Pollster is not only a fine site for poll data, but also has top-notch people writing about polls and poll methodology.
As to today's Zogby poll, which has McCain +5 and which has generated a certain amount of Chicken Littleism, Moore at Pollster says it's a very strange poll.
All pollsters, it seems, eventually find themselves with what Andy Kohut once referred to as "loopy" results.
This time, it's Zogby's turn to confuse the masses. His latest Reuters/Zogby poll, based on a sample of 1,089 "likely voters" drawn from listed telephone numbers, conducted Aug. 14-16, 2008, shows McCain over Obama by 46% to 41%.
What makes the poll especially strange, even after taking into consideration that it has screened out whoever they consider to be unlikely to vote, is that it only includes people with listed telephone numbers.
Folks, this is a very odd method. Virtually all pollsters use some form of random digit dialing in order to make sure that numbers are included that are unlisted. Your sample is clearly incomplete and skewed if you only use listed numbers. And of course this doesn't include cell phones which are all unlisted.
So, while there is evidence of some drops in Obama's numbers, the Zogby poll is quite suspect. Furthermore it is out of line with other polls and has a pretty substantial influence on the poll averages.
And what is its impact? Well, if you take all the polls listed in the current realclearpolitics average except for the Zogby poll, the average difference without the Zogby poll is +2.5 Obama. With Zogby, it's a +1.2 Obama. In other words, the odd Zogby poll all by itself cuts Obama's lead in more than half, as it reduces the lead by 1.3.
Now, I do think Obama has to go after McCain harder than he has and he needs a vp pick that will be a fighter. But let's not get deflated by yet another weird Zogby poll.