Differing Senate poll results may be related to different turnout predictions
This weekend, the Post-Dispatch released our latest and final poll before the Nov. 7 election. The Maryland-based firm of Research 2000 polled 800 likely voters last Monday through Wednesday (a few early Thursday.)
It shows U.S. Sen. Jim Talent, R-Mo., and his Democratic rival, state Auditor Claire McCaskill, in a tie — 47 percent each, while 2 percent of those polled supported Libertarian Frank Gilmour and 4 percent remained undecided.
In the state auditor’s race, Democrat Susan Montee had a narrow lead over Republican Sandra Thomas – 45-40.
And in the closely watched battle over Amendment 2, the stem-cell initiative, support had dropped to 51 percent, with 35 percent opposed and the rest undecided.
The poll’s margin of error was 3.5 percentage points, which means that that each number could be that much higher or lower.
The poll’s results differ somewhat from the Mason-Dixon poll released a week ago, which showed McCaskill with a 3-point edge, and a new SurveyUSA poll that showed Talent with a 3-point edge.
The difference can be explained in one word: demographics.
The Mason-Dixon poll’s partisan breakdown of the 625 poll participants is 36 percent Democrat, 35 percent Republican, 29 percent Independent. That poll was conducted Oct. 17-19.
The Survey USA poll, conducted Oct. 21-23, was of 630 likely voters. The partisan breakdown: 38 percent Republican, 35 percent Democrat and 27 percent independent.
For the Post-Dispatch, Research 2000’s 800-person sample was 37 percent Republican, 37 percent Democrat and 26 percent independent.
All of those demographics are based on the pollsters’ best guess of which side is more enthusiastic, and what the partisan percentage of the final vote will be.
The Nov. 7 results will, no doubt, determine which pollster guessed right.
(No Ratings Yet)
Loading ...
I was polled by phone a few days ago. Couldn’t recall which “organization” it was. I’d been called at least two or three times before in the last few weeks by people looking to survey me on various topics. This last pollster was the first person I agreed to answer. Reason? He gave his name and who he was with. The other people would immediately barrage me with questions, not even telling me their name or who they were with. I cut them off and hang up on them. This guy sounded like he wasn’t from Missouri because he had no knowledge of Amendment 2, but he was genial and listened, even asking me to expound on various positions, especially considering that some of his questions and answers didn’t fit my positions.
I think this is only the 2nd time I’ve given a phone survey during politicking season.
Still don’t know if I’d completely trust any polls though. How are these samples obtained? How truly random are these calls and what methods do the pollsters use to compile the surveys.