OK

With only 15 days left in the election, the chances of Republicans winning control of the Senate are now quite remote. If today was Election Day, Democrats would have a 99.3% chance of retaining the chamber.

Senate Competitive Campaigns Chart

* = 25-day simple average. All other campaigns feature only polls that will be included in the final Senate Snapshot averages--that is, polls with the majority of interviews taken on or after October 8th.

Seat Outcome Odds chart


Here is some more detail on possible outcomes:

56 Democrats: 0.1%
55 Democrats: 1.4%
54 Democrats: 9.3%
53 Democrats: 28.1%
52 Democrats: 35.6%
51 Democrats: 19.6%
50 Democrats: 5.2%
49 Democrats: 0.7%
48 Democrats: 0.04%

The odds of Democrats keeping 55 or more seats in the Senate are twice that of Republicans winning the chamber.

Other forecasters, most notably Nate Silver, give Republicans a far higher chance of controlling the Senate (18% in his latest forecast). However, using the odds that Nate projects for each individual Senate campaign, he actually only shows a 2.3% chance of a Republican takeover if the election were held today, not an 18% chance.

Seat Outcome Chart, using Fivethirtyeight odds

Nate ends up with an 18% of a GOP takeover due to the final step of his methodology. In this step, he makes some adjustments to no longer consider each campaign to be an independent variable. As such, he accounts for the possibility of systematic model error toward one party or the other. This makes what seem to be highly unlikely outcomes in my model (which views each campaign as an independent variable) much more likely.

The point being is that even in Nate's methodology, there would have to be substantial, systematic error across hundreds of polls to give Republicans even an 18% chance of taking the Senate. However, when the polling averages are viewed as independent variables, the chances of Republicans taking the Senate are remote in both our models.

IMHO, the odds of systematic polling error in favor of either party is virtually nil, since each individual poll should be understood as an independent variable. Further, what little chance there is of systematic error favors Democrats, not Republicans, in the form of cell phone only households and early voting. This is especially the case for early voting, which seems to be going quite well for Democrats. If polling does turn out to be busted in 2010, it will because of Dem GOTV.

Notes
--Only campaigns closer than 12.0% are listed. If a campaign isn't listed here, then it is not currently as close as any of the campaigns listed here.

--With few exceptions, all polls used in the averages are taken from Pollster.com.

--Click here for the Senate Snapshot methodology.

Originally posted to Daily Kos on Mon Oct 18, 2010 at 05:46 PM PDT.

EMAIL TO A FRIEND X
Your Email has been sent.