Normal and predictable variations in polling numbers were mistakenly presented by journalists as evidence that presidential candidate Bob Dole was dramatically closing the gap on President Bill Clinton before last week's election, according to political science professors Robert Y. Shapiro of Columbia University and Lawrence Jacobs of the University of Minnesota.
The study found that journalists seized on nearly any sign that the presidential race was tightening, even though the great majority of trial heat polls showed Clinton had a steady double-digit lead throughout the campaign. The study suggests that the overwhelming majority of news stories concerning the presidential polls ignored ample surveys of voters' attitudes toward substantive policy issues and instead opted to discuss "horse race" polls.
"Journalists generally used polls conducted by their own organization, and rarely weighed their polls against comparable surveys from competing media or other external sources," the authors report. "Excessive weight was put on a single poll generated from within a journalist's own organization, and audiences were denied the opportunity to compare results and reach accurate conclusions."
The problem is not the campaign polls themselves, say Shapiro and Jacobs, but journalists' coverage of them. "One solution would be to launch regular poll watches that would capitalize on the vast amount of polls to challenge questionable surveys or media interpretations," said Shapiro and Jacobs. "This would make journalists pause and rethink their treatment of election-year polls." The study was sponsored by the Annenberg Public Policy Center at the University of Pennsylvania. Professor Shapiro can be reached at Columbia's Political Science Department, New York 10027, 212-854-3944, firstname.lastname@example.org.