Poor Polling?

So, this week we have some suggestive evidence that thanks to racism, polls may overstate Obama's real advantage by as much as six percentage points, and some suggestive evidence that thanks to cell phone users, polls may overstate McCain's advantage by as much as 2.8 points. (Though Nate Silver offers a caution about the former.) If both of these things are true (and we can't say with any certainty that they are), the net effect is an overstatement of Obama's support by 3.2 percentage points. I therefore haul out some old advice: (1) take all polling with many grains of salt; (2) look at state-by-state polling data, not the all-but-meaningless national numbers; and (3) look for multiple-day rolling averages to filter out more of the statistical noise.