By Jim Sweeney

     The vexing aspects of using and interpreting numbers in polls, scientific research, statistics, economics and other fields is a topic addressed by John Allen Paulos in several books
and a monthly column, "Who's Counting?", for ABC News' web site.
     Paulos looks at how math and other scientific disciplines play a role in everyday life. One of his related  themes is how the public and the press often misunderstand the  sciences. 
     Before looking more at Paulos' work, consider these recent stories:
     1) A recent Gallup poll supposedly measured opinion in Islamic countries. It got lots of coverage--and lots of criticism. The National Council on Public Polls  denounced the
coverage of the poll's results by CNN and USA Today. A March 22 Washington Post article says that Gallup was the source of the "sensational characterizations" that were the  target of NCPP's criticism.
     2) The March/April Columbia Journalism Review discusses Bjorn Lomborg's book "The Skeptical  Environmentalist." Lomborg claims most environmentalists are alarmists and 
extremists. The book has gotten lots of positive coverage in this country. But the CJR article says Lomborg's critics, who  include many scientists, fault his cost-benefit analyses and extrapolations of data. They believe he selectively chooses studies that support his positions. 
     3) Several courts, including the Supreme  Court, are considering challenges to the 2000 Census. Several state and local governments argue that sampling techniques 
used by the Census Bureau miscounted their populations.  Possible consequences of errors in the Census include fewer seats in  the House of Representatives and less federal aid.
     4) A study done for the state of New Jersey  said that black drivers speed more than other drivers, which  might explain why state police stop black drivers more often. Those  who claim the police engage in racial profiling questioned the  study's methodology.
     Paulos is a professor of mathematics at Temple University. He also is an adjunct professor of journalism at  Columbia University, where he teaches a course on "numbers  in the news."
     The book that will probably most interest  journalists is "A Mathematician Reads the Newspaper" (Basic Books,  1995). He uses news stories as teaching aids to introduce 
mathematical, statistical and other concepts to the reader. He  often comments on how stories are misinterpreted. He clearly  feels readers and reporters should be more skeptical about "facts."
     While Paulos is a mathematician, he has wide-ranging interests and looks at broader issues of science  in journalism. Anyone who has ever read the magazine The  Skeptical Inquirer will
have a good idea of his approach to this topic. 
     In the book, Paulos cites a Spy magazine  story that found that the number of Americans infected with HIV  was given as one million in many news stories--from 1985 to 1993.  The number could not have remained static for that length of time.
     Paulos says that, for many reasons, it would be difficult to determine that figure with any great accuracy.  The implication is that reporters and editors should have asked more 
carefully where the number came from and how it was determined. 
     The book was written long before Enron's  collapse, but Paulos' skepticism about the ruminations of stock  analysts, whether they are columnists or sources for  reporters, could have
been written yesterday. He goes to great lengths  to explain why, despite what some analyst may say, the daily  changes in a stock or the stock market overall have "an extremely 
large element of chance involved."
     Paulos expresses lots of contempt for  pseudoscience stories, especially those that got lots of play because  they were pushed by people with seemingly legitimate credentials.  Most of the UFO abduction books that have been published (and  there are a great many of them) have gotten no attention from the  major media. But psychiatrist John Mack got lots of mainstream  media attention for his book on UFO abductees (possibly because he is a medical professional who teaches at Harvard).
     Paulos feels that many major media outlets,  including The New York Times and Washington Post, weren't very  skeptical about Mack's book. However, some reporters were inspired to look further. Paulos notes that a Time magazine  reporter tracked down one of Mack's "abductees" and found she freely  admitted to making up her abduction tale. 
     How stories are played also affects readers' perceptions. It's often noted that plane crashes get lots of  coverage, but far more people die every year in this country in car crashes. Paulos cites a study by a statistician of how often  different modes of death made the front page of the New York Times in a two-year period. Deaths from cancer got .02 stories per  1,000 annual deaths. Plane crashes got 138 stories per 1,000  annual deaths. (Keep in mind this was before the World Trade  Center attacks.)
     Paulos is also very mistrustful of numbers  bandied about by advocates of one side or another in a dispute.  Trolling news stories, he found that immigrants "either receive  $44 billion
more in government benefits than they pay in  taxes or else pay $25 billion more in taxes than they get in benefits, depending on what you read."
     Other broad ranges he has found include the  percentage of men who are homosexual (1% to 20%) and the number  of homeless people in America (200,000 to 7 million). 
     His conclusion on this matter: "Benchmark figures, operational definitions and simple arithmetic  should be a part of every major story or should at least appear  frequently in the ongoing coverage." 
     Paulos' web site  (www.math.temple.edu/~paulos) offers summaries of his various books, a biography and  other background on him, and links to the current and archived 
columns for ABC News.

      Jim Sweeney is the chief copy editor for Government Computer News in Washington, D.C. Prior to that  he was chief of the proof desk at U.S. News & World Report.  Before that, he was a reporter on various technical publications.