Is Obama's lead melting? Not in Missouri, where a Research 2000 poll for the St. Louis Post-Dispatch shows Barack Obama with a 48-43 lead. Obama had also led Research 2000's only previous poll of the state, taken in January, by a 5-point margin.
I continue to remain relatively skeptical of the the importance of Missouri in the electoral math, and other recent polling has shown it as a McCain state. Nevertheless, Barack Obama does have a fairly robust set of scenarios for electoral victory, with Missouri playing a role in certain of those. The most salient is probably if McCain makes a VP pick that gives him a state-specific advantage in Ohio (Rob Portman), Michigan (Mitt Romney) or Pennsylvania (Tom Ridge). If McCain won Ohio, for instance, and the map reverted to the Bush-Kerry version from 2004, Obama would reach 270 electoral votes by winning Missouri and Iowa. A more radical scenario would be if Obama lost both Michigan and Ohio, which would place him at 232 electoral votes if he won the other Kerry states. Even if he also won Iowa, Colorado and New Mexico, neither Missouri and Virginia would be enough to put him over the top. But winning both of those border states -- Missouri and Virginia -- would be. In the simulated scenarios where Missouri turned out to be a decisive state, it was fairly often paired with Virginia, as well as North Carolina and Indiana.
What to make of those national polls then? In addition to the new Newsweek poll that showed Obama's lead reverting to 3 points, the Rasmussen daily tracker now shows Obama's lead at just one point (and the race dead-even if learners are not counted). Obama retains a 4-point advantage in the Gallup tracker.
Our tracking graph, which is based entirely on an inferential process seeing how the polls move relative to another, shows Obama running about a point-and-a-half off his peak numbers, which occurred in late June. It furthermore concludes that the last couple of days' worth of polling implies a very tight race -- perhaps Obama by a point or so. But it is not yet convinced that the couple of days' worth of mediocre polling gives it enough information to warrant a more serious downgrade in Obama's stock.
Whenever a candidate declines in a poll, there is sort of a Rorscharch blot process that takes place in which every particular interest group attributes the decline to its favorite cause: progressives to Obama's flip-flop on FISA, conservatives to Obama's losing the Wesley Clark dust-up, etc. But the reality is that, if it's before Labor Day, and you want to get the attention of the 80 percent of the public that does not read blogs and does not watch cable news, you either have to hold a convention, name a Vice President, or do something phenomenally stupid. I don't think any of the current controversies rise to that level (if any does, it's the Phil Gramm thing). The respective candidates' favorability ratings have not really moved at all, although it looks like some of Obama's Very Favorable ratings have migrated to Somewhat Favorable.
Finally, our condolences to the family of Tony Snow.