Latest Electoral College Projections

In this, the final days before the election, The Imperial Gazette will try to keep tabs on the ever-changing Electoral College projections. And we'll provide a little analysis of the pros and cons of each source.

Overview

To win the presidency, an absolute majority of 270 of the Electoral College's 538 votes are necessary.

Source	Bush	Kerry	Toss-Up	Date
CBS	222	190	126	30-Oct-2004
CNN	227	207	104	28-Oct-2004
Charlie Cook	222	207	109	29-Oct-2004
Election Projection	259	279	-	30-Oct-2004
Electoral Expectations	272	266	0	30-Oct-2004
Electoral Vote Predictor #1	231	298	9	1-Nov-2004
Electoral Vote Predictor #2	287	244	7	1-Nov-2004
Federal Review	296	242	-	26-Oct-2004
First in the Nation	207	182	149	30-Oct-2004
First in the Nation (projected)	280	258	-	30-Oct-2004
Ed Fitzgerald (current)	255	283	104	30-Oct-2004
Ed Fitzgerald (projected)	259	279	104	28-Oct-2004
LA Times	168	153	217	30-Oct-2004
E. Alan Meece	266	272	-	30-Oct-2004
Mind the Gap	262	276	-	27-Oct-2004
National Journal	217	184	137	29-Oct-2004
New York Times	227	225	86	30-Oct-2004
NowChannel (was 2.004k.com)	245	283	10	31-Oct-2004
Political Oddsmaker	242	238	58	27-Oct-2004
Pollbooth	276	242	20	28-Oct-2004
Pollkatz	251	287	-	26-Oct-2004
President Elect	283	255	-	28-Oct-2004
MyDD (Chris Bowers)	247	291	-	1-Nov-2004
Race 2004 (today)	207	185	146	1-Nov-2004
Race 2004 (projected)	289	249	-	1-Nov-2004
Rasmussen	222	186	130	30-Oct-2004
Real Clear Politics	232	207	99	29-Oct-2004
Larry Sabato	276	262	-	26-Oct-2004
Robert Silvey	290	248	-	28-Oct-2004
Slate	266	272	-	30-Oct-2004
Strategisphere	274	264	-	27-Oct-2004
Triapas	281	257	-	28-Oct-2004
TruthIsAll	217	321	-	30-Oct-2004
USA Today	205	165	169	29-Oct-2004
Sam Wang's Meta-Analysis	255	283	-	30-Oct-2004
Zogby	286	231	-	28-Oct-2004

A Look at Some of the Sources

Some of our sources (like the Electoral Vote Predictor and MyDD) aggregate data from polls conducted by others, while others like Rasmussen conduct their own polls. Both approaches have their own strengths and weaknesses. Organizations that do their own polling have the benefit on following a common weighting methodology in each state. But many of them only target a few states, usually the battlegrounds, but even then they might miss a state that suddenly creeps up on the radar. The aggregators have an advantage of a greater diversity of approaches and more frequent updates, but can sometimes produce wacky results when different states come from different polls using different methodologies. (Even so, respected political analyst Charles Cooks warns against putting too much reliance into any one polling source.)

Electoral Vote Predictor 2004

The Electoral Vote Predictor is one of the largest and most rigorous of the polling aggregators. In order to avoid any possible impropriety, the "Votermaster," as the webmaster calls himself, carefully uses all of the available polls, including paid and premium polls and even the ones that have come under fire for being partisan or using skewed methodology. He always uses the most recent poll available for a given state, which is based on the poll's middle date rather than starting or ending date. When multiple polls have the same mid-point, he uses the shortest poll as the one with the most accurate snapshot. If there's still a tie, he averages the tied polls' results.

EVP currently projects that Bush has at least a narrow lead in states totaling 285 Electoral Votes, while Kerry trails with 247. This includes some surprise projections of Bush carrying Ohio, Wisconsin, New Mexico, and Hawaii, but losing Colorado, and Arkansas remains too close to call. Note, however, that most of these surprises for Bush come from Zogby, which has a Republican bias to their sampling methodology (which we'll discuss in another article).

NowChannel.com

NowChannel.com, formerly know as 2.004k.com, collects and aggregates other pollsters' data like the Electoral Vote Predictor. It uses a slightly differently methodology in that it determines which is the most recent poll by their release date rather than their midpoint. They ignore polls from blatantly partisan sources like Strategic Vision (which in 2000 predicted the day before the election that Bush would win over 400 Electoral Votes) as well as Internet-based polls like Zogby Interactive. One of the nicest things about their interface is the complete transparency - it's the easiest site to "dive under the hood" to check the data they're using.

2.004k.com currently shows Kerry winning with 287 Electoral Votes, including taking all three top major swing states: Florida, Ohio, and Pennsylvania. They also show Kerry taking a few of the other swingers, including Iowa and New Mexico, but losing Hawaii and Minnesota to Bush. Wisconsin remains tied in their most recent poll.

Rasmussen

Rasmussen conducts their own state-by-state telephone polls around the country. However, they do have some limiting flaws. First, they cut down on costs by using automated telephone systems rather than paid employees to call voters. Respondents are then asked by an automated machine to press a button on the phone to indicate their choice (e.g., "Press one if you plan to vote for Bush, press two if you plan to vote for Kerry"). This approach makes it impossible to properly screen for the respondents age and other criteria, but down allow for more frequent polling. The other flaw is that their site limits how frequently they update the free sections of their site. The aggregators who pay for Rasmussen subscription feeds get more frequent data, but those of us who are poor and unemployed have to rely on older data.

Rasmussen currently predicts that Bush will take 222 Electoral Votes while Kerry will take 207 votes. They've put a huge number of states back in the too-close-to-call column, including Florida, Ohio, Pennsylvania, Nevada, New Mexico, Minnesota, Wisconsin, Iowa, and New Hampshire.

MyDD

MyDD may be the most controversial source, as they are unabashedly pro-Democratic. They aggregate data from other polls, though (unlike the Electoral Vote Predictor) they don't include polls from sources like Strategic Vision that skew so far to the right that almost everyone believes they are too biased to be credible.

They take other pollsters data but re-analyzes it to predict the final outcome based on how current undecided voters are expected to skew. It's a controversial approach, but an interesting one worth examining. We'll see on election day whether their approach holds water or not.

MyDD currently predicts that based on how undecided voters are expected to break, Kerry will win with 304 Electoral Votes, including winning the key swing states of Florida, Ohio, Pennsylvania, and New Mexico. Iowa's seven Electoral Votes remain too close to call.

Race 2004

Race 2004 is another independent website that is collecting polls and aggregating the data to predict a winner.

Race 2004 shows a huge number of undecided states (accounting for 128 Electoral Votes). But they also project that on Election Day Kerry will have the advantage in the toss-up states and will narrowly pull out the election with 273 Electoral Votes:

The Imperial Gazette

Articles

Links