« Crunching the Numbers on Jersey | Main | At Long Last »

Sunday, May 23, 2004

General Election Cattle Call, May 23

Posted by Chris Bowers

(May 19 Results in Parenthesis, cross-posted at MyDD)

National Two-Party Vote Popular Projection
Kerry: 51.9 (52.4)
Bush: 48.1 (47.6)
Status: Lean Kerry

Electoral Vote Projection:
Kerry: 311 (337)
Bush: 227 (201)
States Changing Hands: FL, NH, OH

Bush is closing the gap. The race is now teetering on a return to ���too close to call.���

Since I have been asked several times, here is a description of how I project states this far from the election. I do not always trudge through the entire process, as it is very lengthy. If the ���assumed��� standing in the state leaves me confident one candidate or the other is ahead, I stop right there and get on with my life (which, this afternoon, includes seeing Howard Dean!).

--Using the national two-party vote projection along with the partisan index projection, first calculate the ���assumed��� standing in the state. For example, the current national two-party vote projection is +3.8 for Kerry. The partisan index for New Hampshire is +1.8 GOP. This makes New Hampshire an ���assumed��� +2.0 Kerry.

--Assemble all of the post-Super Tuesday trial heats, job ratings and not-unfavorable numbers for the state:

(Kerry-Bush)

ARG Trial (43-48)
ARG Job (34-46)
ARG Unfav (53-53)
Ras Trial (47-45)
UNH Trial (49-45)
UNH Unfav (59-54)

--Calculate the central mean of all these numbers, including the ���assumed��� standings (Kerry 51-49 Bush). For every different polling firm being used, include the assumed standings once. In New Hampshire, with three different post-Super Tuesday polling firms, include the assumed standings three times. In order to balance the calculation, translate every poll number into a 100.00 scale before the calculation.

--Viola! The state is projected:
Kerry: 50.4
Bush: 49.6

--In a situation such as this where the state is ���too close to call,��� I will go ahead and call it for the candidate who is currently ahead as long as the long-term voting trends do not go against that candidate. In this case, New Hampshire has been strongly trending Democrat for sometime now, and so I feel confident in projecting the state for Kerry.

Posted at 02:49 PM in General Election Cattle Call | Technorati

Comments

this has got to be the most ridiculous system for projecting an election that i have ever seen in my poll-watching life. you site all polls as if they have the same margin of error and same record of reliability (for example, zogby predicted dole would beat clinton in '96 -- whoops, gallup is known for having the biggest and most reliable sample in the industry). some pollsters are known for using techniques that pull unrepresentative samples, yet you quote them all equally. you don't even post the date of a poll. taking polls taken on different dates is like apples and oranges! just because people felt one way 5/1-3 doesn't mean they feel that way 5/5-7.

furthermore, if i understand your post correctly, you would take a 50-45-5 result, let's say, and "translate" it to a "100.00" scale? so what happens to the 5% DK's?

in a situation where it's too close to call, you don't just "call it for the candidate who is currently ahead" because that's simply not what the polls are saying. there is a margin of error for a reason: you cannot say that a poll that gives kerry a 2 point lead with a 3 point margin of error means that kerry is winning. The term statistical dead heat means something: that the race is a virtual tie! "calling" it for the guy who happens to come out ahead this time is irresponsible.

you take the mean of a whole bunch of poll numbers measuring different factors correlating to the final vote tally in different ways and give them equal weight, as if right track/wrong track has the same impact as personal negatives.

a word of advice: it might be wise to take advantage of your dc location and talk to some people who really know how to analyze polls and predict elections; might help with your "project." if you do ever get your act together and figure out how to interpret poll results, i'll be interested to read your posts.

Posted by: dchead at May 26, 2004 11:01 AM | Permalink | Edit Comment | Delete Comment

A word of advice to the commenter above: If you want to make reasoned, constructive criticism, please do so. But the unnecessarily vituperative tone of your comment is not appropriate for this site.

Posted by: DavidNYC at May 26, 2004 12:29 PM | Permalink | Edit Comment | Delete Comment

granted the tone was vituperative (i'm not even sure what that means), but if you read the comment there is a lot of substance in there, and constructive criticism abounds. i encourage you to consider some of my critiques because your site would be much better off for it.

Posted by: dchead at May 26, 2004 03:00 PM | Permalink | Edit Comment | Delete Comment

Sorry to keep coming back and posting (since I'd say blog rule number one is "if you don't like the blog, just don't read it"), and I will stop reading, but just after I point you to this article by the polling directors at WaPo and ABC that ran in the WaPo on May 7th. Entitled "Ballyhoo Over 'Battleground States'" it attacks the very idea of being able to determine which states will be competitive this year by using the previous election's results (i.e., the entire premise of this site). It also tears apart the very notion of states "swinging" in presidential elections, in that most states labeled as such do not swing at all, but rather consistently go for one party by relatively narrow margins. Anyway, enough from me, I'll provide the link and be on my way: http://www.washingtonpost.com/wp-dyn/articles/A7010-2004May6.html

Posted by: dchead at May 26, 2004 03:26 PM | Permalink | Edit Comment | Delete Comment

If you examine the archives of this site, you'll find that that op-ed has been addressed in great detail, here and here.

Furthermore, Chris has examined past voting trends in very careful detail.

Posted by: DavidNYC at May 27, 2004 04:17 PM | Permalink | Edit Comment | Delete Comment

football gambling interaction pro basketball betting fantasy football gambling gambling problem sports professional sports gambling best online sports betting college football gambling tip basketball betting line offshore sports betting sports gambling bonus free football gambling basketball gambling gambling on football game sports betting web site football gambling selections.com college basketball gambling sports gambling site mlb baseball online betting online sports betting football sports gambling online baseball betting pick buffalo bills football betting sports betting line online basketball gambling baseball betting site sports book betting line ncaa college football betting line pinnacle sports betting football gambling portal basketball gambling software ncaa football gambling sports betting information ncaa basketball betting college football sports betting betting bill buffalo football nfl online atlantic city sports gambling american football betting mlb baseball online gambling sports betting spread england football gambling new patriot sports betting football gambling football gambling line nfl

Posted by: Dan at November 12, 2004 02:42 AM | Permalink | Edit Comment | Delete Comment