Not very much, potentially-- if you rig up the charts that present your data just the right way. Take a look at Poblano's latest diary, for instance. http://www.dailykos.com/...
Here is the Poblano diary we're discussing: http://www.dailykos.com/...
Let's say the Republicans were worried about a serious Democratic challenge from Barack Obama in the Presidential election this year. Say further that they were prepared this time to actually rig the election if they had to-- they didn't actually do it before, because they weren't willing to go that far yet or because the exit poll results would have exposed it, but this time, they're ready to do something to undermine the polling outfits.
They can't actually infiltrate every polling outfit in the country, because the bribery or intimidation or blackmail it would take to do so would be so large-scale that it would be hard to keep the effort entriely under wraps. But, they are prepared to undertake a Plan B- use the entree they've gotten at undermining the mainstream media to start a new conventional wisdom: that public opinion polls in general are almost useless, due to lying respondents and intentional pollster bias and so forth, so that an accurate poll is really just a fluke occurence. They really only have to convince all the Joe Blows out there, who don't know or think too much about statistics, polling, or pollsters in the first place, to accept it, too. All they have to do is shake people's faith in the polling outfits, and then produce some pollster that they can make confirm the (fake) election results.
What form might an effort like that take? Well, for one thing, you might have some guy who you've gotten to be well-known on a website like Daily Kos post something like Poblano posted today.
Poblano's post purports to compare two pollsters, Rasmussen and Survey USA. He includes the dates of their respective polls in his charts- but only in tiny black letters. Then he starts his comparison with a Survey USA poll from 2/27/08, side-by-side with a Rasmussen poll from 3/18/08. In colored figures, he lists the poll results for these polls showing a 29-point-spread favoring McCain in one poll, but a 20-point-spread favoring McCain in the other pollster's poll, listing the figures side-by-side. But the polls aren't even from the same month, and were actually taken about 20 days apart.
Then he goes down the entire chart doing this, listing spreads side-by-side to look like similar polls show wildly different spreads, when the polls listed side-by-side for comparison are typically days, and even weeks apart. Then at the bottom of the chart he concludes with an average of all these numbers- following 3 Rasmussen polls from a few days ago as the last 3 Rasmussen listings, while the most recent Survey USA results are three polls from 3/15!!!
But just a quick look at these Gallup national polls show you that since 3/15 is when everything changed for Hillary-v.-McCain and Obama-v.-McCain match-ups!! So comparing an average for one pollster that leaves out this information with an average for one pollster than includes it is, of course, going to make a difference as to what kind of average you have to compare-- just like starting the data examined in the chart weeks earlier for one pollster than for the other, like Pobalano did, might!
For comparing the two pollsters, Poblano's side-by-side listings of individual polls is misleading, because the polls compared are on totally different dates, days or weeks apart, when things can change. For comparing the two pollsters, Poblano's bottom-of-the-chart averages are misleading, because the data for one pollster both starts earlier (when the Democrat was doing better), and ends later (after the Democrat was doing worse).
That diary is misleading and if people were trying to undermine people's feeling that they could rely on polls- which might belie eventual elections "results"- to tell them what's going on, those charts and that diary entry are just the type of thing they might write.
I don't know why that guy gets so much stuff in his tip jar, but then, Internet polls can be ganged-up on, too- like when Atrios asks his readers to go to CNN's polls, and then we have CNN polls telling us that 99% of respondents don't favor George Bush's Iraq policy, or whatever.