Why Another Blog?

I've decided to set up another blog, (my other one is called Writer's Musings), because there are some topics just too weighty for that blog.

So here it is. In this space I'll explore more serious issues in more detail. I do not expect visitors to agree with me in all cases.
In this forum feel free to take off the gloves, grab a handful of mud and fight for what you believe in.

Simple rules, rather like cage-fighting in the blogosphere:
No direct name calling. No excessive profanity. No whining when smacked in the face with mud.
Sling inuendo. Feel free to ask leading questions even if in a snide tone.

Saturday, September 20, 2008

Electoral Map Review

In my last post I discussed how to look at poll numbers. One source of bias that I neglected to mention, and one which further muddies the water, is the difference between polling “registered voters” and “likely voters.” Polls that use the “registered voter” model will almost always show a Democratic bias. Even in states that are strongly “red states” the strength of the state’s “redness” is reduced when using the “registered voter” model.

The “likely voter” model is thought to eliminate this bias, but it can also be biased if the pollster simply asks, “are you likely to vote?” A preferred method is to use historical data to identify the demographic groups with stronger likelihood of voting. This is where the “weighting” you may have heard of being discussed comes into play. Faulty weighting can reintroduce bias.

Electoral Vote Maps

The first map is the Real Clear Politics Electoral count for 19 September 2008. My map for the same day appears below. Note that the maps are not really too different, but a couple states change the numbers. On my map, I applied all the factors I discussed in the previous post. I threw out outdated polls (pre-convention) and I scrutinized polls likely to contain partisan bias or flagrant statistical outliers. In some cases I looked at the historical trends both in previous elections and poll trends to break what currently look to be dead heats.

Real Clear Politics Electoral Map 19 September 2008

My Electoral Map 19 September 2008

Let us begin with some obvious and easy to justify differences. MT and ND are shown by Real Clear to be “pink” or leaning to McCain. I contend that both are strong “red” states. Neither status changes the Electoral Vote count.

MT: The MT average includes one poll from February where McCain had an 8% lead and one from May again showing an 8% lead. The latest Rasmussen poll conducted after both conventions shows McCain up 11%. Obama had a brief moment in the MT sun in July, but neither before nor since has he polled in positive territory in MT. MT is a “red” state, not a “pink” state.

ND: The ND average includes a February poll where Obama had a 4% lead and a preconvention poll paid for by the Democrats where he had a 3% lead. The only post-convention poll was conducted by Rasmussen and it shows a 14% McCain lead. Again, this state was not ever in danger for the Republicans and remains a “red” state.

The Map Differences

NV: McCain holds a slight lead, but I’m letting history say that this state is actually leaning to McCain instead of being a tossup. Over time most polls show McCain on top and the state usually goes Republican. The InAdv polls look to have significant statistical bias. However, in this case I left their results in the numbers pending newer polls. If future polls show results closer to the Rasmussen numbers, then I’d toss the InAdv poll numbers. Either way, I think that McCain wins this state by about 4%.

IN: Real Clear shows this one a tossup, but in the middle of their numbers is a poll by the IndyStar that deviates from other polls by about 200% showing Obama up 3%. Their own data shows they relied heavily on interviews in Indianapolis and they admitted that Indianapolis is the only place in the state where Obama has any lead whatsoever. After tossing this statistically flawed poll, the state is back to “pink.” I’m calling this one McCain by at least 5% in November. The last time IN went Democratic was in 1964.

VA: Given Obama’s steep falloff the last few days here, I’m putting this one in the McCain column. Most polls since the conventions are showing McCain with about a 4% lead. I’m going with 44 years of Republican victories here and saying that McCain wins VA by at least 7% come November, which is a falloff from Bush in 2004 and 2000.

In Agreement For Now But…

OH: Most polls are showing McCain up by an average of 4%. A recent CNN/Time poll is showing Obama by 2% which brings the numbers back into a range making this look closer. I didn’t throw any numbers out on this one yet and though I’m also showing it as a toss up. Waiting on new numbers, at this point I’m predicting McCain by 4% in November.

CO: Real Clear and I agree on this one, but Real Clear has added the clearly biased InAdv poll showing Obama up 10%. This number is about 1000% higher than any other poll so I tossed their poll numbers. All other polls show this one a dead heat. Almost all InAdv poll numbers in every state are showing a deviation of 200% or higher over other polls. Either they are deliberately biased or they have faulty weighting. Either way their numbers are suspect. I think this one will tilt to McCain, but I’m holding it in tossup until I see more numbers from other polls with more reliable credentials.

PA: This state was only ever “baby blue” so there was always a chance for McCain. The most recent numbers are showing a dead heat with between 6-10% undecided. Most significant is that Obama had as much as a 9% lead before the conventions and that has evaporated. Anything can happen here. If this state goes McCain on November 4, McCain will win the Presidency.

MI: Has gone from “baby blue” to “gray.” Combined with increasing gains by Republicans over the past two elections, this is a trend that should worry Democrats. Obama had a 9% lead at one point. I don’t think it likely to flip, but Obama will be forced to expend resources to defend this state.

MN: This year is probably the Republican’s best shot at taking this state. Obama had a solid 15% lead in this state that has evaporated and the state has trended Republican the past two elections. It is now a dead heat. This one may flip, but it will still be close. Neither candidate will win by more than 3%.

WI: This is another state that Democrats should worry about and is the Republicans’ second best chance to flip a traditionally Democratic state. Obama had an 11% lead in July that has steadily eroded. This one is a statistical dead heat. McCain has come on strong here the past week. Even if Obama hangs on, he will be forced to spend resources in order to defend this state. Like MI and MN, this state has been trending Republican in recent elections.

NH: I suspect NH will move into the Obama column within the next week and that he will win it by about 4% in November.

Closer Than They Should Be

WA: Now that McCain’s convention bounce has worn off, this one is back to leaning to Obama. WA was looking like it was moving to being a tossup. Still, considering that he once held a commanding 12% lead that is now only a little over 4%, one must wonder. This one could tighten again, but there is little play in the numbers.

OR: I’m thinking Obama wins this one by about 5% come November. His numbers, while consistently positive, have been sliding. Rasmussen had Obama up by 10% prior to the conventions, but on 15 September had him at only 4%. That may be the tail end of the McCain bounce, but we’ll need to see more data. The Portland Tribune gave Obama a 10% margin in a survey done prior to Rasmussen. I discount that number until I see more.

The bottom line is that the election remains close. Will we see a shakeup following the debates?

1 comment:

Tempest in a Teapot said...

WRT my last comment, I don't men to suggest that I expect to see those exact figures on election night. Just that judging from numbers one week out from the election, my reading at the beginning of September was pretty damned accurate.