(Update (8/26): My prediction has changed, and is now at 280-258, but I have not included this in the survey's totals or averages.)

hR's (8/18)
Kerry 243 - Bush 167 - ?? 128
(243-196-99)

Matthew Hubbard (8/21 - new):
Kerry 316 - Bush 213 - ?? 9 (CO)
(new to survey, was: 327-211 on 8/14)

LA Times (8/22):
Kerry 161 - Bush 147 - ?? 230
(was: 172-147-219)

Leip Atlas (1103 user predictions compiled; 8/22):

• state by state medians
  Kerry 264 - Bush 274
  (no change)
  
  
• overall medians (new):
  Kerry 289 - Bush 249
  (noted in last survey, but not included; no change)

mattb25 (8/19)
Kerry 327 - Bush 211
(was: 332-206)

MyDD (8/22)
Kerry 316 - Bush 222
(was 327-211)

Our Campaigns (8/22):
Kerry 264 - Bush 274
(no change)

Pollbooth (8/20)
Kerry 281 - Bush 222 - ?? 35 (OH, WI, WV)
(was: 281-211-46)

Pollkatz (8/19):
Kerry 297 - Bush 241
(was: 347-191)

President Elect 2004 (7/29 - not updated since last survey):
Kerry 264 - Bush 274
(not updated)

Kenneth Quinnell (8/8 - not updated since last survey):
Kerry 303 - Bush 207 - ?? 28
(not updated)

Race2004 (8/22):
Kerry 332 - Bush 206
(*was: 318-207-13 as of 8/7)

(*Note: This site recently underwent a redesign, and because of that, I believe, last week I took the wrong numbers from it. I had been taking the "If the election was held today" projection, but instead I extracted the more immediately apparent summary numbers in the box at the top of the home page. Since I can't find an archive to see what the proper numbers for last week should have been, for the purpose of my stats on votes gained and lost I've substituted the numbers for the week before, which I am certain were correct.)

Rasmussen (8/20):
Kerry 223 - Bush 183 - ?? 132
(was: 232-197-109)

Running the Numbers (8/20):
Kerry 296 - Bush 242
(was: 312-226)

Larry Sabato (8/16 - updated):
Kerry 274 - Bush 264
(was: 290-248)

(Note: Sabato finally updated the numbers on his site -- they hadn't been changed since June -- but only to keep them the same as they've been all that time. This, I guess, rescinds his comments reported in this article in the Mobile Register, which quotes him as giving Kerry a 290-248 win. Also, a curiosity: although the update is dated "August 16, 2004," it was not on the site when I initially prepared this iteration of the survey in the wee hours of August 22nd.)

Samboni's State-by-State (8/21):
Kerry 307 - Bush 222 - ?? 9 (CO)
(was: 327-211)

Benjamin Schak (8/18):
Kerry 316 - Bush 222
(was: 322-216)

Robert Silvey (8/19):
Kerry 296 - Bush 242
(was: 301-237)

Tradesports/intrade (8/22):

• Fitzgerald conversion [50 = toss-up]
  Kerry 264 - Bush 269 - ?? 5 (WV)
  (was: 264-247-27)
  
  
• Geekmedia conversion [45-55 = toss-up] (new)
  Kerry 264 - Bush 222 - ?? 52 (FL, OH, WV)
  (noted, but not included in last survey; was: 264-231-43)

(Note: For further information on converting Tradesports data see here and here.)

Tripias (8/22):
Kerry 296 - Bush 242
(was: 316-222)

TruthIsAll (8/21):
Kerry 307 - Bush 221
(was 337-201)

Sam Wang (8/20):
Kerry 313 - Bush 225
(was: 318-220)

Washington Dispatch Page Two Politics (8/22 - new):
Kerry 296 - Bush 242
(new to survey; was: 316-222 on 8/15)

Wayne in Missouri (8/20):
Kerry 291 - Bush 247
(was: 311-227)

David Wissing (8/20):
Kerry 296 - Bush 242
(was: 316-222)

Young Conservatives (8/15 - not updated since last survey):
Kerry 188 - Bush 237 - ?? 113
(not updated)

Zogby/WSJ (8/23 - updated):
Kerry 286 - Bush 214 - ?? 38 (FL, MO)
(was: 318-220)

2.004k.com (8/22):
Kerry 306 - Bush 207 - ?? 25 (AR, CO, WI)
(was: 321-211-6)


STATS

44 SITES
(was: 39 - 5 sites added)

Kerry winning: 32 (was: 28)
Kerry ahead: 4 (was: 5)

Bush winning: 4 (was: 5)
Bush ahead: 4 (was: 1)


Kerry gained: 5 (was: 21)
Kerry lost: 20 (was: 3)
Kerry no change: 9 (was: 8)

Bush gained: 18 (was: 1)
Bush lost: 11 (was: 22)
Bush no change: 6 (was: 9)

?? gained: 11 (was: 5)
?? lost: 2 (was: 2)
?? no change: 0 (was: 3)

Site no change: 7 (was: 7)
Site not updated: 4 (was: 4)
New: 5 (was: 7)
Reinstated: 0
Temporarily dropped: 1


AVERAGES: ALL SITES
MEAN: Kerry 284 - Bush 227 - ?? 27
(was: 291-222-26)

MEDIAN: Kerry 296 - Bush 222 (remainder: 20)
(was: 301-220-17)

MODE: Kerry 316 - Bush 222
(was: 264-211-63)


AVERAGES: SITES WITH NO ??
MEAN: Kerry 301 - Bush 237
(was: 306-232)

MEDIAN: Kerry 307 - Bush 227 (remainder: 4)
(was: 316-222)

MODE: Kerry 316 - Bush 222
(was: 327-211)

SUMMARY

This week has seen improvement in Bush's position and a erosion of Kerry's lead, but not so much so as to change the status quo: a supermajority of sites still show Kerry winning (71%) or leading (81%), all averages show Kerry clearly winning, and Kerry's position is still around 300 votes -- but whereas last week he was probably a little over that mark, this week he seems to have fallen a bit under it.

Note that of the 4 sites that show Bush winning, 2 have not been updated recently, and 2 are compiled from group averages and presumably change very slowly (and not at all if the group members don't update their predictions regularly). Of the 4 sites that show Bush only ahead but not winning, 2 have not been updated recently, one (DC Political Report) greatly increased the number of toss-up states it reported (indicating significant uncertainty), and the last (Tradesports) is based on group activity.

It's worth talking briefly about Tradesports. Although my scoring system (the "Fitzgerald conversion") still shows Bush ahead (264-269-5), there's been significant weakening of his position, as shown by the "Geekmedia conversion," which defines a toss-up state more broadly. States that Bush would still win if the election were held today (and was based on Tradesports data) are being less strongly held by him than they had been. WV, for instance, was 52.9 and is now exactly 50.0, OH has fallen from 58.8 to 55 and NV was 59.9 and is down slightly to 58.0. Florida remains precariously poised at around 50 (currently 50.4), which is down from where it was several weeks ago.

Kerry, on the other hand, is either holding his own in the battleground states he "owns" according to Tradesports, or getting slightly better. In other words, overall the Tradesports momentum seems to be very slightly shifting Kerry's way, but, in any event, the data certainly no longer supports the clear 264-274 win for Bush it did as recently as two weeks ago.

Nevertheless, despite these caveats, Kerry's lead has clearly slipped a little.


UPDATES

• For a few days after I publish the survey, I'll update figures, make corrections, and add new sites that come to my attention. I'll note these changes here, and mark the entry for each altered site as well.
  
  
• Chris Bowers, Larry Sabato and Election Projection updated their sites, and the survey's been updated to reflect their changes.
  
  
• (8/23) As expected, Zogby/WSJ updated their battleground poll, but, unexpectedly, without the addition of four new swing states (AZ, CO, NC, VA) they had said they would be adding. This time out, Zogby didn't make an explicit Electoral College projection (as they did for their last iteration), so I assumed the status of the non-swing states to be as in 2000 and totaled up the vote myself and included them in the survey. (8/27): I found an article on the Zogby website which gave their official count, so I've updated the entry here to reflect that.
  
  
• (8/23) Also, my own prediction has changed, but I have not included the new figures in the survey's totals or averages. (It changed again on 8/26.)
  
  
• (8/24) I went back and figured averages for the past iterations of the survey I did before I started calculating them (and thanks to Winston and Stephen from MA for encouraging me to start doing so), and I've included the results as a graph, just below, which is the first image to be officially posted on unfutz. I'm hopeful it will come out ok.
  
  
• (8/26) Matthew Hubbard wrote to alert me to his Electoral College tracking site. Since his most recent numbers date from 8/21, I've included them in the survey and updated the totals and averages.
  
  
• (8/26) I've also added CK Rairden's figures from Washington Dispatch Page Two Politics.
  
  
• (8/26) See the note below about National Journal/Hotline.
  
  
• (8/26) JeffR in comments (on another post) brings another site to my attention, My Election Analysis. Since the latest numbers here (311-227) are dated 8/26, I won't add it into this survey, but will do so for the next one.

PREVIOUS

• August 15
• August 7
• August 1
• July 25
• July 20
• July 6
• June 25

E.C.

FEC
NARA
Web Zine

NOT INCLUDED

National Journal/Hotline (7/30):
Kerry 261 - Bush 195 - ?? 82
(awaiting access to site)

(An article on 8/25 in the Atlanta Journal Constitution quotes the latest Hotline -- that is, National Journal -- numbers as 252-249-37. Since I can't seem to get the free access to NJ during the convention, as I was promised I'd get, and I won't in any case get consistent access to their figures because I can't afford the exorbitant price of their subscription -- unless someone slips them to me under the table -- I'll probably end up dropping NJ from the survey, after using these AJC-reported numbers in my next interation.)