Pollster bats 10 for 10 in the Reuters/Zogby telephone poll package of competitive Senate races; Interactive polling model shines
Zogby via BBSNews 2006-11-09 -- Pollster John Zogby continued his record of political polling success Tuesday, accurately identifying the winners of 84% of the 73 congressional and gubernatorial races the firm tracked in the weeks leading up to Tuesday's elections.
Zogby International's telephone polling of crucial U.S. Senate contests was particularly stunning – batting 100% in identifying the winners of all 10 of those toss–up races, including the victories of Democrat Claire McCaskill in Missouri, who won by 2% (Zogby had her up by 3% a week before the vote) and Democrat James Webb, who defeated Virginia Sen. George Allen by 1% – the exact margin identified by Zogby.
Zogby also correctly showed Democrat John Tester holding onto a slim lead over Republican Sen. Conrad Burns in Montana. Zogby had Tester up by 2%; the unofficial count in Montana has the challenger up by just under 1% over Burns.
In other Senate races, Zogby nailed the Connecticut contest between incumbent Joe Lieberman, Democratic challenger Ned Lamont, and Republican challenger Alan Schlesinger. The pollster had Lieberman ahead by 12% – he ended up winning by 10%. Zogby polling in the other races produced similar results.
As the election drew near, Pollster Zogby was quoted widely saying that he believed the Democrats in the U.S. House of Representatives would pick up between 25 and 30 seats, and that Democrats would almost certainly pick up three or four seats in the Senate, and perhaps get as high as six. As in earlier elections, Zogby's estimates turned out to be right on the mark.
Another bright spot for Zogby came from its online polling division, Zogby Interactive, which correctly identified the winners in 18 of 19 Senate races. And the only race that Zogby did not pick correctly – the McCaskill/Talent race in Missouri – was still well within the margin of error. Zogby Interactive had Talent winning re–election by a single percentage point, instead of losing by 2%.
This separate polling methodology, under research and development at Zogby since 1998, showed particular value in mapping close contests – it correctly identified the winners in four of the five races that were won by single digits (the McCaskill race being the exception). Taking into consideration the last–minute twists and turns that took place in that race after the Zogby Interactive poll left the field, its result was remarkably precise.
There is still work to be done in perfecting the interactive polling model – for instance, the interactive survey had Minnesota Democrat Amy Klobuchar leading in the Senate race there by 8%, while she coasted to an easy 20% victory. And in New York, incumbent Democratic Sen. Hillary Clinton led by 24% in the poll but actually won with a 36% edge. Some of that can be attributed to the fact that, for a variety of reasons, the larger the advantage one candidate holds over another, the more difficult it is to pinpoint what the exact margin between them will be on Election Day.
Zogby Interactive also correctly identified 16 of 21 winners in races for governor around the country, correctly choosing the winner in six of the nine closest races that ended with single-digit victories. For instance, in Illinois, Zogby had incumbent Democrat Gov. Rod Blagojevich up by 7% heading down the stretch; he ended up winning by 9%. In Maryland, Baltimore Mayor Martin O'Malley was leading in the Zogby Interactive poll by 5%; he won by 7%. Even in the blow-out race for Massachusetts governor, Deval Patrick led in the interactive poll by 25% and ended up winning by 21%.
Again, in the governor's races, there were misses by the interactive model. In Arkansas, the last poll before Election Day has Republican Asa Hutchinson up by 3%, but he lost to Democrat Mike Beebe by 14%. And in Colorado, where gubernatorial candidates Bob Beauprez, the Republican, and Democrat Bill Ritter were tied in the last interactive survey. When votes were counted, Ritter won by 15%.
Pollster John Zogby: "Our analysis of our interactive polling data over time, compared to the public results from other firms, shows that we were ahead of the curve in identifying movements in several key races. We were first to identify weakness in the campaign of Virginia Sen. George Allen, for instance. Then, just before the election, everyone's polling showed that race had turned from a cake-walk to a toss-up. In the Michigan Senate race, we were on the leading edge seeing that race tightening, and then widening again as Stabenow regained her footing. In Ohio, we were early in noting the Republican Mike DeWine was facing trouble with his political base. In the end, he could not regain that support. And in Arizona, we always had the Kyl-Pederson race much closer than other pollsters, who gave Kyl big leads. From Oregon to Florida, we were in the front seat as this election unfolded.
"We think the strength of this interactive polling model stems in part from the fact that our pool of respondents tends to feature people who are more intensely partisan - which means they are somewhat more motivated than the average American to pay attention to politics and to vote. This means our results are more likely to reflect actual election outcomes.
"While we had some great successes with both our telephone surveys and our interactive polling this year, to be honest, it does appear that the interactive polling has fallen short in some states, notably Colorado and Arkansas, as well as overstating the count for minor parties in several states. Nonetheless, it is a solid achievement and a worthy experiment that we will continue to develop and perfect. This is truly the next wave for the polling business."
Overall, Zogby correctly identified 85% of the winners using his interactive polling methodology.
Comments (0)