Monday, October 25, 2010

8 Days and Counting With No Major Changes, What the Gamblers Say, Poll Selection 101, On Macroeconomics, Just for Fun Predictions

Projection Totals
Senate: 50+ Democratic Caucus (48 Dems, 2 Ind, VP Tie-Breaker), 50 Republicans
House: 231 Republicans, 204 Democrats

Several new polls in the past 72 hours but not a lot of new news to report. My House projection got closer, with generic polling data tightening slightly, leading to a 3 seat swing to the DEMs, but still solidly a projection for GOP control.

In the Senate, only 2 race rating changes and they are both pretty minor and technical. Ohio fell behind my magic (and arbitrary) 20 point threshold again and slips back to a "Likely Hold" versus a "Safe Hold" for the GOP, but at this point, 19 points is pretty much out of reach. Missouri also slips slightly below a threshold, with the current 9.9 point GOP lead moving it back into the "Lean" column versus the "Likely" column, but this is still a long shot for the Dems to flip this seat.

I thought rather than my normal comparison to other major political sites (our friends at, and, I'd take a break and bring back in the gambling money.

For frequent readers, you will know that is the website that lets you bet on anything, including politics, in a stock-market style format. Gamblers tend to follow the polls pretty closely and are often pretty good at predicting election outcomes, although, I would argue based on the record of this site over the past 2 years, not nearly as good as a statistical study of poling data.

At any rate, here are intrade's odds on our closest races (all calculated based on the ratio of the contract price of one candidate to another)

California: DEM favored 3:1
Washington: DEM favored 5:2
Alaska: GOP favored 2:1 (vs. all others)
Colorado: GOP favored 3:2
Nevada: GOP favored 2:1
Pennsylvania: GOP favored 3:1
Illinois: GOP Favored 2:1
West Virginia: GOP Favored 4:3
Kentucky: GOP favored 6:1
Wisconsin: GOP favored 6:1

So, the gamblers pick the same winners as our projection, but not necessarily but the same order in terms of odds. I find a few of the odds a little surprising (how can Kentucky have the same odds as Wisconsin?), but generally not that surprising.

Why I Don't Use Partisan Polls
You may notice that in several races, my projections differ pretty significantly from those of other sites. This is partly due to the use of multiple methods of weighting and averaging, versus many sites that simply use a "most recent poll" or a simple average of recent polls. However, one issue that is pretty fundamental is the use of partisan-affiliated polls.

I do not use polls from partisan-affiliated firms such as Public Policy Polling in my averages. My reason is simple...if you are paid by or affiliated with a party or candidate, I cannot presume objectivity in your polling methodology.

In the West Virginia race in particular, there is a Public Policy Polling poll that shows Gov. Manchin leading by a whopping 10 points, whereas non-partisan polls all show him trailing by small margins. Now, it could well prove out that PPP is right and all the other polls are wrong. But PPP is a Democratic-affiliated firm, so I can't trust the sample selection and weighting, especially so close to an election.

Macroeconomic Issues
A fair criticism of my analysis of GDP growth versus parties in power (and one that I owned up to in my long list of caveats on the limits of such an analysis) is the impact of macroeconomic events far beyond the control of government on economic growth.

In that spirit, I'll commit to doing a similar analysis on three statistics that are directly in the control of government: taxes, spending and deficits. Results this weekend.

Some Just for Fun Predictions
None of these will happen (I hope), but just to have some fun in the current political environment, imagine what could happen:
(1) Alvin Greene wins the South Carolina Senate race with 76% of the vote, despite polling 30+ points behind immediately before election day. Greene attributes his wins to grass roots campaigning and then goes on to admit that he doesn't know what grass roots are or what a campaign is. Diebold swears by the unfailing accuracy of its voting machines.

(2) Christine O'Donnell pulls off a surprise upset in Delaware and immediately declares that she is, in fact, a witch and cast a spell on voters entering the booth to vote for her. She commits to making the cause of her life over the next 6 years Witch and Warlock rights.

(3) Marco Rubio wins big in Florida on Tuesday and announces Presidential exploratory committee on Wednesday. Sarah Palin criticizes his lack of experience.

(4) Harry Reid loses to Sharron Angle and declares that he will spend time with family and work on important social issues such as "breaking down barriers for negros".

(5) Rand Paul wins in Kentucky and declares that he really meant all the crazy stuff he said during the campaign (okay, this one probably will happen!)

More updates later in the week.
If you like this site, tell your friends.

No comments: