Posts Tagged ‘statistical modeling’


Also from Crossen

May 26, 2011

Well, really Benjamin Barber via Crossen.

Contrast these two questions:

1. Do you want a drug rehabilitation center in your neighborhood?

2. Do you think that the community needs drug rehabilitation centers, and if so, would you accept one in your neighborhood if you were persuaded that the policy process by which the locations were chosen was participatory and fair?

Pollsters assume that people can only answer questions of private preference.  If people are constantly asked to evaluate public polices in terms of their prejudices, they unlearn the art of civic judgment.

I don’t really blame polls for the questionable quality of our collective civic judgment, but I do find it persuasive that our questions are probably incorrectly oriented, that we would be better of if we’d explicitly require people to separate out what they believe to be best for society from what they believe to be best for themselves.



The real “other side”

May 26, 2011

Every now and then, there are rumblings about passing a law that would prohibit unions from using member dues to make political contributions without the written permission of the member.  For those of us who’ve been exposed to the portrayal of unions as corrupt special interests (think The Wire, Season Three), this makes some sense.  I admit to thinking that I could see where the impulse for such legislation might come from when I first heard of it.  For me, the two sides were: constrain political contributions or don’t.

Flying Whale, not surprisingly, was able to put it in context much more quickly, responding, “Are we going to require shareholders to sign something before corporations can make political contributions too?”  For Flying Whale, the two sides were: constrain both opposing powers or neither.

Initially, I was confined to a narrower scope, that of limiting union corruption, when imagining the other side of the argument.  Flying Whale was working from a broader and, I think, more robust understanding–that the other side was really about keeping opposing powers balanced.

The conversation reminded me that I really do think it’s a skill to be able to see the real “other side,” not the one embedded in the frame someone else is using.

Cynthia Crossen’s book, Tainted Truth, of which I’ve admittedly only read a few chapters, really made this point for me.  In her discussion of polling, she explores how poll results are affected by question wording.  Old news, right?  But I was surprised by how difficult it was for me to spot the less egregious slants.  For example, in Chapter Five, Crossen explores the public opinion polling that surrounded the Clarence Thomas/Anita Hill scandal/accusation.  Regarding a question from a New York Times/CBS News poll:

“Some people say Anita Hill’s charges should not be taken seriously because she did not make them years ago at the time she said the incidents happened.” (So far, so good.  That, indeed, was a popular argument against Anita Hill’s case.) The question continues: “Other people say the charges should be taken seriously even though they were made for the first time just recently.”  This second sentence is supposed to be the other side of the coin–the reason Anita Hill should be taken seriously.  Instead, it simply restates the negative point–she took a long time to complain.  But what would the results have been if the second part of the question had read, “Other people say the charges should be taken seriously because women sometimes have reasons to delay reporting such behavior?”

Once Crossen points it out, it’s so clear.  But just the “Some people say…other people say” structure had me fooled into taking it as an even-handed question.

Knowing that you’re susceptible to being duped certainly helps, but I’m finding this to be a slow skill to acquire.



The people you know…

August 24, 2009

The statistical modeling folks over at Columbia are comparing, by age, people who support gay marriage with people who actually know someone who identifies as LGBT.  While I’m not surprised that there might be a relationship between knowing an LGBT person and supporting gay marriage, I am astonished that more than 40% of folks in their twenties claim NOT to know an LGBT person.

Obviously, folks who openly identify as LGBT aren’t evenly distributed throughout society, but I still find that percentage to be startlingly high.  I’ll be curious to see if those percentages look very different five years from now.