How Accurate Was the Obama Delegate Prediction Spreadsheet?

On February 7, an internal Obama campaign spreadsheet leaked in the press. It contained the campaign’s predictions for all of the remaining primaries. Now that the primaries are over, we have the opportunity to judge the accuracy of Obama’s prognosticators, who, as everyone knows by now, showed remarkable prescience in their planning this campaign season.

Below are the spreadsheet’s popular vote and delegate predictions compared to actual results. The numbers show that the Obama campaign strategists were routinely too conservative: they underpredicted both the margins of their victories and their losses. They often anticipated a close to 50-50 split in a state that turned out to seriously favor one of the two candidates.

Of the states they predicted correctly, they underpredicted their margin of victory (aka were too pessimistic) in 16 states and underpredicted their margin of loss (aka were too optimistic) in six. They only overpredicted their margin of victory in two states, and never overpredicted a loss. In total, they got 24 of the 27 primaries after February 5th correct.

Of the ones the campaign got wrong, they were too hopeful in South Dakota and Indiana, where they predicted victories but suffered losses, and were too pessimistic in Maine, where they predicted a close loss but actually saw a substantial victory.

They nailed the delegate count exactly in five states, and were within one delegate in five more. They predicted their delegate count to within five delegates in 23 of the 27 primaries.

Washington’s numbers are complicated by the fact that the state held both a primary and a caucus.

Texas’s numbers are complicated by the fact that the state held both a primary and a caucus.

All results numbers, both popular vote percentages and delegate counts, come from the New York Times.