Can Cultural Anthropology Scrogg Population Genetics?

James Mullooly invented the word Scrogg, meaning something along the line of “anthropologist who catch geneticists playing fast and loose with the data.”  In my experience, Scrogging is fairly easy to do on open-source turf of the Biological Sciences journals where there are often places for comments.  These comments are typically reviewed by editors, and while not strictly “peer-reviewed,” would in my mind contribute to the academic record of aspiring academics.  The editors I have had contact here and here were fair and open to critiques of articles which were well-done technically, but missed out on a more social scientific perspective, as do many articles about human population genetics.  The editors were  quick to respond to me—it seems the biological sciences are much quicker than the social scientific journals at making editorial decisions.

 

I got away with two Scroggings because population geneticists wrote about a group that I knew just a little about, the Mlabri of Thailand.  But I knew enough to know that lab-based geneticists who missed key points, and did not cite standard ethnographies.  I am sure that there are many sociologists, anthropologists, and others who have similar experiences with geneticists writing about groups with which they are familiar.  What I would encourage you to do is to go into Google Scholar, and PubMed and search for genetic studies of ethnic groups you are familiar with as a result of your field and library work.  Then evaluate carefully how the data (typically blood samples) was handled on its way to the lab.  Does the “sample” reflect social relations on the ground?  Is it consistent with historical, geographical, and anthropological data with which you are familiar?  If not, the article deserves a carefully written Scrogg highlighting how conclusions might have been different if anthropological data were also considered.

Scrogging by cultural anthropologists should result in a number of well-reasoned postings.  More to the point, it is hoped that geneticists will be more careful about how they handle data, and editors more consistently solicit ethnographers and cultural anthropologists as peer reviewers.  While scrogging, of course, be as narrow, precise, and gracious as possible given the circumstances.  The point is not to embarrass, but to highlight the importance of cultural anthropology and qualitative data in evaluating populations.

19 thoughts on “Can Cultural Anthropology Scrogg Population Genetics?

  1. justaguy

    This strikes me as an example of where a more critical understanding of race would be helpful for geneticists. That is, the geneticists seem to have assumed that the Hmong were fungible enough that any sample of Hmong genetics could be used to represent the groups they were talking about. And your criticism of that sampling issue is something which should be obvious to any geneticist, and I assume it is when you call their attention to it. But the idea of race as a stable and monolithic category is so pervasive within our culture its easy for people who should know better to make those kinds of mistakes.

  2. Tony

    It “should be obvious,” but apparently is not. There is a level of abstraction in Genetics which is part of their, um, “culture.” When they look at a blood sample, they see a “population,” without always considering conditions under which the sample was collected. This apparently is difficult for them to do, or they wouldn’t have made this mistake twice in print. In neither case, either, was it caught in peer review, or the editorial process. With all those co-authors, there were at least ten geneticists looking over each paper, and none raised a flag. As you are finding out on the other thread, there really really is a different way of thinking in the world!

  3. the general point you make is valid, and i’ve made it myself (i.e., “this paper should have been sent to area studies specialists, etc.”). that being said, there is also the secondary problem that a lot of area studies specialists (i’m bracketing historians-of, ethnologists-of, etc.) don’t understand statistics well enough to comprehend tha their critiques are not relevant to the specific question/result at issue in a historical/pop gen paper.

  4. But the idea of race as a stable and monolithic category is so pervasive within our culture its easy for people who should know better to make those kinds of mistakes.

    no population geneticist believes this. if you can find me a population geneticist who believes, tell me who they are, and will email them to confirm they actually believe this ludicrous proposition.

  5. justaguy

    @razib,

    My point wasn’t that any population geneticist would believe that – I would be very surprised if any did. But per Tony’s critique, the geneticists in the article acted as if that were true. That is, instead of using a Hmong sample specific to the geographic area they were looking at, they picked a generic “Hmong” sample which was taken far away, and acted as if it were representative of the Hmong group they were looking at. If that critique is valid, that would be acting as if Hmong genome were fungible.

    I suspect that whoever did that, didn’t do so consciously noticing the discrepancy between the genetic sample they were using and the question they were using it to answer and say, “well, a Hmong is a Hmong is a Hmong”. And I recognize that scientists are human, and mistakes happen for any number of reasons. I don’t see the need to attribute to malice or incompetence what could easily be explained by a sleep deprived grad student.

    My point was that a more critical attitude towards race might prevent a mistake like that from being overlooked by the various co-authors and reviewers.

  6. well, a Hmong is a Hmong is a Hmong

    this is an empirical question. after further research it turns out that these researchers probably didn’t make an error of result, though their method was criticizable (the pan-asian data set as 50,000 markers are a much larger population data set).

  7. also, for the purposes of ‘hypothesis testing’ you often run ahead of the data. being wrong or being right is fine, as long as it’s clear and distinct. i’m moderately skeptical of hypothesis testing, but it’s going to take 20 years for this to go out of vogue as the fundamentalists for hypothesis testing retire or die off. lots of scientists who might have provisional views on a conjecture need to strength their position to make it publishable.

  8. hey, does anyone read “Cultural Anthropology: The Human Challenge”?

  9. Tony

    The social boundaries of “hmong” are fluid in Southeast Asia. There are also sub-groups which in turn has implications for endogamy and exogamy. These sub-groups include clans, dialectical differences, and residential differences. Then there are also of course norms regarding marrying into other groups. The problem is as Guy points out, that these sub-categories are often small, and not detailed well in the “sample” which often involves showing up in a rural village, offering incentives, and a human subjects form to sign. Little attention is paid to local conditions.

    I don’t know about “Cultural Anthropology: The Human Challenge.” Who wrote it and what is its thesis?

  10. Tony

    @Razib
    “But the idea of race as a stable and monolithic category is so pervasive within our culture its easy for people who should know better to make those kinds of mistakes.”

    I get it that few if any population geneticists believe this. But as you note, the term “race” has a lot of other meanings in modern culture and “discourse.” Maybe it is time for an alternative term? Has there been any discussion of such a switch in genetics?

  11. Maybe it is time for an alternative term? Has there been any discussion of such a switch in genetics?

    yes. pop gen ppl naturally use the world “population” or “subpopulation.” the word “race” usually pops up in medical genetics contexts, where the US census classifications have had an inordinate influence (because NIH money is tied to gov. forms i assume?). but really, you basically mean what geneticists would call race before 1970….

  12. btw, the problem is not abandoning the word race. the problem is that it is now common among cultural elites in the west to assert that “race has been disproven by biologists.” this is true insofar as race is a categorical/platonic construct. but, you get totally stupid assertions regularly of the form: “two people of differences ‘races’ might be more related than two people of the same ‘race.'” when say race does not exist these sorts of assertions aren’t ridiculous. but they’re totally false for non-trivial levels of inter-population genetic difference. populations are reifications of relatedness, so by definition it’s false.

  13. Tony

    I think that there are plenty of people who are not in the cultural elite who equate race with genetics and/or census categories, just as there are people int he elite who miuse the verb “disproven.”. The US Census and the way the word has been used in law for 200 or 300 years also leads many to equate race with genetics, despite what geneticists, anthropologists, or sociologists may assert.

  14. Tony

    @justaguy
    Razib thinks social scientists don’t do statistics, just like we think that Geneticists can’t do social theory or ethnography. Perhaps we an prove each other wrong?

    @Razib
    I share your skepticism about the all-purpose nature of hypothesis testing. There are other ways of “knowing” being developed.

  15. Razib thinks social scientists don’t do statistics, just like we think that Geneticists can’t do social theory or ethnography.

    i didn’t say that. i’m saying that the sort of people who do history or ethnography often lack perspective on questions of statistical power, representativeness, etc. obviously lots of social scientists don’t have a problem with statistics.

    I think that there are plenty of people who are not in the cultural elite who equate race with genetics and/or census categories, just as there are people int he elite who miuse the verb “disproven.”

    the people who are not in the cultural elite don’t matter that much on these issues. they’ll follow the elites’ lead. no point in talking about them. these are people who think there is a ‘hispanic race’ because the nixon administration created the category in 1970 for political reasons. this is the sort of detail which is interesting, but not my primary focus.

  16. justaguy

    I freely admit to not knowing that much about statistics – although I did get an A in my 12th grade stats class. And that is fairly common among cultural anthropologists I know, although it is by no means universal. Getting a basic competency in statistics is one of my goals for the coming months – although its made difficult by the fact that it has no application whatsoever to the research I do.

    My problem with Razib using the lack of quantitative analysis in ethnography as a critique is that he assumes that since stats are central to fields he’s familiar with, fields that don’t use stats aren’t valid. And he’s certainly not the only person who feels that way – its common with people in quantitative fields. But whenever I get into a conversation with a quantitatively inclined person about how I could use stats to answer the questions I’m asking with the data that’s available to me, I never arrive at an actual answer.

    And that comes from anthropology’s history of looking at small scale societies where there just wasn’t that much to count – sociology, which started out from similar theoretical foundations but with a focus on Western societies, has a robust quantitative element. It also comes from our general orientation away from simplifying the messiness of our data – as a philosopher professor said to me in exasperation, “You anthropologists, for you nothing is ever an example of anything!”

    But different questions require different tools. A good friend of mine is a data scientist at a social networking app. He’s hiring an ethnographer, because he can see trends in usage, but has no idea what the social or cultural factors are which are driving it (e.g. Why are they big among 14-24 year olds in specific cities in Guatemala?). And those qualitative elements are central to developing their business.

    If by prove him wrong you mean prove that anthros and geneticists can productively benefit from each other’s work. Sure, quantitative and qualitative researchers can play nice together – I understand that a lot of tech firms who use ethnographers do just that.

  17. Tony

    Here is Alexis de Tocqueville on the limitations of statistics:

    General ideas are no proof of the strength, but rather of the insufficiency of the human intellect; for there are in nature no beings exactly alike, no things precisely identical, no rules indiscriminately and alike applicable to several objects at once. The chief merit of general ideas is that they enable the human mind to pass a rapid judgment on a great many objects at once; but, on the other hand, the notions they convey are never other than incomplete, and they always cause the mind to lose as much in accuracy as it gains in comprehensiveness. http://xroads.virginia.edu/~Hyper/DETOC/ch1_03.htm

    All statisticians should mull the nature of this definition!

  18. justaguy

    My sense is that most people doing research that relies heavily on stats aren’t naive about problems with representation and equivalency. Razib certainly describes those issues more succinctly and eloquently than I can. Its just that they often acknowledge those issues, put them aside and get to work.

    And I think there’s a definite cultural difference between fields – anthropologists see ambiguity and messiness and want to dwell on that, whereas economists see it and want to make abstractions which allow them to build models in spite of it. I once asked a friend who was almost finished with an econ PhD at the Sorbonne whether the folks in his department thought economic theories of rationality were real and universal, or were useful assumptions for modeling. He replied that he was pretty sure they were meant as assumptions, but in all his years as an economist nobody had ever brought the question up.

    And then, of course, when statistics get out of the hands of experts and into public conversation it often turns out poorly. When Kuznets developed the GDP it served an incredibly useful purpose – nobody had a way of conceptualizing aggregate economic activity, and they didn’t know if things were getting better or worse after the depth of the depression. Kuznets said he was afraid people would mistake GDP for a measure of the health of the economy, which was exactly what happened.

  19. Tony

    @Justaguy:
    You are very magnanimous with the statisticians! But I think that the tendency to push “culture” aside as statistical noise in order to protect basic assumptions is a big problem.

    The good news is that at least some editors seem amenable to pokes from anthropologists, and the new technology of on-line publishing has the means (i.e. comment sections) to do so.

    To enliven things, here is more of the writings of De Tocqueville on the statisticians among us:

    “Having superficially considered a certain number of objects and noticed their resemblance, [humans] assigns to them a common name, sets them apart, and proceeds onwards.

    [Such g]eneral ideas are no proof of the strength, but rather of the insufficiency of the human intellect; for there are in nature no beings exactly alike, no things precisely identical, no rules indiscriminately and alike applicable to several objects at once. The chief merit of general ideas is that they enable the human mind to pass a rapid judgment on a great many objects at once; but, on the other hand, the notions they convey are never other than incomplete, and they always cause the mind to lose as much in accuracy as it gains in comprehensiveness.”

    By the way, De Tocqueville in his day (the 1830s) used census and other data in his descriptions!

Comments are closed.