Blog: Expanded Universe

Reverse alphabetical order

One of the sillier biases in today's world is that people whose names come earlier in the alphabet might be favored in a variety of situations, theoretically because they come earlier in alphabetically sorted lists (school rosters, some author lists), and thus get more options and more attention.12345678910

The effect is weak, if it exists. But I was thinking about this in high school when I listed my yearbook dedications in reverse alphabetical order in an attempt to counteract the bias, however microscopically. At least, I think I was, but maybe I was just being contrarian for the sake of it.

As far as I can remember, I've only met one other person who has done something in reverse alphabetical order for plausibly similar reasons. It was a special moment of kinship.

  1. I'm fairly privileged along this axis. I got to be first of seven authors on that one cryptography paper I helped with in 2022, despite being far from the main contributor.↩

  2. Yuret, 2019: economists with alphabetically earlier surnames are more likely to become full professors, but the effect is not significant; also, no such effect was found for mathematicians, even though both fields have similar author ordering conventions. Also compare earlier work, Yuret, 2016, which found no effects at all.↩

  3. Urbatsch, 2014: political leaders's names tend to be alphabetically earlier than those of their constituents↩

  4. Shevlin and Davies, 1997: authors with earlier surnames are cited more often, but the effect becomes statistically insignificant when controlled for the natural nonuniform distribution of surname initials over the alphabet↩

  5. Huang, 2014: first authors of scientific papers with earlier surnames receive more citations (but not non-first authors)↩

  6. Einav and Yariv, 2006: economics professors with alphabetically earlier surnames are more likely to receive tenure in top institutions and win awards (but the effect mostly goes away considering all institutions instead of just top ones)↩

  7. Cauley and Zax, 2018: men with alphabetically later surnames get better education and jobs↩

  8. None of these even deal with the effects of having a name that is more or less stereotypically associated with some demographics — which seem very plausibly larger, and which can easily confound any effects one might want to attribute to alphabetization, since names from different demographics are distributed differently over the alphabet.↩

  9. An amusing part of reading several of these papers is seeing how they describe the A1Z26 encoding that every puzzlehunter knows in fancy scientific terms in order to explain how they do statistics on the names. Yuret, 2009: "numbers are assigned to letters in an ad hoc fashion". Urbatsch cites Grannis et al., 2002: "letters will ... be treated as digits in a base-27 (septemvigesimal) system" — though Grannis et al. appear to use a base-27 system only in a tiny subcomponent, to produce a jury-rigged order-invariant hash of (first name, last name) by summing them, since their paper is about matching names between different data sources and they wanted to ignore errors where the two names were swapped.

    Sometimes the logarithm of the A1Z26 is taken instead. I think the theory is, the important variable is how often you're alphabetically first in a random sample of yourself and several other peers, in which case the difference between A and B is more important than the difference between Y and Z. (Exercise for the reader: Mathematically formalize this and decide whether the logarithm is the correct transform.)↩

  10. This is more of a bit than actual research, in part fueled by me randomly remembering that Elicit exists. You shouldn't believe a claim like this just because I linked to lots of papers; I aggressively simplified each paper's reported effects, plus publication bias and all that.↩