Wednesday, October 1, 2008

Here's the Deal...

Alright folks. This is a change election, or so I've been told, and to have an electoral college analysis that does not respond well to changes in polling, is not necessarily a plus. I can see the writing on the wall and have now for a few days.

So here's the plan, both short and long term.

Tonight, I'll update the map as if there was no change to the formula and then have a few words to say in a separate post about the Muhlenberg polling discussion that sprang out of last night's update. It will likely not be tonight, but tomorrow I will revise the formula behind the map and see if we can kick start things around here.

Now the big question now is, well, why didn't you make this change after McCain's convention in St. Paul? Things were moving in his direction then. Why favor Obama with a change to the formula now? These are all valid and good questions. [I should have thought of them myself.] The reason is that conventions are part of what Jim Campbell would call the predictable campaign. We expected McCain to get at least something of a bounce out of the convention. The average's job in that scenario wasn't to mute the shift toward McCain, but to account for the likely temporary nature of the fluctuation. What we are witnessing now in the polls is something different and the mountain of past polls in our data set are too much of an anchor on the new -- and different -- data we now have. In other words, some revision is necessary to capture the true nature of the change. Whereas the convention bounce was temporary, the movement now likely isn't.

I'll be back shortly. I need to add in the afternoon polls to the averages.


Recent Posts:
The Electoral College Map (10/1/08)

The Electoral College Map (9/30/08)

The Electoral College Map (9/29/08)

5 comments:

Jack said...

For some reason, I thought your methodology was to use only the last three polls and put more weight on the most recent one. Which is why I was really confused by the extreme stability of the electoral map and your comment that VA would need an Obama +10 to flip despite already being on the watchlist.

When you change methodologies, even if just going back to an old one, please detail exactly what you will do, to avoid confusion.

Josh Putnam said...

Jack,
The switch from weighting the three most recent polls to only the most recent poll took place just after the conclusion of primary season. But both have always included the polling from Super Tuesday on. The most recent ones aren't weighted so much as the earlier ones are discounted.

Here's the link to that post..

Jack said...

Yeah, I remember that, but I must have misread it to mean that you were discounting all polls before the last three. My mistake.

Wayne said...

Sounds good, I look forward to reading your next post.

Josh Putnam said...

Actually, there are some interesting changes today without any tweaks, Wayne. Stay tuned.