Inferring an AQ

Back in December Greg and I posted a draft of a paper on assortative mating, class, and caste. We asked for input and we got a lot, for which we are grateful. In that manuscript we described a thought experiment in which the rate of boiling off of young Amish each generation might act like truncation selection in favor being Amish, whatever that might be. We suggested that one might construct an “Amish Quotient” inductively by analogy with the way that IQ was discovered and developed, purely from correlations among observable traits. Pie in the sky but all is fair with thought experiments.

A week later an undergraduate student of mine, Michael Weight, walked into my office with a 1970 paper from the American Anthropologist (Wittmer 1970) reporting on a personality test given to 25 young Amish men and 25 Rural Indiana men as controls. We were able to obtain Wittmer’s 1968 doctoral dissertation from Indiana State University with the raw data, scores on the PF16 personality test. I have never had much faith in personality tests but, with nothing better, Weight thought it was worth a look. The PF16 was developed in mid-century by Raymond Cattell without modern machinery of multivariate analysis: this is perhaps and advantage rather than a disadvantage. On the other hand the test is proprietary so it would be very difficult for an unlicensed practitioner to try to make sense of it. The output is in the form of scores on 16 variables, and the supposed key to interpreting these variable is widely available.

In the spirit of exploratory data analysis Weight computed a correlation matrix from the 50 young men by the 16 PF16 scores and looked at the first pair of principal components. The are what Steve Hsu calls “lossy compression” of the original data, yielding an optimum (in the sense of least squares) picture of the differences among the subjects and, in the dual space, of the differences among the variables.

Here is the result: each letter is the placement of an individual subject in the PCA space. The “a”s are young Amish men while the “n”s are young non-Amish men. There is almost a complete separation between the two groups: a Fisher discriminant analysis between the two group misclassified two of 50 subjects with its hit-miss table.d There are black circles around the two Amish men misclassified by the discriminant function.

There is a paucity of comparable PF16 data on the internet, probably because so much of it is proprietary, used in personnel selection. Weight did find one paper giving PF16 scores for a sample of managers as well as a random sample in the UK. The group average manager is shown on the figure with the large ‘m’ while the average random Englishman is shown with a large ‘e’. We entered the UK group means into the data table 12 times each so all three groups, Amish, non-Amish, and UK managers and random people were given roughly equal weight. Interestingly the UK random sample group is right in the center of the generic Indiana non-Amish. The UK managers, all of whom were finalists for high management positions, are more distant from the Amish.

I am not sure that this is much evidence in support of our selection by boiling off hypothesis about gene-culture coevolution among the Amish, but we certainly have failed to falsify it. We wonder especially why generic Indiana young men are so similar to citizens of the UK yet so different from Amish young men who likely live right down the road.

A convenient property of PCA is that we can look at the differences among the variables (called “factors”) on the same axes. This figure plots the two-dimensional reduced space of the PF16 variables, labeled from “a” to “q4.” Since the axes are the same we see right away that Amish men are high on factor g and low on factor q1. From the standard interpretations given for the PF16 tests Amish men, relative to their neighbors, are “..exacting in character, dominated by sense of duty, persevering, responsible, playful.” He is also “conscientious and moralistic.” His low q1 score suggests he is also “confident in what he has been taught to believe, and accepts the ‘tried and true'”. He is “cautious and compromising in regard to new ideas”. He “tends to oppose and postpone change, is inclined to go along with tradition, is more conservative in religion and politics, and tends not to be interested in analytical ‘intellectual’ thought.”



Bartram, D. The personality of UK managers: 16PF norms for short-listed applicants. Journal of Occupational and Organizational Psychology, 65(2):159–172, 1992.

Wittmer, J. A Comparison of the Variability of Perceived Parental Behavior Characteristics and Personality Traits of Twenty-five Non-Amish and Twenty-five Amish Youth, Between the Ages of Eighteen and Twenty, from the same Community. Dissertation, Indiana State University, 1968.

Wittmer, J. Homogeneity of Personality Characteristics: A Comparison between Old Order Amish and Non-Amish1. American Anthropologist, 72(5):1063–1068, 1970.

This entry was posted in Uncategorized and tagged , , . Bookmark the permalink.

45 Responses to Inferring an AQ

  1. melendwyr says:

    These results seem to be pretty much what we’d expect from a casual glance at Amish culture. Which unfortunately makes it something of a “dog bites man” story.
    Is there any indication that the Amish population’s composition is actually changing? From what little I know about their parallel society, they’ve having an increasingly hard time finding adequate farmland to match their growth rates. I would guess that there’s subtle but increasing pressure to dissuade youths from coming back from rumspringa, in the same way that polygamous Mormons work to rid themselves of excess male youth. If the need for tradition and novelty were being strongly selected for, we might expect desertion rates to drop off with time, but politics and competition might mask that effect.

    • gcochran9 says:

      Defection rates have decreased over time, not increased.

      They’ve almost certainly changed over time, become more psychologically fitted to their niche. There are only two ways of avoiding that conclusion: abjure behavioral genetics, or give up arithmetic.

      • I was afraid of that. I’ve long been concerned about the long-term effects on population of high-fertility groups like Amish, Laestadians etc., and have suspected that selection will lower the defection rates in the long run, speeding up the growth rate of groups like these further. Moreover, this selection can be expected to make them even less inclined to listen when it becomes necessary to limit their fertility to avoid catastrophic overpopulation.

        That crisis is still a few centuries ahead, and much will happen before then, but at present it looks as if mankind will eventually jump back into the Malthusian trap with both feet.

        Maybe I ought to get a blog of my own and write a long post about this issue, but neither my knowledge of demographics nor my wrting skill is really up to doing it right, and anyway I don’t expect anybody will pay attention.

      • melendwyr says:

        ” There are only two ways of avoiding that conclusion: abjure behavioral genetics, or give up arithmetic.”

        Which is why it’s a shame you don’t have data showing change over time. Because you know there are genetic influences, and I know there are… but the people I argue with frequently do not. Every little example showing that culture isn’t a plausible explanation helps.

      • MontyLives says:

        I’m not arguing that the Amish haven’t changed over time. However, it’s worth noting that the Amish themselves explain a portion of the decrease in defections to the widening gap, caused by technological progress, between their way of life and the English way of life. The “English” world is apparently more alien to them today than it was one hundred years ago.

        • gcochran9 says:

          I had thought about that, too – it sure seems reasonable. But the central point, that this situation would almost inevitably select for a shift in personality, seems inevitable.

    • I don’t know what the data are on farmland prices in the US, Canada, and Mexico. In my lifetime a lot of farmland has been taken out of production and, in my part of the world (upstate New York) arable land is cheaper than it was in the 1970s.

      Hard to say about “dog bites man”. Certainly the folk belief that the heavy hand of culture shapes people seems to have little or no empirical support. Certainly the old belief that families shape people has almost complete fallen by the wayside, replaced for a while by the mysterious “non-shared environment”. That would better simply be called “noise” or “error” or “random effects.”

      We do have empirical data on heritabilities of personality traits and we do know the evolution of quantitative traits works. Lo and behold what we have in the Amish data, such as it is, is just what we ought to find. That is of course no confirmation, we can still say “just what we expect”, but that gets us nowhere.

      I am coming around to the view that almost all of social science has been just as much a harmful disaster as Freudianism was in psychology and Fascism and Marxism were in government.

      • panjoomby says:

        social science & education are swayed/led by “sweet-talkers” & anecdotalists over scientists (most TED talks are by sweet-talkers, malcolm gladwell is an anecdotalist). Perhaps they should be called Testimonialists. Jon Haidt did good work antagonizing social psych by pointing out how left-leaning social science academics are & how that can effect what gets published. Psych journal articles with elaborate socio-cultural hypotheses get published over occam’s razored biological explanations. I feel bad for teachers – rampant environmentalism means they are expected to turn lead into gold, b/c – it’s all environmental, so teachers just need to do better!
        that 16PF was considered pretty empirical in its day, but pretty over-factored (about 10 or 11 factors too many:)

      • Anthony says:

        Farmland in Iowa has been getting more expensive lately, partly due to the effect of fuel alcohol subsidies on the price of corn (maize for the non-Americans in the audience).

        There’s affable land, and then there’s arable land. It is possible to feed yourself from stony clay soils with a 4 month growing season, but if there is better land really available, why bother?

      • Anonymous says:

        “Psych journal articles with elaborate socio-cultural hypotheses get published over occam’s razored biological explanations.”

        What’s a good illustrative example of this?

      • melendwyr says:

        “In my lifetime a lot of farmland has been taken out of production”
        Which is a disaster in the making, as the places where food is produced become ever more distant from the places where it is consumed, making us even more dependent on tech- and energy-dependent transport.

      • harpend says:

        Melendwyr above is absolutely right: we need longitudinal data. That is almost here now with lactase persistence from fossil DNA in Europe. Soon……..

  2. Mike Johnson says:

    Very interesting. Seems plausible.

    It might also be valuable to look at the effect of the Amish population on the US’s overall AQ. With an average of ~7 children per family (compared to a national average of 2.3), even a modest defection rate could significantly increase the AQ of the US as a whole, given 100 years. Yes/no?

  3. Interesting thought. I doubt that it has had much effect yet, but we would have to have population estimates for the other Anabaptist groups like Hutterites and Mennonites to do the number. Amish numbers in the US are pushing 300,000 this year and the 10% estimate of the boiling off rate refers mostly to young adults.
    Here is another plausible scenario. The rate of population growth of Amish is such that the population is doubling about every 25 years. At this rate in a century they will number nearly five million, having overtaken the US Jewish population. Put this together with increasing (as I see it) bifurcation of the US population into haves and have nots.
    The haves, or SWPLs, eat a lot of organic food. Half the vegetables in my neighborhood grocery store are organic and our local farmers’ markets prosper. I struggle against this in my own home and I lose big. My logic is that the 1960s are past us but the boss lady rules here and I have eat that stuff as well as pay for it.
    If this keeps up there will be a growing market for organic food and the Amish and other Anabaptist groups will have much of the market. Do we have a future of a countryside full of Anabaptist organic farms supplying a substantial part of our foods? Probably not of our caloric supply but of our nation’s high priced SWPL food.
    I would be happy to see it. Amish are not going to carjack me or assault my children and grandchildren.

    • Boris Bartlog says:

      The Amish are none too rigorous in their application of organic rules. Certainly they are small-c conservative in their use of pesticides and artificial fertilizer, but since they have no religious injunction against either, their use of them varies widely.
      That said, the fact that they have an abundance of labor, a good work ethic, and familiarity with old-school farming means that they are certainly well-suited to become niche producers of organic food. I’d also trust their assertions about the nature of their product over those of some other groups who might also try to compete in this space.

  4. Incidentally I trust that our remote neuralizer will erase memories of the first version of this post in which I confounded the UK sample of managers and the UK norms of a random sample from the UK. I have corrected the post: now the ‘e’ is random English centroid and the ‘m’ is the sample of finalists for high management positions.

  5. Greying Wanderer says:

    culture driven evolution – cool

    • That’s precisely correct! It’s *incredibly* interesting to me. It seems culturally driven evolution is good enough to allow the Amish to create a whole new personality trait (the trait measured by AQ) in a period that extends a little longer than the age of the United States.

  6. j3morecharacters says:

    I think we should first establish if the Amish, Hutterite, Mennonite and Anabaptists in general are a solid, stable, identifiable ethnic group. In their origin, they were a mixed group of Austrian, Moravian, Swiss, etc. religious refugees, and I think there has been always considerable internal movement. I know of Anabaptists buying land and colonizing in the Bolivian Chaco, some came from Canada, others from Kazakhstan and so. They are not impermeable as you seem to assume, they do assimilate converts and of course, some move to other Christian churches or drop out.

    • gcochran9 says:

      The total number of people who have joined the Old Order Amish in the last century is in the tens, at most. Many have left, but biased leaving is the point.

      As far as I know, nobody before me ever suspected that they were experiencing selection for enhanced Amishtude. I’ve been talking about this with Henry for years.

      • JayMan says:


        The thought had occurred to me, too. A few weeks before I first saw your post on it. 😉 Though not years, no.

        • JayMan says:

          Hmmm, I guess the phone app doesn’t like smilies… :\ 🙂

        • Actually I suspected something like it long ago (many years, maybe a decade or two), though specifically with reference to Laestadians rather than Amish.

          I was already thinking about the possibility that some subcultures might be much more fecund than the general population and the possible consequences if that fecundity was a lasting feature, whether due to genetic or cultural inheritance or both. Then one day when I (unusually) happened to be watching TV, the speaker said that families with 10 children or more were common among old-style Laestadians. That scared the hell out of me (I tend to take distant dangers seriously).

      • jb says:

        The total number of people who have joined the Old Order Amish in the last century is in the tens, at most.

        I remembered reading something about an an Amish or Mennonite community of converts in Japan that was small but quite serious about it. Turned out to be Mennonites. I couldn’t find the original article, which I read because I was interested in Anabaptist demographics, but I found a number of references, such as this one. I remember from the original article that the Mennonites in America are aware of the Japanese community and considere it legit. Don’t know if there has been any mixing though.

    • Boris Bartlog says:

      They’re inbred enough to warrant their own database of genetic disorders:
      One oddity I noticed in my dealings with them was that they had an awful lot of short men, to the point where I wondered whether the men were actually taller than the women on average. But I haven’t seen anything published on that.

      • gcochran9 says:

        One way to avoid leaving the Amish would be to be relatively uncompetitive in the outside world.

      • Boris:

        An extension of the AQ research/results presented here is seeing if the selective pressures the trait AQ measures was adaptive to are the same pressures that clustered the genetic diseases plaguing the Old-Order Amish together. In particular I’ve been suggesting to Henry for the past 4 months (or so) that Ellis-van Creveld Syndrome is a balanced polymorphism (and by a nice possibly property, others in the cluster would be balanced polymorphisms also). I’ve been writing a proposal for funding that would allow me to go to Lancaster to see if heterozygotes have an “Old-Order Amish Body Plan” that’s strongly sexually selected for. In other words, not only is AQ a mental index but it may also be a physical index. Did you notice anything else about Amish men, or did your observations end at their heights?

      • Boris Bartlog says:

        Not much else. Nothing terribly unusual about their builds; the younger men were leaner and more muscular than your average American, as you’d expect from people that did a lot of physical work. The older people (both men and women) were about as overweight as typical US whites of that age. Lots of brown hair and brown eyes; blue eyes were a minority, actual blond hair was rare. They looked kind of Tyrolean, which fits their geographical origin.
        I would be surprised given their beliefs to see strong sexual selection for some particular male physique. Conservative religious cultures don’t tend to leave that much room for female preference to operate, and on top of that I think there would be a tendency for any really handsome men to leave the Amish entirely.

  7. melendwyr says:

    Even if we attributed any differences strictly to culture, culture is still transmitted from parents to offspring – and it can be remarkably difficult to get people to change their deep beliefs. So if people having unSWPLish beliefs are reproducing at rates far above SWPLish people, they will eventually dominate culturally.
    We can’t even argue that their cultural traditions will eventually shift to favor non-reproduction as the rest of the Western world has, because they are reacting to that sort of change, explicitly rejecting it. The circumstances are totally different.
    I can only assume that the faith that people can be made to conform runs bone-deep among the mainstream… or they’re simply not thinking that far ahead. I can’t tell the two stances apart from this distance.

  8. So, any other inbreeding, enclavistic ethnic groups where this kind of “boiling off” model might see some use?

    Shall we go alphabetically, and continue down the A? Some other tightly knit, endogamous ethnic group starting with A? Pick some new letter a few notches down from M? Group with “unique” personality traits and high boil-off?

  9. I would be cautious about concepts like “doubling every 25 years” when it comes to populations. The arithmetic is solid enough, but we never get to the second half of the chessboard, and seldom off the first row. In this case. “Doubling in the next 25 years” is about the most we should dare. Relatedly, growth trends do not persist forever (@ Monster from Polaris), and population is expected to peak at 9 billion a few decades out. Relax. Imaginative people can conjure up nightmare scenarios of hordes of unfed people storming about and ruining the best vacations spots, but A) Likely not and B) We don’t actually care as much as we think at first glance. Rhetoric about legacy and stewardship aside, people care about their children and some small similarity groups, plus some larger, grander group they hope sustains in recognisable form, such as a church or a country. Beyond that, there is no evidence that we care about posterity very much. Look at your own family and chosen institutions. Who cares much about their great-grandchildren? Only in theory. When they are young they have mythic value. When they are 25, whether they are saints or sinners affects us far less than we would have predicted.

    What we call caring about posterity turns out to be very much tied up with how we want Yellowstone, or William & Mary, or the Amish Ordnung to look now.

    • “population is expected to peak at 9 billion a few decades out”

      Yes. But as I understand, this is based on national TFRs. You really need to consider subculture TFRs for long-range predictions. An initially small group that keeps growing fast for long enough can eventually become a majority, especially if the overall population is declining.

      Still, it’s true that long-range (multi-century) predictions are unreliable anyway.

    • aisaac says:

      If you look closely at the predictions that expect the population to peak at 9 billion, you can see that they’re bullshit. This document explains the assumptioins behind the UN population projections which predict that peak; they assume, just because, that all countries will undergo the demographic transition (maybe true) and that all countries will hover around replacement fertility forever (undoubtedly false, the existence of the Amish alone disproves that).

  10. harpend says:

    The Amish have been doing their high fertility thing for several centuries with little sign of slowing down. Fertility has been declining a lot for other ethnicities in the US. We can hope that it continues. For Amish I don’t really wish fertility decline: they don’t crowd Yellowstone, they don’t beat up our highways and bridges, they don’t burn much coal and oil, they don’t migrate to our cities and suck welfare. What is this about William&Mary?

    • My alma mater. Very concerned with posterity. As for Amish fertility, it has indeed been high. But with 150K in 1995, do we really think that halving that in 1970, 1945, 1920, and 1895 really means that there were only 9.5K Amish then? They may have a high baseline, but they are still in a bubble of growth now. I’d bet against it continuing at quite that pace for another century.

      I agree that more of them would be a net boon to society.

    • Huh. Nice doubling then. Thanks.

    • MontyLives says:

      Amish fertility is falling among most of the larger groups. Many of them do take up birth control at some point, and the majority of Amishmen no longer work on farms. That said, they are highly decentralized, ultra-conservative groups do not practice birth control (e.g. the “Swiss Amish” of Adams County, IN, the Swartzentrubers, a lot of unnamed others I’m sure…), and one would expect the ultra-conservative Amish to eventually outbreed the merely conservative Amish.

      Which makes me think, incidentally, that any estimate of “boiling off” might want to consider intra-Amis boiling off. If a liberal Swartzentruber Amish kid joins the progressive New Order Amish and has only three or four kids (not at all uncommon among New Order Amish) instead of ten (which is the Swartzentruber average t.f.r.) I would think his traits are still being selected against, ultimately, within the Amish community.

      • gcochran9 says:

        The population growth rate is running at about the same rate, so it can hardly be the case that fertility has declined much.

      • MontyLives says:

        It doesn’t seem to want to let me reply to Greg, so I am left looking like I am having a conversation with myself.

        Well, the fertility fall-off that I’ve seen has all been in very recent studies, in the past ten years or so, and it’s been from around 7 kids to 5. Not sure how much of an effect such a recent change would have on population growth rate. Also, this has happened mostly among groups that have switched out of farming, and well after the switch was made. I suspect when these groups reach their fertility “floor” it will still be well above replacement.

  11. Oh, and I would have some suspicion about a personality test that has a purported factor that predicts both conscientiousness and playfulness. I have had to administer any number of stupid psychological “instruments” over the years because someone thought they needed that data for a “study.” Many of the items seem circular. I admit, I don’t know a thing about this one.

    • harpend says:

      I don’t know anything about it either. I assume it is mostly snake oil stuff, commissioned by the same kinds of people in government and business who pay Malcolm Gladwell big money to give talks. Good information about what these sorts of tests actually predict is scarce because they are proprietary.

  12. Greying Wanderer says:

    A larger scale version of this kind of “boiling off” in the centuries before the industrial revolution

    “West of this line, the average age of marriage for women was 23 or more,[3] men 26, spouses were relatively close in age,[4] a substantial number of women married for the first time in their thirties and forties, and 10% to 20% of adults never married…The Western European pattern of late and non-universal marriage restricted fertility massively, especially when it was coupled with very low levels of childbirth out of wedlock…The origins of the late marriage system are a matter of conjecture prior to the 16th Century when the demographic evidence from family reconstitution studies makes the prevalence of the pattern clear”

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s