So you know Mark Penn? The high-powered pollster who runs the most high-powered campaign in our party? The man whose data analyses substantially impact Democratic political strategy? Well, uh... heh-heh... turns out he's making shit up.

Ezra Klein writing for In These Times:

Unlike most pollsters, Penn never releases his raw numbers, only his analysis. So we must take it on faith that his methodology is rigorous, his polls accurate and his interpretations fair. [Penn's new book, "Microtrends,"] is our first opportunity to observe, at length, how adroitly Penn handles raw data. And the answer is stunning, even to a doubter like me. Mark Penn cannot handle numbers. If this book were turned in as the final to an entry-level statistics class, Penn would not only be failed, but the professor might well retire in shame.

[...]Penn was talking about actual lefties—people who are born left-handed. Increasingly grim, I absorbed the first hard blows of Penn’s interpretative technique: “More lefties,” he enthuses, “could mean more military innovation: Famous military leaders from Charlemagne to Alexander the Great to Julius Caesar to Napoleon—as well as Colin Powell and Norman Schwarzkopf—were left-handed.” He uses the same thunderingly awful logic to argue that we’ll see more art and music greats, more famous criminals, more great comedians, more “executive greatness,” and better tennis and basketball players.

This is what statisticians—or anyone who has taken a statistics class—call a “correlation/causation error.” It is not enough to cherrypick a couple famed military leaders, notice that they’re lefties and assume that something intrinsic to their handedness caused their tactical genius. It is not enough to say that past cultures discouraged left-handedness and use that as a stand-in for discouraging creativity of all sorts. To say that Bill Gates is right-handed does not suggest that a greater proportion of right-handed people would mean more Bill Gateses. For a professional pollster to imply that correlation equals causation is like a firefighter trying to put out flames by tossing a toaster into the blaze—it bespeaks a complete unfamiliarity with the relevant techniques.

What’s more amazing is this: A page earlier, Penn argues that the rise in lefties has nothing to do with there being more lefties, and everything to do with more permissive parenting. In other words, where children used to be trained out of left-handedness, now parents “shrug their shoulders, saying it’s okay.” So not only does Penn fail to prove that lefties are genetically different in some important way, he also suggests that the gene pool is no different, and that there are as many of them around now as always. It’s a fallacy atop an error built around something that isn’t happening.

Klein has several more examples, all equally hilarious -- my favorite is where Penn declares, apropos of nothing, that “ten people with bazookas can overcome 1,000 people with picket signs, but they can’t overcome 10,000 people with picket signs.” (Counterpoint: Yes they can. They have bazookas.) It's not just that this is flawed polling, rather it's TRANSPARENTLY flawed polling, which is subsequently used to back up insane conclusions that the data wouldn't support even if it WERE any good. The mind reels.

I am trying to convince myself that this is all just some giant Karl Rove-style headfake, an attempt to convince everyone that Hillary's campaign is obviously doomed and thus win the expectations game. It better be. Because failing that, I'm force to conclude that Mark Penn is not a brilliant pollster but rather the world's greatest bullshit artist, and I'm not ready to face such a discomforting fact.

Read the whole review. And if you're really a masochist, buy the book. But I caution that I am not legally liable for any brain injuries which may ensue -- you expose yourself to this stuff entirely at your own risk...