Publications

When a report prepared by former Senator George J. Mitchell indicated that Roger Clemens and more than 80 other Major League Baseball players used illegal, performance-enhancing drugs, the former Cy Young award-winning pitcher went on the offensive to clear his name. Added to Clemens' testimony before Capitol Hill lawmakers was a voluminous report prepared by a marketing agency that relied on statistics to make the case for Clemens' innocence.

But an article in the February 10 edition of the New York Times written by four Wharton faculty makes another case: The validity of a statistical analysis is only as good as its individual components. It's a distinction that is likely to gain in importance as organizations and individuals try to make sense of an increasingly large and complex barrage of information.

"Today, consumers of information are drowning in data," says Justin Wolfers, Wharton professor of business and public policy. "Terabytes of data are being generated from the constant measurement of businesses, workers, government and other activity, and there are many ways to draw inferences from the raw data. Unfortunately, many of them lead in the wrong direction."

Read more.