/* Google Analytics Code asynchronous */

Saturday, May 10, 2008

SuperDay+14: The view from the sidelines

I'm late getting to it, but Jay Cost has a nifty web-spreadsheet that lets you project the aggregate vote (it would be wrong to call it a 'popular vote' for reasons already outlined).

CONFIDENCE INTERVAL

Playing with the figures, one could guesstimate that the range of adjustment in favor of Clinton might be +350K to +500, yielding an all-in bracket of +87K to -89K in the "all in" bottom line of the sheet. We could call that our one standard deviation range.

MORE MEANINGFUL ESTIMATES

It's silly enough to try to add up these numbers and pretend they mean something, because they are apples-and-oranges, caucuses and primaries, both open and closed.

Jay doesn't say how he got to his estimates for the difficult states of IA, WA, NV, and ME, who have not provided a breakdown of the popular votes. However, I have backed into numbers quite different than Jay's, digging up the following disclosed numbers of people who participated in the caucuses: 239K, +250K, 177.6K, 43.8K.

Still, this is unsatisfactory.

MORE PEOPLE VOTE IN PRIMARIES - ADJUSTING CAUCUS NUMBERS UPWARD

A better result (more consistent, but not necessarily more meaningful) would be to adjust for ALL the caucus states.

Caucuses turn out fewer voters than do the primaries. Caucuses this year have turned out about 53% fewer voters than have primaries, using the actual primary/caucus results compared to Kerry 2004 actual votes in the general.

To adjust, all we have to do is apply the margin of victory in each state and a supposed turnout figure to pretend to have apples-to-apples numbers (basically normalizing the results around the 2004 Kerry vote turnouts).

It's also true that the caucuses have turned out wider margins. I don't know why that is. The average caucus margin of victory is nearly double that of primaries. Who knows why, but let's adjust, to be "conservative".

MORE VOTES DUE TO OBAMA

I end up with over 560,000 more votes due to the Obama column.

1. Revised "aggregate vote" figures for caucus states. Old vote margin is either estimated (IA, NV, ME WA) or reported. New vote margin is based on (a) 50% adjustment (reduction) in caucus victory margin (b) an adjustment to the caucus turnout figures, based on actual votes for Kerry in 2004, in each caucus state (i.e. normalization of the average caucus-state turnout to the average primary-state turnout observed so far this year).

StateActual MarginHaircut MarginTurnoutKerryMargin NewMargin Old
Alaska51%25%-39%111,02517,0994,482
Colorado34%17%-35%1,001,732111,38741,201
District of Columbia52%52%14%202,970119,27063,870
Hawaii52%52%-31%231,70883,86919,519
Iowa8%4%-29%741,89821,36019,407
Idaho62%31%-35%181,09836,51213,228
Kansas48%24%-39%434,99364,41117,699
Maine19%10%-29%396,84227,4578,552
Minnesota34%17%-32%1,445,014167,55973,195
North Dakota25%12%-30%111,0529,5804,680
Nebraska35%18%-32%254,32830,68913,661
Nevada-6%-3%-29%397,190-7,967-6,649
Washington36%18%-29%1,510,201194,77345,703
Wyoming24%12%-35%70,7765,4602,067
Total



881,459320,614





Net Change560,845
Figures do not include Guam, American Samoa, Virgin Islands, or Democrats abroad. No adjustment taken for margin of Victory in D.C. and Hawaii, as these margins seem directly related to Obama, rather than some unspecified "caucus factor". "Turnout" is the margin above/(below) the general election turnout for Kerry in 2004.

WHY DO THIS NOW?

Well, Hillary is likely to say that she can still win, by taking her needed votes from among the supers. Her argument is probably going to be that she's won a "popular mandate", is ahead in the current polls in key states, is strong enough to beat McCain, and is 'ready on day one'.

This will be the "cover story" that super-delegates need to vote for her.

While that is clear a change of the rules, I suppose it's fair, because the supers can vote based on astrology, if they want to.

The question is how convincing it might be.

I don't find it much, after looking at these numbers. (Puerto Rico would have to turn in over 1.5M votes or something at a 20% margin ...).