<img src="
">
I don't buy this, but if true there's a lot of work to do.
KERRY SURGE FALLS SHORT
Bush 47.77% - 274 | Kerry 48.52% - 264
From the May 11 to the May 18 Composite Polls, there was a net change of 4.13% in favor of Kerry, taking him from a deficit of 1.41% to a lead of 2.72% -- the highest lead for either candidate since I began tracking the numbers on March 8. Since the Composite Poll methodology constrains weekly swings by factoring in the previous week's composite result, this should be seen as a significant swing to Kerry. That week, Kerry led by 5 in the CNN/Time poll, 4 in the Zogby poll and 1 in the Newsweek poll.
Since then, his lead has shrunk each week. To 2.38% on May 25, and now to 0.75%. What looked like a real surge for Kerry has pulled back to dead even.
And the Electoral Vote prediction is back to 2000, with Kerry gaining only New Hampshire. Welcome to the status quo.
During Kerry's surge, he had taken a lead in the Electoral Vote Prediction, 284-254. But this week, Bush jumps back ahead by the slimmest of margins, 274-264. The difference was the recent Mason Dixon Ohio Poll, showing a Bush lead in that all important Battleground of 6 points. Coming two weeks after an American Research Group poll giving Kerry a 7 point lead in Ohio, and therefore, weighted more heavily, the Composite Poll analysis gives Bush a meaningless lead in Ohio of 0.27%. If Ohio is that close on November 2, you better hope you paced yourself and didn't take too many shots early that night.
And Ohio is why I've changed the map coloring scheme. Any state in which I predict a margin of less than 2%, I'll color white. What's the use in calling it Lean Kerry or Lean Bush when it's going to continue to fluctuate like Cesium 133? The bottom line numbers will still predict a fully allocated electoral vote, but you'll be able to get the details, as usually, at the bottom of the Composite Poll page.
I'll also start reporting a few more numbers this week from the Composite Poll. In the numbers that I highlight, I allocate the undecideds. From now on, I'll give you the raw numbers before that allocation. In addition, I'll also give you an estimated actual popular vote based on the individual state analysis. As I explain in the methodology, the Composite Poll avergage of national polls is used to help determine the margin of lead in each state. But state polls are also used. Thus, the Resulting Popular Vote includes not just national polls, but also the effect of state polls. Here are the numbers.
<a href="http://www.federalreview.com/compositepoll.htm">Here is the link with the details</a>