It's probably not party line for me to say this but the issue of SurveyUSA's age weighting on its Firedoglake polls is just kind of a distraction from the poll results.
To be clear I don't think turnout in the 18-34 group this year will be nearly as low as those polls have it- but I also don't think it has that significant an effect on the poll results. In his weekly column today Mark Blumenthal helpfully figured out what the actual breakdown by age was in 2006 in those districts. For Baron Hill's it was 16% 18 to 34 as opposed to 5% in the poll, for Steve Driehaus' it was 17% as opposed to 3%, and for Tim Bishop's it was 10% as opposed to 1%.
Let's say for the sake of argument that all three of those Democrats are winning 60-40 with voters under the age of 35.
In Baron Hill's district if you make voters 18-34 16% of the sample and give him 60% of the vote with them, while reducing the representation of seniors in the poll from 32% to 21%, you get Hill trailing Mike Sodrel by 4 points instead of 8 points.
In Steve Driehaus' district if you make voters 18-34 17% of the sample and give him 60% of the vote with them, while reducing the representation of seniors in the poll from 41% to 27%, you get Driehaus trailing Steve Chabot by 17 points instead of 11 points.
In Tim Bishop's district if you make voters 18-34 10% of the sample and give him 60% of the vote with them, while reducing the representation of seniors in the poll from 41% to 32%, you get Bishop leading Randy Altschuler by 3 points instead of 2 points.
The takeaway from the SurveyUSA polls in these districts based on the numbers as they were released was that Steve Driehaus and Baron Hill were in big trouble and Tim Bishop was in the most trouble he's been in in quite a long time. Reweight the numbers for age and while the numbers improve for the Democrats, I don't think the overall conclusions do. And I'm not surprised, because I rarely find when we weight for age that it has an impact of more than 2-3 points on the topline numbers. I may see a younger electorate than SurveyUSA does, but I don't think its impact on those poll results is that great.
Subscribe to:
Post Comments (Atom)
3 comments:
I think this should be not a surprise, and I think this is not sign of massive loses for dems.
They are 9 republican challengers with the level of former congressman or statewide officer (and someone more fighting for open seats). Pro republican pollsters show negative results for 3 of this 9 (IN-09, MI-07 and OH-01), while other pollsters show positive results for 2 (ND-AL and SD-AL). And I think PA-08 (D+2) and MA-10 (D+5) seems too leaning dem. Surely pro republican pollsters will not lose their time for show bad polls from ID-01 and MN-02, but will be difficult here republicans win more than 5 seats.
They are too 7 D open seats in R+ districts after Murtha's dead. Not more.
And of course they are more seats in risk for dems, but still only MD-01 shows negative polls, while other targeted seats like SC-05 or NC-08 give positive polls for dems.
PPP find not still a district where democrats lose. A bad poll from VA-05 would not be a sign of high loses for democrats.
Until now we have not sign of Democratic Party can lose more than 15 seats after the elections.
If someone wish tell dems will lose more, he should add a list with the districts without forget the districts what republicans can lose (LA-02, DE-AL, IL-10 and maybe someone more).
We will go you one better and list 60 seats that the Democrats should lose for a net loss of 58 seats since they probably will win in DE and LA-2. The list is too long for here, but you can find it rank ordered by the likelihood of the Republicans capturing the seat at: http://www.marstonchronicles.info/index.php?option=com_content&task=view&id=296&Itemid=1130 or by state at: http://www.marstonchronicles.info/index.php?option=com_content&task=view&id=308&Itemid=1144.
Win 58 house seats is in the same level than win CA-Gov, CA-Sen and likely HI-Gov.
Really optimistic republicans.
They think all democratic candidates are Coakley.
Post a Comment