Saturday, 14 January 2017

Empirical Searches for Nominative Determinism

Nominative determinism is the idea that a person's name influences their life or career. The Seven Dwarfs are an example, and humorous lists have been written about people with names that suit their vocation. The world's fastest man is named Bolt, there was a Russian hurdler named Stepanova, a meteorologist named Julie Freeze, and eugenics pioneer Eugen Fischer. Considerable attention is given to a urology paper by Weedon and Splatt.

To my knowledge, the idea is not taken too seriously. However, I was recently curious about the extent to which people have looked for evidence of nominative determinism. I went on Google Scholar and found some results which I'm sharing with you now.

Most analyses of nominative determinism focus on medical professions, looking at whether specialists tend to have names associated with their speciality.

Examples of nominative determinism in medical research, with statistics of N=1. Ref.
Although this study was not entirely serious, it had a number of methodological flaws as it failed to consider the prominence of each surname in the medical fields relative to its prominence in the population at large.

One recent study was by four members of a family surnamed Limb, all doctors (two of whom named C. Limb are avid climbers). They surveyed a directory of over 300,000 medical doctors grouped by speciality, and went through the list and independently noted which were suited to medicine in general and to their speciality. The lists were then merged by consensus. They found that 1 in 149 doctors had medically appropriate surnames, and 1 in 486 had names relevant to their field. They noted:

"Specialties that had the largest proportion of names specifically relevant to that specialty were those in which the English language has provided a wide range of alternative terms for the same anatomical parts (or functions thereof). Specifically, these were genitourinary medicine (eg Hardwick, Kinghorn, Woodcock, Bell) and urology (eg Burns, Cox, Dick, Koch, Cox, Balluch, Ball, Waterfall). Some terms for bodily functions have been included in the frequency counts but are not mentioned because they are colloquial terms that may lower the tone of this publication."
Another study looked at nominative determinism in patients rather than doctors, trying to see if surname influences health. Based in Ireland, the authors looked through the Dublin phone book to estimate the percentage of people surnamed Brady (0.36%), and then tried to ascertain what fraction of them had been diagnosed with bradychardia (slow heart rate). The only data they had available was for pacemaker implants, and found that 8 out of 999 pacemakers had gone to people  named Brady. Despite the small numbers, this was statistically significant (p=0.03) relative to Brady's prominence in the Dublin telephone directory. The authors don't speculate on the causes of this, but neglected to discuss the idea that doctors may be subtly biased towards diagnosing Bradys with bradycardia.

"This increased bradyphenomenon in Bradys could be attributable to increased levels of bradykinin"

In the analysis of particle physics experiments there is something called the "look elsewhere effect" where any significant effect has to be considered in light of all the other effects that are being looked for (this xkcd gives an example of the folly), which reduces the overall significance of the interesting part. I suspect that if they had looked at what other surnames were over-represented in bradycardia diagnosis, the significance of Brady would be diminished. I think this proviso should be applied to any search for nominative determinism: look at the boring names as well as the silly ones.

There are a few other studies out there, mostly tongue-in-cheek. These do go a step beyond compiling humourous lists, but not too much further.