Thursday, February 11, 2010

Diversity XXIII: The Checklist

I was reminded by the Guardian's blog post today about our last dust-up about diversity, which started with Scott's lottery proposal. (I won't link to all the significant conversation it started, but the Guardian blog post is a good overview)

At the time, I hashed out pretty clearly what my disagreement with it was, but I forgot to say the one philosophical point on which I agreed with it -- that it tried to find away to address implicit bias in the selection process. As I'm sure I don't need to say, even casting directors/hiring managers/etc. who think that they are open-minded demonstrate biases in their hiring practices. To cite just one example, a study in the American Economic Review found that blind auditions significantly increases the chance that women will be advanced or higher. The reason for this is because it takes away all of the extraneous biasing factors out of the hands of the auditioner and allows them simply to judge on the music. This is why many theaters that judge plays also judge blindly, and that's not exactly a new tactic.

But recently, I heard Atul Gawande's interview on NPR (and was reminded of it by The Daily Show's interview) about his book, the Checklist Manifesto.

The crux of the manifesto is this: every time a checklist is developed of things that you have to remember to avoid infection, and you force doctors to go through that checklist, rates of infection go down. The increase in positive medical outcomes is measurable, and at the end of the process, 80% of doctors say that the checklist improved outcomes and saved lives-- and 92% say that they want the doctor who's operating on them to use a checklist.

But the key revelation, I found, was the difference between those two numbers - the 12% of doctors who think the checklist is hogwash, but still want their doctors to use it. Partly that might be Pascal's Wager at work, but Dr. Gawande has some insight on why else that might be.

You see, even though he developed a checklist and insisted on its use, he didn't think he needed it himself. He thought the checklist was for other doctors. And yet, he too couldn't escape the data: his patients did better when he had a checklist. He didn't make as many mistakes.

The case in which checklists were used were to prevent carelessness. But I think the checklist idea might be able to address diversity... if you can figure out what belongs on the checklist.

So I call upon you, internet: say I'm the head of an organization who wants to ensure diversity in hiring practices. What would you have me put on my checklist?