Thursday, July 24, 2008

Lies, Damned Lies, and Statistics

Of all the things I never dreamed of accomplishing in my life, passing a doctoral exam on statistics ranks very high. Yet, here I am reporting that I have indeed accomplished this feat. It doesn't rank as high, however, on the list of accomplishments that I will boast about at a party. It ranks somewhere between being named Safety Patrol captain in fifth grade and making a perfect score on Minnesota driver's written test. I'm just saying...

Stats satisfies a foreign language requirement in my degree plan. Seriously. No, really.

Some of you aren't surprised by this, because you've taken stats or lived with a person who loved stats (I report this in the past tense because these people tend to have an enmeshed relationship with inanimate objects that makes them less than ideal roommates). It truly is a world of its own, with a language of its own, and a citizenship of its own.

I'm no master in the language of stats. I've learned the equivalent of knowing how to ask where the restroom is in Italian or German, the difference being that I could tell you the odds of my being able to find the restroom in comparison to the population at large. I could also factor in a variable like prostate health, or, perhaps political party affiliation (one of these being far more important than the other to me at this stage in my life). Safe to say that stats doesn't qualify as one of the romance languages.

My sense of accomplishment, meager as it is, was enhanced, however, by the latest issue of Newsweek which contained an essay proclaiming this as the age of the statistician. While our essayist acknowledged the well deserved, maligned reputation of statistics in the world at large, she trumpeted the ways that stats has cleared up much of the fog created by overstatements related to dna and illness. We will owe a great deal to statisticians in the very near future. I'm sure that she's right, and I can see how some of the powerful analytic models used by statisticians are ideal for judging these types of things.

Still, I am struck by the fact that this counts as a foreign language for me. Some of my scientist friends, who "prove things everyday," enjoy some of the lingering luxuries of modernity. As my stats teacher muttered from time to time to help us cut through some of the complexity, "this is just math." As if math were this value free zone, the last safe harbor for those hoping to see the world as it actually is.

Still, at other times our teacher responded to our vacant gazes by telling us that some of these numbers didn't correspond to anything real. The formulas were kind of an invented language that allowed communication to take place across studies that varied in units of measurement. Even the number determining statistical significance was kind of made up. It could have been just as easily been drawn at another point on our vaunted bell curve.

Like all linguistic nets, it seems that stats has a certain inner coherence that makes it work. There is no place to justify the scheme outside of the scheme itself. Stats, it seems, is hermeneutical. It makes sense if you give up your need to have it make sense anywhere else.

Until I started treating it like a foreign language, I was scandalized by some of what you're asked to do in stats. Take for instance, how you are taught to treat an outlier. There is nothing like a big, fat, juicy, outlier to mess up your model, keeping you from attaining the holy grail of statistical significance. Now in theology we might call this outlier a prophet, maybe even a messiah (a highly significant finding in and of itself), but in stats you have to somehow eliminate the outlier. And there are several strategies available and acceptable for making an outlier insignificant. Sheesh. I felt like a character on The Sopranos, rubbing out all of these outliers.

Once I remembered, however, how rule based a language has to be to function as a language, I was able to relax a bit. I put my protest t-shirt, "Save the Outliers," back into the dresser drawer. After all, there are limits to every language, and none of them attain the power of precision to describe the world exactly as it is.

All that to say, "What language shall I borrow, to thank Thee, dearest Friend."

4 comments:

Redlefty said...

Congrats! Stats was one of my favorite parts of the MBA program I completed several years ago.

Have you ever seen the speech by Hans Rosling on TED.com? Good glimpse into the kinds of things you mentioned about owing part of our future to these guys.

Amy S. Grant said...

A doctoral exam on statistics, sounds like loads of fun.

I found your blog from a link on Richard's - glad to find you over here. Hope you're enjoying school! Other than the stats.

Richard Beck said...

This is one of the most disappointing things you've ever written. ;-)

Mark Love said...

Wow, RB, I've written a lot of things that have disappointed people before. I've really accomplished something here.