This by itself is not yet proven reliable (it's never been used to predict a future event) but when combined with the data from the polls it is real cause for optimism. This article, excerpted here for background, was published in the New York Times on Spetember 17, 2002.
>
Research Brings a New Dimension to 'a Candidate's Voice'
>"A new study by researchers at Kent State University suggests that at least some of the way that people perceive social standing can be detected in the way their voices change when they are talking with people they see as more confident or higher on the social ladder. The changes are barely detectable, the researchers say, but those who are most inclined to make them typically lose at the ballot box."
(more after the cut, including 2004 analysis)
"The new paper, published in the current issue of Social Psychology Quarterly, analyzes the voices of presidential candidates in their debates since 1960. The researchers, Dr. Stanford W. Gregory Jr., a professor of sociology, and Dr. Timothy J. Gallagher, a colleague, found that, in every case, the politician whose voice was most steadfast, as measured by the researchers, won the most votes."
>"Predicting the past may not seem to be a stunning achievement. Seven of 10 patrons of any corner bar can "predict" the outcomes of the last three Super Bowls. But while the patrons are really just recalling known results, the researchers are using correlations with known facts to test a theory."
>"It has long been established that humans communicate not only by words, but by expressions, tics, posture and more. "These other sources of information - the sound of the voice, the gestures, the expression - they help form our impression, but without our being as aware of them as the words they speak," said Dr. Paul Ekman, a professor of psychology at the University of California at San Francisco."
>"The researchers focused on the tendency of people in conversation to alter their pitch, volume, pace and other characteristics of speech to emulate one another. This area of study, which has been going on for several decades and is known as communication accommodation theory, has previously revealed that people of lower social status unconsciously alter their nonverbal vocal patterns to sound more like people of higher status, while those of higher status don't budge much."
(snip)
>"In the new paper, the researchers took three recorded samples from presidential debates and created a score that described how much each subject's voice varied. The vocal data "predicted the popular vote outcomes in all eight elections," the researchers found. In the 2000 election, Al Gore's voice changed less than that of George W. Bush, and Mr. Gore won the popular vote; the Gregory analysis does not appear to be affected by Florida elections law or the Supreme Court."
I love when the media uses blatant election theft as humor to be gratuitously sprinkled throughout their stories.
Anyway, I emailed Dr. Gregory for this year's results.
Today he emailed me back.
Here it is.
>
Timothy:
>Thanks for your note. Yes, we have worked on this year's debates and have come up with the following:
Debate Acoustic Anal.
1 Bush .64
Kerry .84
2 Bush .91
Kerry .95
3 Bush .39
Kerry .85
>We have also worked on another statistic based on the relation between the acoustic analysis results and popular vote results for the last 8 elections and predict a Kerry win with a 4.35% spread in the popular vote.
>We will see if the anterograde value in any way matches the retrograde.
>Incidentally, we found that in the vice presidential debates Cheney had .85 and Edwards .84. In our 2002 analysis of Cheney vs. Lieberman, the former was further ahead than the result this year.
>Stan Gregory
Hell yeah.