One of the best courses, that ought to be standard fare in High School, is statistics. An education in what they mean, how sometimes they do in fact lie, and how they can be used and misused, is fairly critical in our quantification obsessed community.
Just yesterday, the NYT/CBS Presidential poll had Sen. Obama holding a 14 point lead. Yet CSpan/Zogby results this morning, similar to their tracking over the past week, show a difference of only 4 1/2 points, with Rasmussen consistently showing a 5 point difference.
Something is clearly amiss. Reminds me of the early, and unbelievably flawed, exit-polls in the 2004 election.
All of the three aforementioned tallies can’t be right, and at least one is totally wrong, by an enormous margin. Naturally, it’s also possible that all are seriously amiss.
Sometimes I think bona-fide news agencies ought to leave polling to outside entities, who, unlike mainstream media, don’t have a stake in the outcome, apart from establishing a track record of reliability.
Papers and broadcasters do best when they report the story, apart from making the story.
Either the folks at CBS/NYT have suddenly acquired an unusually unique talent and population sampling for data collection, or they have adopted a more ACORN-style approach to their “craft”.
Their data may in reality say more about them than the race.