The U.S. Census Bureau recently released lists of names occurring 100 or more times in the 2020 Census returns.
Michael is listed as the most common first name, reported by approximately 3.4m people (~1 in 100).
For each name, there’s a count of males and females. ~0.2% of people named Michael reported their sex as female. So, in the case of Michael, if we used name to infer sex, we’d probably have a margin of error.
That name is not a one-off. In fact, the first 2,113 names have both sexes recorded against them. Some of these might have had small adjustments, so the underlying data might look slightly different. As it stands, the first name that only has one sex is Lynnette, ranked 2,114 with just over 12k people recording that name, all female. The first all male name is Geoff, ranked 3,469 with just under 6k people.
Interestingly, some names are balanced. For example, names like Casey and Quinn are close to parity; that is, they each have more than 40% male and more than 40% female. So it’s near impossible to accurately use those names to predict sex. There are others like this: 60 names have more than 10% of each sex.
This is U.S. specific, of course. Other countries may have different profiles. Some languages have sex built in, which makes it a bit more deterministic.
In general, accuracy will depend on which names our customers have. In many cases, it will be difficult to work out which individual records are wrong. So, first names are very likely to be unreliable indicators of sex.
Disclaimer: The info in this article is not legal advice. It may not be relevant to your circumstances. It was written for specific contexts within banks and insurers, may not apply to other contexts, and may not be relevant to other types of organisations.