Hacker News Reader

Comment by a_bonobo | Hacker News Reader

Thanks for the paper!

So only 4 US ethnicities were tested: African Americans, Caucasians, Chinese and Latinos. That feels extremely broad - especially "Chinese" and "African Americans" combine many, many different genetic groups. Edit: The Chinese government, for example, recognizes 56 ethnic groups, which may or may not genetically be distinct.

>The subgingival microbial community was able to predict an individual’s ethnicity with a 62% accuracy, 58% sensitivity and 86% specificity.

Wait, that doesn't seem very good - I could flip a dented coin and get similar results?

View on HN

What's with the popular usage of latino or hispanic as a separate race?

From wiki: "The United States Census uses the ethnonym Hispanic or Latino to refer to "a person of Cuban, Mexican, Puerto Rican, South or Central American, or other Spanish culture or origin regardless of race."

Also, remember that there are many white/caucasian people in Mexico and other latin american countries that also are hispanic. Ricky Martin is an obvious example of a white Puerto Rican, for instance.

I am not saying a mestizo should necessarily be classified as "white", but there seems to be an idea these days that hispanics are a separate race, which is incorrect.

In general "race" is a bogus term. Usually, when people (especially Americans) use the word race, they actually mean something closer to ethnicity.

> What's with the popular usage of latino or hispanic as a separate race?

Mostly that came from the US government deciding to create a pressure group by lumping various central and south american nationalities together.

However, US hispanics do form a group which is noticeably different from, say, US whites. Many are indios. In particular, we get a huge amount of mexican immigration, but it's mostly the poor mexicans who choose to come. White mexicans are generally doing pretty well in Mexico.

> there seems to be an idea these days that hispanics are a separate race, which is incorrect.

This depends on what level of population structure you want to label "race". It's not difficult to distinguish US hispanics from US whites with a genetic test; that's a good enough marker of "race" for me. As you point out, the category has a lot more mixing than you might expect in another racial group, such as Germans, but it's still a workable category.

The chinese government recognizes many ethnic groups, but Chinese are overwhelmingly Han, and less genetically diverse than Caucasians. Why are you more worried about the chinese group than the caucasian group?

> I could flip a dented coin and get similar results?

No idea what you're trying to say there, but (a) there's a fun paper out there essentially showing that it's not possible to bias a coin; (b) in the absence of demographic information on the community, a biased coin is no better or worse than an unbiased one; and (c) yes, you can obviously do enormously better at distinguishing american whites, american blacks, american chinese, and american latinos, than those figures. But chance would predict ethnicity with 25% accuracy, so they're doing a lot better than that.

Reply on news.ycombinator.com