Polling is about determining a population's belief by asking a representative sample, so polls that don't accurately represent the population they are polling shouldn't be counted as accurate representations of that population's sentiment. Case in point (again): PPP's latest poll of Colorado.
The PPP poll out today claims Obama is beating Romney in Colorado, 49-42, but that's when they oversample Democrats by 7.3%, undersample Republicans by 2% and undersample unaffiliateds by 4.4% compared to the active voter registration totals in Colorado as of June 1.
Every other recent poll — Rasmussen, Purple Strategies, NBCNews/Marist — has had the race tied or within the margin of error. Yet somehow, after weeks of bad press, and Colorado's unemployment rising in April and May, Obama is up 7 in the PPP poll?
The Democrat flacks over at PPP must also have missed veteran Democratic pollster Peter Hart's focus group in Colorado last week where Obama was dubbed a "slick salesman, but his words didn't match his actions" by a group of undecided voters, 10 of 12 whom voted for Obama in 2008.
And for some godforsaken reason the media is reporting the poll without any mention of the serious sampling flaws.
For example, Alicia Caldwell at The Denver Post blindly wrote up the results without bothering to mention the fact that it polled a Colorado electorate that doesn't exist. Do they teach statistics in J-School? Did she not scroll down her own blog to read Curtis Hubbard's coverage of the brutal Denver focus group? Didn't that make her wonder about such a wildly divergent result in the PPP poll versus the focus group?
Even the normally solid reporter at Politico, Alexander Burns, reports the poll without even mentioning the enormous statistical problem it has.
It may be an easy post and click generator to write up poll results, but reporters should at least know what they're reporting before they click the publish button.