Baby Name Statistics, Part 4

Part 1 Part 2 Part 3

Contrary to popular belief, racism is far from dead in America. Bertrand & Mullainathan (2004) conducted an experiment on the effects of "African-American" names versus "white" names on employer callbacks from sent-out resumes. In total, 5000 resumes were sent out for 1300 positions in Boston and Chicago for jobs in administrative support. Their findings are summarized below.

1. African-American resumes yielded one callback every 15 resumes (6.45%), compared to one callback for every 10 white resumes (9.65%). This overall difference of 3.2% is significant for a p-value of <0.0001.
2. High quality white resumes yielded a larger increase in rate of callbacks (27% more) over low quality white resumes than was the case for African-American resumes (8%). These differences were present across all areas of employment, all types of employers, and both cities--with the exception of slightly less pronounced differences for employers in black neighborhoods in Chicago.

Resumes sent from addresses in higher-wealth or higher-educated neighborhoods also had a higher callback rate than resumes from other neighborhoods. More amusingly, Equal Opportunity Employers and federal contractors actually had higher racial gaps in callback than other types of employers. So much for the debilitating effects of affirmative action. Here is a relevant chart detailing the breakdown of callbacks by name used in the study:
 
People tend to react more favorably in personal interaction with people who share their name. It is possible this difference (if it is not due to random chance) is due to the person reviewing the resumes sharing the names of the applicants. I would compare the data for baby names to this list, but without knowing the ages of said resume reviewers, it would be hard to draw any good conclusions. Nevertheless, this explanation would be a pleasantly non-institutionalized racist situation--so it's probably not the case.

On the other hand, Steven Levitt of Freakanomics fame co-authored a journal article where he concluded, among other things, that "Black names provide a useful signal to employers about labor market productivity after controlling for information on the resume" (Fryer, Jr. and Levitt, 2004) (translation: blacks are lazy and shiftless (fuck him)). To be less pointed, the paper is emblematic of the fetishization of the mean that is so common in American discourse. The fact that people, for instance, with the name "Shakira" are likely to have lower socioeconomic indicators or labor market productivity tells you absolutely nothing, unless:

1. You plan to hire a lot (say, thousands) of people named Shakira, and you hire them in a way that these Shakiras are representative of the population of Shakiras in the U.S., both in their collective descriptive statistics and in the distributions all of these different indicators
2. You know the distributions of said indicators (all of them) have very, very, very low variance

Both conditions are utterly laughable: the first, self-evidently; the second would never be the case for the process that generates adult human beings, or any process that has any non-trivial level of randomness. In any event, nobody has these data, it would be a huge pain in the ass to compile them, and on balance, it's ridiculous to think that such a detailed analysis of the underlying distributions would be successful enough to give someone enough confidence to make a judgment in the case of one name--let alone any significant number of them. Also, this paper uses a set of names exclusively from California; and no, the authors never bothered to show that California is a representative sample of the population of the United States on all of the variables they used, especially over the time period (1961-2000) that they considered; and no, it probably (definitely) isn't.

Now that we've finished all four parts of this series, I can sum up everything that I've discussed in a few bullet points. To wit:

• Don't give your kid a unique/weird name. People are, at best, indifferent; if you feel the need to give your kid a unique name, you should give them one that "connotes pleasant and/or dominant qualities." That's hard to get right, so you might as well not bother.
• Don't give your kid a name associated with black culture. People really don't like those; everyone, including you and me, is at least a little racist against somebody (some, of course, more than others). And over a lifetime, that adds up.
• Do give your kid a historically popular name--a Biblical, Latinized, or Roman-derived name. In effect, you're future-proofing your kid's name. It will not in any circumstance hurt them and it may have a bunch of positive effects during their development into adults, as well as on their interactions with others.
• Steven Levitt is either intellectually lazy or he doesn't know shit about probability/statistics.

However, it is entirely possible that these bullet points will avail any prospective parent not. Kids are mean assholes, and they can and will make fun of anybody for any reason. Or maybe your kid just ends up an idiot or a jackass because of your shitty genes. So, the first and last rule of naming your child should be analogous to the first rule of ergonomics:

• If the name sounds good, go with it. This is a function of cadence, alliteration, how adjacent phonemes flow together, etc. I should stress that unless you have a unique appreciation for these things, or a poetical bent, you should only follow this rule after applying the others. Examples: Chadwick Gandall, Barkevious Mingo, Bourbon Bondurant.

WORKS CITED

Bertrand, M., & Mullainathan, S. (2004). Are Emily and Greg more employable than Lakisha and Jamal? A field experiment on labor market discrimination. The American Economic Review, 94(4), 991-1013.

Fryer, Jr., R., & Levitt, S. (2004). The causes and consequences of distinctively black names. The Quarterly Journal of Economics, 119(3), 767-805.