Jim Keyes, the chain's head honcho, told the Dallas Morning News. "We sell a million cups of coffee every day, so our sample size was huge."