Monday, January 18, 2010

Massachusetts Update

A few new pieces of polling data have come in today, so I thought I'd update my projections. The two new polls are as follows:
(1) Research 2000 has conducted a Sample Size 500 poll that shows the race dead even at 48%-48%-3% with 1% undecided.
(2) Management Research Group has conducted a Sample Size 565 poll that shows a solid Brown lead at 51%-41%-2% with 6% undecided.

It is worth noting that Management Research Group has significantly less experience doing political polling than the big firms and universities (Gallup, Rasmussen, Research 2000, Quinnipiac), although that certainly doesn't mean that they are wrong. It is odd that they still show 6% undecideds in a race that is happening tomorrow.

At any rate, the aggregation of all the data now looks as follows for all non-partisan polls:
Sample-Weighted Poll Average: Brown +2.3%
Unweighted Average: Brown +2.9%
Median Poll Margin: Brown +3.0%

Our Statistical Projection: Brown 50.3%, Coakley 47.2%, Kennedy 2.1%

I don't personally believe that in a race this close, that 2.1% of voters will case a protest vote for Kennedy. History shows that independents tend to underperform their polling data, but I have no way to estimate how much, so I'm not going to attempt it. It would seem logical that Kennedy bleed-offs would break disproportionately for Brown.

There are three scenarios that I can envision:
(1) A Coakley squeaker -- all the undecideds break for Coakley, turnout is on the high end and the Kennedy voters stick with the Libertarian rather than reverting to Brown. Margin: Coakley 49.3%, Brown 48.6%, Kennedy 2.1%
(2) Brown wins close -- the numbers fall more or less as we project
(3) Brown rally continues -- the undecideds all break for him and he peels off half the Kennedy voters. Margin: Brown 53.2%, Coakley 45.8%, Kennedy 1.0%

I actually view these three scenarios as fairly equally likely, although obviously I think scenario #2 is the central scenario, which is why I'm projecting it.

The intrade odds on the race have moved from even-money mid-day yesterday to favoring Brown by almost 2:1 odds. This kind of supports my three equally-likely scenarios theory -- in 2 of my 3 scenarios Brown wins, giving him the same 2:1 advantage.

I'm expecting one more poll to be released today and if it is, I will update my projection, but for now, this race remains a Lean GOP Pick-Up, by a slightly better margin than yesterday.

Regrettably, I will be traveling tomorrow night and may not be able to publish a post until later in the evening. It appears unlikely that we will know a winner for several hours after the polls close regardless, unless it turns out to be an unexpected blowout.

No comments: