I was reading through the new Brookings Institute study on the value added by certain colleges. Their report makes very clear what steps were followed in the analysis of their data. Basically, get the average lifetime earning of college students with certain majors and compare that number with the average lifetime earnings graduates of a given college. The result is an estimate of the value of going to that college, the amount of boost that college will supposedly give to your lifetime earnings.
There are many angles where a suspicious and doubting person like me can approach such a study. This same general idea is behind quite a few "value added" studies, including attempts to put a figure on particular teachers as to their ability to add value to students in their classes.
What struck me as I looked at various colleges and the figure the analysts associated with them was how I was being sucked in by the numbers. Here is a college estimated to add $50,000 to my lifetime earnings and here is one that will add $150,000. A doubter can be told, honestly, I think, that these are just the way the figures fall out, that nobody fudged the data for or against any college.
One way that statistically grumpy people might begin to try to get unsucked by the data is to ask about variances, also about extremes. The main approach of correlational analysis is to use what I wrote: averages. So, just for my sake, what are the five or ten lowest lifetime earnings of graduates of that college in that major?
In matters of education, it doesn't hurt to keep in mind that I myself might not do my homework in that major. I might be under the spell of a little redheaded girl, my father and I might be in a continuous battle over what I should major in. Despite the difficulties of conceptualization and of measurement, there is still a large component of my life that rests with me alone: my feelings, my interests, my diligence, my drinking, my ability to eat right, sleep right and clearly discern what my internal compass directs me to do.
Numbers are popular these days and our machines can churn out more numbers in one minute than you can digest in a week. I think of the book "How to Measure Anything" by Douglas Hubbard. Mr. Hubbard does a good job explaining how to go about gathering evidence on just about any question. Of course, that type of specialist gets paid big bucks to render an opinion on slippery matters, such as "Is the morale of my employees rising or falling?" An executive, a leader, a politician or any person responsible for a collection of people needs an average, a summary, a general feel for the tone of the group. The morale or lifetime earnings of an individual are of little help. But it is helpful to remember that in many cases, averages and other summary variables are approximations, estimates and as such, are myths. They can be helpful myths but they are human constructions.
--
Bill
Main blog: Fear, Fun and Filoz
Main web site: Kirbyvariety
