My Moment of Self-enhancement

I worked my way up from nothing to a state of extreme poverty.

I work for Brown University, which is a reputable institution. Though Brown belongs to the Ivy League, it has less name recognition and less money than its peers. According to the latest rankings posted by the Times, Brown is 49th in the world, a rank it shares with the University of Manchester and the University of Peking. Overall, the English universities seem to be doing rather well in this Times-of-London ranking, but that's perhaps another story of self-enhancement. My own story of self-enhancement is that I can claim to have risen from lower ranking universities to higher ranking ones as I moved from undergraduate education (University of Bielefeld) to graduate school (University of Oregon) to a proper job (Brown University). This autobiographical note makes me feel good because it suggests steady advancement against the head wind of regression.

Statistical regression is a phenomenon that occurs when two variables are imperfectly correlated with each other. Instances (people, objects) that have an extremely high (low) value on one variable tend to have a somewhat lower (higher) value on the other. Francis Galton discovered and described regression to the mean in his studies of hereditary genius. Hereditary it may be, but extreme genius is followed by more modest talent in the next generation, much like the tallest men have sons who are a bit shorter on average.

The rank of the degree-granting university and the rank of the job-offering university are also two variables that are imperfectly correlated. So we should expect to see the graduates of the most august universities taking jobs in lesser places. At the lower end, graduates tend to land jobs at places that are ranked more highly than the places where they studied—if they get jobs. With the University of Oregon being a great but not an elite University, my statistical expectation was to land a job at a mid-range university rather than at one where the air is so thin. But I did the latter, and hence my moment of self-enhancement.

Now take a different perspective, looking back in time. If a university is ranked very highly, you'd assume that its faculty graduated from programs that are ranked a little less highly on average. From that perspective there is no surprise, and no reason for self-congratulation. The difference is exactly what one would expect as a result of statistical regression. To induce a feeling of self-enhancement, I have to project myself back in time and look forward to my prospects. With that perspective, I can feel pleased to have swum anadromously like a salmon. Merely looking back in time—which is easier to do—I can only shrug at the statistical inevitability of my mediocre origins.

There is evidently an asymmetry in bivariate relationships. As another example, take mating. For the moment, accept the simplifying assumption that men and women are ranked on a single dimension of desirability as a partner. Also assume that most, if not all, are interested in mating (marrying) up. Half of them will manage to do that, but will be stuck with a partner who mated down. The probability with which a person will mate up is predictable from his or her own rank. Those with the lowest rank are most likely to mate up - if they mate. Since low ranking men are least likely to mate, those who are most likely to mate up are low-ranking women. Those who are most likely to mate down are extremely desirable individuals (note: if the ranks of two mates are highly correlated, the expected difference between own and partner rank will be small).

In the arena of mating, the asymmetry is this: Since extremely high-ranking individuals will mate with somewhat less highly ranked individuals, these less highly ranked individuals are those who have mated up, although their own ranks are already pretty high, though not extremely high. From their perspective, it is amazing that they were able to mate up, when their statistical expectation was to mate down. If they maintain their personal perspective, they can retain their sense of wonder and gratitude toward the providential universe; if, however, they take their partner's perspective, they can only shrug at the damn statistics. Luckily, one's own perspective is the easier one to take.

Back on campus, there is another asymmetry between professors and students. At the most highly ranked universities, the professors are likely to have studied at somewhat less selective schools on average. This means there are many among them who are teaching students who have succeeded where they, the professors, did not. These students won admission to a most selective school. From this perspective, it seems that the professors might feel inferior to their students, in a Groucho Marx sort of way. Viewed from the alternative perspective, however, the professors might point out that they were among the few who won jobs at institutions that would not have accepted them as students. In other words, they went upstream, anadromously, against the pull of regression. By contrast, the students' expected place of employment will be institutions of somewhat lesser distinction.