POLLSTERS AGREE, MAYOR'S RACE IS CLOSE : INTERPRETING RESULTS CAN BE TRICKY, AS LATEST SURVEY SHOWS Series: POST-DISPATCH POLL ST. LOUIS THE RACE FOR THE MAYOR OF ST. LOUIS Second of Three in A Series
Jo Mannies And Harry Levins Of The Post-Dispatch Tim O'Neil Of The Post-Dispatch Contributed Information For This Story., St Louis Post-Dispatch (MO)
Pollsters and politicians agree on one thing in the contest between Mayor Freeman Bosley Jr. and former Police Chief Clarence Harmon:
It's a close race.
How close depends on who's interpreting the polls. Take, for example, the Post-Dispatch's latest poll, published Monday. Terry Tomazic, professor of research methodology at St. Louis University, said, "If I looked at the numbers from this poll, I'd probably say that Harmon is leading at this point by a small margin - a very small margin." Esther Thorson, who supervised the poll, said Monday that she interpreted the Harmon-Bosley results a tad differently: "They are neck and neck," she said. Thorson is director of the Center for Advanced Social Research of the School of Journalism, University of Missouri at Columbia. The center conducted its poll of 412 city residents Jan. 23-31. Monday's story said "former Police Chief Clarence Harmon holds a lead of more than 5 percentage points over Mayor Freeman Bosley Jr." Maybe. Maybe not. The story failed to make clear that the poll's margin of error of 5 percentage points made many head-to-head scenarios possible, including: Harmon leading by as much as 16 percentage points. Bosley and Harmon tied. Bosley leading by as much as 4 percentage points. Such is the uncertain world of polls, particularly so-called "horse race" polls that pit candidates against each other. Post-Dispatch Editor Cole C. Campbell said, "I think the paper's presentation was accurate and misleading at the same time." Tomazic says that although the Post-Dispatch's poll was accurate in its math, the paper's words may have overstated the poll's numbers. The headline over the story read, "Harmon Ahead of Bosley." As an expert in polls, would Tomazic have used those words? "Not quite," he said. "I might have said, `Harmon Leads Bosley.' To me, `Ahead' connotes that he's winning - as opposed to leading at this point." Complicating everything is the big slab of undecided voters - 15 percent. "That's a large bloc," Tomazic said. "Even part of that bloc is enough to sway the election - especially when you figure in the 6 percent who declined to answer." Thorson agreed. But she emphasized that she stood by the poll's general findings, as reported, particularly those indicating that the candidates' supporters are split along racial lines. Of those polled, Monday's story said, "about seven of every 10 whites favored Harmon while about seven of every 10 blacks supported Bosley. Just one of every 10 whites planned to vote for Bosley, while about one in 10 blacks preferred Harmon." "That's statistically significant," Thorson said. "The very least you can say is that blacks support Bosley much more than whites support Bosley, and vice versa for Harmon." A poll's margin of error depends largely on the sample size: the more people polled, the smaller the margin of error. A margin of error of 5 percentage points requires a polling sample of at least 400 people. The margin of error is slightly higher in the portions of the Post-Dispatch poll that examine how white and black poll participants view the contest. The margin of error is 7 percentage points for each group, Thorson said. …