Empiricism: Abstract Nonsense Part II?

There are two types of friends-- those that tell you the truth, and those that tell you what you want to hear. The latter are enjoyable to be around, but in the long run are probably not so good for you.

You can learn a lot about society from its conceptual metaphors, and they are more influential than you might think. The common expression that numbers don't lie not only personifies data, it characterizes data as that first type of friend, the rare kind that tells it to you straight. Many a college admissions essay, my own included, are essentially an eager theme-and-variations on that sentiment. With my wordy ode to empiricism, I got into college and learned all about data, got to know numbers really well, only to discover that maybe they're not exactly what I thought.

At work I have the onerous task of trying to determine the number of jobs that can be attributed to digital technology, and to analyze the trends and the potential for future employment in the information economy. So lately I've been deep in the databases of the Census, BLS (OES, CES, BDM, CPS etc.), BEA, OECD, even AARP. I've fought the good fight with Excel.

The result? Whatever you want to hear. I can't post the graphs up here, but widely disparate trends and figures can come out of some straightforward analysis. As an exercise, go to BLS and use the data to justify that digital media-related employment is growing. And then use the data to justify that it is in decline. Anything I want to tell you, I can tell you, and back up with statistics and a graph. The problem is, I don't know what I want to tell you. I want the data to tell me what is true!

I ran into similar frustrations studying measurement and metrics at the National Center for Education Research and working at the Budget Office.

I know the surface-level explanations. Sometimes data collection is shoddy. Sometimes metrics are poorly designed, methods are lacking, people are careless, tools are faulty. Sometimes, goes a twist on the expression, numbers don't lie, liars use numbers. But I think the issue is more fundamental.

Data implies codification, representation, and reduction. Numbers, like I mentioned in the last post, are abstract-- so bogglingly abstract, that when you think about it, our haste to represent the world with them is remarkable. Empiricism, reductionism, and data are a critical part of understanding our world, and I know no good way around them, but I think they are ingrained too deeply, and too rarely questioned by the very people who use them most.