Reviewing the Key Senate Races Posted by: McQ
on Monday, October 30, 2006
So how are those key Senate race numbers trending this week (with 8 days to the midterm election)? Let's look at the RCP averages for 10/30:
CONNECTICUT - Sen. Joseph Lieberman, running as an independent and leads in the race. Lieberman, a three-term Democratic incumbent, lost the party primary in August after Lamont attacked his support for the Iraq war.
So much for Netroots and single issue (Iraq) candidates. While there is a slight erosion in the trend, still a strong lead for Lieberman.
MONTANA - Democrat Jon Tester leads Republican Sen. Conrad Burns after Burns suffered a series of problems, from returning donations from associates of convicted lobbyist Jack Abramoff to comments seen as insensitive to some ethnic groups and to out-of-state firefighters.
Still leaning Democrat but tightening. Could be a +1 for Dems if Tester can hold on.
OHIO - Republican Sen. Mike DeWine has fallen well behind Democratic Rep. Sherrod Brown.
Brown continues to widen the gap. If Tester holds on in MT it could be Dems +2.
[RCP 10/10: Brown +3%, RCP: 10/20: Brown +10.6%, RCP 10/30: Brown +11.7%]
PENNSYLVANIA - Rick Santorum, the third-ranking Senate Republican, trails Democrat Bob Casey Jr., the son of a popular former governor by a fairly significant margin.
Casey continues to widen the gap over Santorum. Up over 4 points in 10 days. +3 for Dems.
TENNESSEE - In a race for the open seat of retiring Senate Republican Leader Bill Frist, Democratic Rep. Harold Ford Jr. and Republican Bob Corker, the former mayor of Chattanooga in what has been a very tight race.
What a weird couple of weeks in Tennessee. Ford may have done himself in as Corker now leads with a shift of 3 points (still very tight but leaning Republican).
[RCP 10/10: Ford +2%, RCP 10/20: Ford +0.8%, RCP 10/30: Corker +2.3%]
VIRGINIA - Excerpts published from a Webb novel which are, well, explicit and causing waves, politically. However that doesn't seen, at this moment, to have hurt Webb particularly.
Some erosion and almost a tie as Webb has gained some ground even though Allen still has a slight lead.
[RCP 10/10: Allen +5.2%, RCP: 10/20: Allen +4.7%, RCP 10/30: Allen +1.5%]
So Tester (D) is seeing his lead erode in MT as is Allen (R) in VA. Talent (R) has taken a slim lead over McCaskill in MO (D) as has Corker (R) over Ford (D) in TN.
But it still adds up, at the moment, to +4 for the Dems.
I'd remind you that the RCP numbers are average numbers from a number of polls. But I think they reflect trends pretty fairly. I'd also remind you that a heck of a lot can happen in 8 days (see TN and VA this past week).
I think most of these races are close enough that predictions beyond guessing that the polls (mostly within the margin of error) will hold are not much more than wishful thinking.
That said, these races, like most close races will come down to who gets out the vote. Republicans have been historically better at this, and their voters, even when there are fewer of them, are generally more motivated than the Dem’s.
Not many people are making the case that Republicans are motivated this time around.
I am reminded of an old axiom used frequently by my professors when cautioning against trusting scientific equations to give correct answers - garbage in, garbage out.
For example, SurveyUSA has Brown up 20%(+/-4.2%), Rasmussen has Brown up 12% and both LATimes/Bloomberg(+/-4%) and Mason-Dixon(+/-4%) are showing Brown up 8%. Someone’s polling is way wrong, and to include all in an average means the average (ie RCP’s #) is meaningless. That’s not to say Brown isn’t the likely winner, just that RCP’s numbers dont tell us much.
I agree - in theory. I’d suggest however, identifying one polling outfit that you find methodologically consistant to use for identifying trends. for example, in Ohio Rasmussen has Brown trending up, whereas Zogby Interactive has him trending down. What does averaging the two do to trend analysis?
I have no real confidence in any single poll.
And that’s my point - If one has no trust in any single survey, why should one trust an aggregate of all surveys?
I agree - in theory. I’d suggest however, identifying one polling outfit that you find methodologically consistant to use for identifying trends. for example, in Ohio Rasmussen has Brown trending up, whereas Zogby Interactive has him trending down. What does averaging the two do to trend analysis?
But I’m using an average of 5 polls every 10 days. The averages seem to be matching reported trends.
And that’s my point - If one has no trust in any single survey, why should one trust an aggregate of all surveys?
Because they tend to ameliorate the outliers through averaging. Whereas any single poll in any single race could be the outlier.
Because they tend to ameliorate the outliers through averaging. Whereas any single poll in any single race could be the outlier.
Listed on RCP for Ohio there are 14 different outfits conducting 43 surveys dating back to Oct 24, 2005. In the last RCP average are Rasmussen, SurveyUSA, LATimes, and Mason-Dixon. In the past year Rasmussen has conducted thirteen surveys, SurveyUSA five, LATimes ONE, and Mason-Dixon three. Rasmussen has Brown +11(10/25/06), +6(10/12/06), +7(10/03/06), +6(9/13/06), +3(8/22/06), Dewine +7(6/20/06), Brown +3(5/08/06), DeWine +2(4/19/06), DeWine +3(3/28/06), DeWine +9(2/16/06), DeWine +5(1/03/06), and DeWine +2(11/30/05). SurveyUSA has Brown +20(10/23/06), +14(10/09/06), +10(9/18/06), +8(8/5/06), and +9(6/10/06). Mason-Dixon has Brown +8(10/18/06), +2(9/24/06), DeWine +11(4/24/06). LATimes first poll, which RCP deems fit to incorporate shows Brown +8(10/20/06).
Zogby, NOT included in the latest RCP average has Brown +4(10/10/06), +4(9/19/06), +4(8/29/06), +8(8/15/06), and +12(6/13/06). That’s trending down, in case you didn’t notice.
For what it’s worth, Zogby Interactive has conducted the second most reported polls of Ohio.
Are they the outlier? Or is, perhaps, the LATimes/Bloomberg poll? Surely we can trust Time purchased RCP.