Statistics Really Can Predict Oscars, Says U of Oregon Statistics Professor.

EUGENE, Ore. -- Don't fill out your Oscar prediction ballots without first consulting Iain Pardoe's statistical analysis.

Pardoe, an assistant professor of decision sciences at the University of Oregon's Lundquist College of Business, has applied a statistical model to

predict Academy Award winners in the four major categories. Past Academy Awards results confirm that his analysis is 93 percent accurate in predicting winners of the Best Director category and 77 percent accurate in predicting Best Picture, Best Actor and Best Actress over the past 30 years.

Pardoe's statistical analysis predicts the following winners at the 78th annual Academy Awards, scheduled for March 5:

--Best Picture: "Brokeback Mountain" (91.1 percent probability)

--Directing: Ang Lee for "Brokeback Mountain" (90.5 percent probability)

--Actor in a Leading Role: Philip Seymour Hoffman in "Capote" (94.8 percent probability)

--Actress in a Leading Role: Reese Witherspoon in "Walk The Line" (75.9 percent probability)

The model's predictions are based on statistical analysis of influential factors associated with past Oscar winners, such as additional Oscar nominations in key categories, total number of Oscar nominations for the nominated movie, previous Oscar nominations and wins, Golden Globe Awards and Producers Guild, Directors Guild and Screen Actors Guild Awards.