Yesterday the Gallup Tracking Poll showed Obama ahead by 9 points (49-40). Earlier today (Monday) that dropped to Obama ahead by 8 points (48-40). Both were of registered voters.
Now Gallup/USA Today released a new poll of LIKELY voters, showing a 4 point lead for McCain, his first lead in any major poll since early May.
The switch from registered to likely voters explains most of the difference.
Yet this jumping around does not inspire confidence in Gallup:
Republican presidential candidate John McCain moved from being behind by 6 points among "likely" voters a month ago to a 4-point lead over Democrat Barack Obama among that group in the latest USA TODAY/Gallup Poll. McCain still trails slightly among the broader universe of "registered" voters. By both measures, the race is tight.
The Friday-Sunday poll, mostly conducted as Obama was returning from his much-publicized overseas trip and released just this hour, shows McCain now ahead 49%-45% among voters that Gallup believes are most likely to go to the polls in November. In late June, he was behind among likely voters, 50%-44%.
Among registered voters, McCain still trails Obama, but by less. He is behind by 3 percentage points in the new poll (47%-44%) vs. a 6-point disadvantage (48%-42%) in late June.
Results based on the survey of 791 likely voters have margins of error of +/- 4 percentage points — so McCain's lead is not outside that range. Results based on the survey of 900 registered voters also have margins of error of +/- 4 percentage points.
Gallup editor Frank Newport tells Jill that "registered voters are much more important at the moment," because Election Day is still 100 days away, but that the likely-voter result suggests that it may be possible for McCain to energize Republicans and turn them out this fall.
Note that Obama's lead among even registered voters has dropped precipitously from 9% to 8% to 3% in data from Gallup added to Real Clear Politics since midnight Sunday night. Indeed, both the 8% and the 3% results were for polls taken the same three days surveying apparently different samples of the same population, registered voters — strange but not surprising, given the low response rates for most public opinion polls these days.