The Black-White Mortality Crossover

Most of social science still operates as if there were no biological differences among human groups. This so-called “uniformitarian” assumption is pervasive and it works in the sense that a lot of research money falls into the social science hole. On the other hand there are several glaring instances in which it seems to fail badly: a typical response to these is to look embarrassed and quickly change the subject else to social-science-talk away at them, hoping they can be made to disappear.

A recent such instance is the “Hispanic Health Paradox” in which Hispanics in North America enjoy better health and live longer than Anglos. For example, the table on page 11 shows that Hispanic females have 2.7 more years of expectation of life at birth than do Anglo females, 83.1 to 80.4 years. If poverty is the cause of health disparities, how can this be?

An older instance is the “Black-White Crossover” in mortality between US Black and White populations. In North America death rates for Whites are notoriously lower than death rates for Blacks, so that the expectation of life at birth is about five years greater for Whites. This generalization holds at every age up to 70 to 80 at which times the hazard of death curves cross. Eighty year old Blacks live longer, on average, that 80 year old whites. A classic treatment is the monograph by Manton and Stallard (1984). Much effort by demographers has been directed at wishing this away but they just can’t get rid of it. Indeed, today a Google Scholar search of “Black White Crossover” returns a recent paper about breast cancer: Black women have a higher annual risk before about age 50 and a lower annual risk after that age.

A few years ago Renee Pennington and I (Pennington and Harpending 1993) did a lot of ethnographic research among Herero, about whom I often write in this blog. They are prosperous Bantu-speaking ranchers in the northern Kalahari of Namibia and Botswana. Unique in Africa, they have a system in which every year is given a name. Everyone knows not only the name of the year of his or her birth, they know the birth year names of their parent and grandparents and occasionally even further back. Herero are family fanatics, and you would know the birth years of your friends in the same way that in North America you are likely to know the middle names of your friends. In line with other posts at this blog about Herero, “family” has nothing much to do with marriage, and hardly anyone knew years of marriages and divorces. Year names are known back to the early nineteenth century.

Few Herero know how the year names map to calendar years, but there are specialists who curate year name lists. An informant might tell me that his grandfather was born in “onburra ojomevaomengi”, year of heavy rains. I could look it up in my copy of the name list and find it was 1854.

North American data about the crossover come from censuses mostly. How many 85 year olds were there in the state last year and how many deaths of 85 year olds? We were able to take advantage of the year system to estimate mortality statistics from family history interviews. We used life history information from living people and patched in information from partial lifetimes of other family members using a Kaplan-Meier kind of procedure. Details are in the book. At any rate our information was completely different in kind from census data from large North American populations. Both have weaknesses, and it is interesting to compare our estimated hazard of death curves with census results. Here are the estimated curves for US White and Herero males and females: The pattern, for what it is worth, is a great exaggeration of the North American pattern: the crossover occurs way earlier in the lifespan.

We have a brief discussion of the phenomenon in the book but we have never looked any further into the matter. I have been meaning to put a graduate student on the issue for years but I always forget. At any rate we did a quick and dirty sampling of world censuses and produced this scatter plot. The horizontal axis is expectation of life at birth, longer lifespans going east. The vertical axis plots the probability that a 75 year old will live to 80, higher survival from 75 to 80 going north. There is a lot of scatter but there is an interesting regularity. African populations are the furthest north, European populations in the middle, Asian populations at the bottom. At oldest ages African live the longest, Europeans are intermediate, and Asians have the shortest remaining lifespans.

This needs to be looked into. BTW the point with label “Botswana census 1981” is raw data from the census. What national censuses do in much of the world is cook their data until they follow like a standard generic European pattern. There are, after all, no biological differences between populations: we all know this as a sacred truth. The data point from the Botswana census is data before cooking: you will not see it in the final census product.

Next, is the African pattern of aging related to late age at paternity?


Manton, Kenneth G., and Eric Stallard. 1984. Recent Trends in Mortality Analysis. New York: Academic Press.

Pennington, R., and H. Harpending. 1993. The structure of an African pastoralist community: demography, history, and ecology of the Ngamiland Herero. Clarendon press.

This entry was posted in Uncategorized. Bookmark the permalink.

40 Responses to The Black-White Mortality Crossover

  1. The fourth doorman of the apocalypse says:

    If there are genes that contribute to this longevity, it might make sense for the posessors of those genes to recognize other posessors of such genes.

    The crossover might occur because those with genes/alleles for greater longevity are not so evenly spread among the members of that group, compared, say, with Europeans/Whites. Or perhaps there are more copies of the bad alleles floating around in the population.

  2. gcochran9 says:

    I never believed in this before – thought it was error – but now I can imagine how it might be so.

  3. typal says:

    “Next, is the African pattern of aging related to late age at paternity?”
    According to Michael Rose populations with an earlier last age of reproduction stop aging earlier and live longer, and vice versa. Here

    • harpend says:

      By “stop aging” Michael means that the hazard rate stops increasing. I am not sure that the basis of that effect is very well worked out yet.

      On the other hand Rose is well known for breeding his “Methusaleh” flies with prolonged lifespan by restricting mating to older flies. Pushing the peak age of reproduction to older ages causes selection for longer lifespans, in line with Hamilton’s 1966 theory of senescence.

      • typal says:

        If it’s just longer lifespans why would there be a crossover? No, there is a transition involved. In Europeans the hazard rate stops increasing at an age when most people are not alive. For blacks the hazard rate stops increasing somewhat sooner, because if you might get a spear in the back tomorrow you better reproduce today, so they cross over and get more of a longevity effect .

  4. albatross says:

    So what’s the explanation you have in mind, Greg?

    There’s a plausible environmental explanation in the US, I think, if you assume that the cultural/environmental stuff that leaves blacks with shorter lifespans that whites tends to hit a lot harder early in life than later in life. (More crime and risky sexual behavior and higher risk tolerance generally all affect you a lot more in your 20s than in your 40s, and by your 80s probably most of the effects have been washed out.) But I haven’t thought much about this, so maybe that’s just all wrong.

    If we’re thinking in terms of genetic load, longevity might be one of those things like IQ that falls out of lots and lots of genes all summing up their effects. (Though the summing up model is wrong–it’s more like lots and lots of complexes of genes each get a veto on continuing your life–once that blood clot lodges in the narrowed bit of your coronary artery, the other genes probably stop getting a vote on your life expectancy.) Then, we’d expect lots of slightly harmful mutations to sum up to a shorter lifespan. Is the idea here that at some age, we’ve filtered out most of the people with a lot of these mutations? But then, why would they have lower mortality after that?

    • harpend says:

      You are hinting, I think, at the heterogeneity model of the crossover. According to this model, associated mostly with James Vaupel, individuals vary in frailty. In a harsh environment more frail individuals die early so that older cohorts are made up of the hardiest, who then life longer.

      This model has not fared so well in the sense that explicit tests of it all seem to have failed, in humans and in flies. There was an interesting exchange about this in SCIENCE a few years ago between Vaupel and other, on one side, and Finch and Crimmins on the other. Finch and Crimmins had Vaupel for lunch.

    • Zombie Shane says:

      You can’t be serious about analyzing any African-AMERICAN biological statistic without examining the relative Caucasian Admixture of the subjects in question: Nearly pure African black through mulatto through quadroon through octoroon etc etc etc:

      To expect pure-blood Africans to have similar biological outcomes as, say, the children of Susan Rice and her husband, Ian Cameron, would be simply nuts.

  5. dearieme says:

    Survivorship bias?

  6. J.G. Quiggin says:

    There’s probably higher genetic variation among the Herero, right? It seems plausible that this ramifies into a broader (than whites’) distribution of longevity. Then by the age of 75, when the left 3/4 or so of this distribution have died, what remains is a distribution with a higher mean than the (likewise clipped) white distribution.

    • harpend says:

      Right, great idea, seems to have failed when looked at explicitly. See my reply to albatross below (above?).

      • J.G. Quiggin says:

        I’ll take a look at the Science exchange, but might you provide a one-line summary of how the data tell against this “heterogeneity thesis”?

        Here’s a second shot, anyway. The lower early life expectancies are plausibly explained by higher-risk/shorter-time-horizon behavior (via high T levels, or etc.), or again by mutational load (or etc.). But then perhaps the higher later life expectancies result from selection for a few “super-longevity genes”—selection which only has much to work on because polygyny has allowed these unusually long-lived males to keep having children (whereas in monogamous cultures, having “super-longevity” [extra-life-after-75] genes wouldn’t affect the gene pool).

  7. Jehu says:

    Don’t blacks in general have heavier bones? An awful lot of mortality in very old age is due to a death spiral after breaking a hip or similar injuries. Also, there’s the question of number of nearby relatives—a lot of the old people I’ve seen seem to basically expire after what a wargamer would call failing a morale check, which seems to happen an awful lot in the various old folks and nursing homes to the lonely.

    • Anthony says:

      The post hinted that cancers were a factor in that crossover, though one type of cancer crossing over at a much lower age than the general mortality crossover isn’t conclusive of anything. Of course, even if cancer *is* a factor in the mortality rate changes, bone structure could play a role in that – the body does try to repair the damage from broken bones, and the system that takes care of that could fail in ways which lead to cancer.

      I wonder if *that* connection has ever been studied – incidence in cancer after breaking a bone, by age groups.

    • pablo brandt says:

      Actually blacks are more prone to bone diseases after 30.

  8. typal says:

    If older biological fathers are common in Africa, as Gregory Cochran is saying, then Rose’s ideas are not the explanation for the crossover. Ethnographic Surprises : The Child Bride says that an old (Herero) man who is the husband of a young girl is likely not going to be the biological father of his wife’s children. Has that pattern been common among Africans? Going by the sky high peak in testosterone of Black African men in their early twenties and the subsequent rapid fall to below white levels I think it has.

    Paper from Rose’s lab. Does Aging Stop?

  9. Steve Sailer says:

    Another issue worth looking into is the shortage of black males relative to black females from birth on up. Among whites and Asians in the U.S., about 105 boys are born for every 100 girls. For blacks, it’s about 103. And then the death rate of black boys is quite high relative to black girls. So, for whites and Asians during the prime mating ages, the sex ratios are pretty close to 100 to 100, as theory would predict. For blacks, there are fewer than 100 black young men for every 100 black young women.

    I don’t have a theory to explain this anomaly, but I’ve long had a hunch that it’s important.

    • harpend says:

      Same pattern in (much of) sub-Saharan Africa. Terrible high mortality of young males.

      • typal says:

        “US White and Herero males and females: The pattern, for what it is worth, is a great exaggeration of the North American pattern: the crossover occurs way earlier in the lifespan.”

        The crossover is sooner because Hamilton’s forces of natural selection bottom out and stabilize only when you are long past the age of reproduction. Rose’s theory would predict that transition happening earlier in a population like the Herero where reproduction happens at young ages.

  10. DB says:

    One should make a distinction between crossing over of (i) conditional probability of surviving x more years, given that they made it to age (y-x), and (ii) overall probability of surviving till age y. From your description, it sounds like only (i) crosses over, which is mildly interesting but not nearly as demanding of explanation as (ii).

  11. Rob King says:

    Could the crossover have something to do with fast vrs slow life histories (Stearns, 1976) combined with some sort of pleiotropic effects? E.g in the US black people tend, in certain ecologies, to be prompted down a fast life history track. This carries with it risk taking behaviours, early menarche, etc. However all of these effects only apply to the first half of life–pre menopausal–so if you do happen to make it through them you are golden. Just a thought.

  12. typal says:

    If the effect on fitness of an increase in mortality is zero, as is the case after the age when survival affects reproduction, then natural selection does not weed out the genes that cause mortality at these ( post reproductive) ages. On the ‘fast track’ the post reproductive ages are reached sooner, and the genes that have bad effects at post reproductive ages are exerting their bad effects from an earlier age. So the ‘fast life history track’ people suffer the post reproductive stage of aging from an earlier age. It seems to me it would be neccesary to claim that blacks have been selected for reproduction at later ages to explain blacks being healthier in old age, as they are.

    But how explain eighty year old Blacks living longer than eight year old whites when Blacks have 5 years LESS life expectancy at every age up to 70 – 80? You can’t reproduce at late ages if you don’t live to those ages!

    Rose’s proposed explanation is that at a stage of life beyond the potential to reproduce, the effect of natural selection is no longer falling as it has ‘bottomed out’. Rose suggests that DECLINE in the effect of natural selection is responsible for aging. When this decline finally ends, at post reproductive age, mortality-rates plateaus. Rose says that, relative to the age of reproductive maturity, a transition to the late life stage of life occurs much later in humans than in the populations of flies for which there is data. In humans the ‘late life’ stage of life is only reached at 90 years old whereas the data for flies scaled to humans would predict a ‘late life’ stage for humans at 40–50 years old. Rose suggests that human populations’ adoption of agriculture led to more children surviving to adulthood, and to reproduction occurring later in life, and to have resulted in high population density, thereby increasing the range of ages not under selection. So the better your ancestoral culture was at keeping most children alive the longer you have to live to get to the mortality-rates plateaus. European culture was better at this, hence they don’t get to the mortality-rates plateau as soon as blacks.

    It would be interesting to know how Bushmen age when they are following their traditional lifesyle and diet. They should do very well according to Rose.

    • harpend says:

      Interesting discussion: there are two models on the table, both appealing to Michael Rose. One is that later reproduction should select for longer lifespan, the other that early termination of reproduction should lead to the hazard rate going flat sooner.

      I favor the first, but no matter. The second is sort of testable from our data since we should see the hazard curve turn East and go flat at older ages.

      Here is a link to a slight redraw of hazard rates. They are separated by sex and by rates before and after 1965. No trace of turning east in either case—hazards just keep on increasing.

      Truth be told I don’t remember right now how we smoothed/fit the raw data. Perhaps they do turn East at oldest ages and our methods cover it up? Someone please send us a graduate student who could look into the matter with our old raw data.

      • albatross says:

        I’m confused by the second explanation. What’s the proposed mechanism by which 80 year old black men have a lower hazard rate than 80 year old white men?

      • typal says:

        albatross, I suggested it may be that, in evolutionary time, blacks had earlier cessation of reproduction. Based on fruit fly experiments Rose says earlier cessation of reproduction turns off natural selection, and the the genes not being subject to natural selection means they reach a plateau where aging is ‘halting’. In humans it is at 93, maybe blacks reach that plateau sightly sooner. His explaination is based on mathamatical models and as he says ‘counter intuitive’ – so don’t expect to understand it !

        Smaller effective population size also has that effect, according to Rose, maybe that’s another difference between the ancestors of blacks/Hispanics and whites . (But don’t hold me to the fire with any of this.)

  13. Anonymous says:

    The factor of an agricultural diet might explain the effect of early reproduction on mortality in later life being blunted , and too subtle to show up in that data set. Michael R. Rose says selection pressures decline after we cease reproduction, hence at older ages bodies become progressively less adapted to agriculture. There has only been 10,000 years to adapt to agriculture. The Herero diet (sour milk and maize porridge) is not pre-agricultural. (Rose says grass-derived foods, such as grains, rice, corn and sugar cane, and anything made from milk should be avoided by everybody after the age of 30.)

    The Bushmen’s ancestral hunter gathering diet was genuinely stone age and they are presumably better adapted to that diet than any agricultural people are to agricultural foods, given how long Bushmen had to adapt to it. ( they must be very badly adapted to agricultural food). I don’t think there are any hunter gatherers peoples still following their ancestral diet.

    • typal says:

      Anonymous was me.

    • Greying Wanderer says:

      “The factor of an agricultural diet..There has only been 10,000 years to adapt to agriculture.”

      Completely random thought sparked by that comment but are there crossovers of a similar pattern to this between other ethnic groups where one has been agricultural for much longer than the other e.g. Finns and Greeks?

      (The thought behind this being…HGs are much more healthy to make up for the harsher environment. In that harsh environment it doesn’t make them live much longer both because the environment is harsher and they get killed off young by other HGs. However if they were taken out of that environment and placed in a safer one the increased healthiness would make them live much longer *if* they weren’t killed off by the food.)

  14. albatross says:

    One explanation I can kind-of imagine for the situation where blacks live shorter lives than whites, but blacks at 80 have a better chance of making it to 90 than whites at 80 involves the same gene having different effects at different ages. Antagonistic pleiotropy happens when the same gene makes things better one place and worse someplace else.

    Suppose there is an allele that improves my chances of surviving from 40-50, but decreases my chances of surviving 80-90. If over the last couple thousand years, whites had a noticably better chance of surviving to 40 than blacks, then there would be stronger selection pressure in favor of this allele that gets me to 50 but then doesn’t let me make it to 90.

    Does this make sense? I’m certainly no biologist, so I could be missing something….

    • typal says:

      I think natural selection is supposed to systematically favour genes for survival at ages close to the prime reprodutive young age over genes for survival at all or older ages. So genes for surviving at all ages would not have an advantage over one for surviving while reproducing (ie surviving while under the forces of natural selection). I think this is why Rose says that genes for survival at later ages cause a mortality rate plateau at a later time in populations where natural selection has ideal conditions such as large effective breeding population size, survival of most children. Reproduction into later ages keeps thoses ages under selection. Agricultural society had stronger selection than isolated groups of hunter gatherers according to Rose .

  15. Jason Malloy says:

    Isn’t a mortality crossover an eventual statistical probability given any group that dies at a disproportionate rate for prolonged stretches of life? If you take any high mortality group, the subset of that group that you’re left with after decades of attrition is going to be a much hardier and healthier than the larger and more typical subset left in a lower mortality group.

    I have come across papers in the past that have observed similar crossovers for smokers, Native Americans, Australian Aborigines, and even lower SES groups, more generally.

    • Jason this makes a lot of sense: it is the Vaupel heterogeneity theory. Generally it has not fared so well when folks have tried to test it. See e.g. the reference to the Finch and Crimmins versus Vaupel exchange above.

    • Jason Malloy says:

      Then again, going by that list of supercentenarians above, blacks are considerably overrepresented at record ages.

      135 White (68%)
      55 Black (28%)
      2 Mixed Blacks (1%)
      2 Puerto Rican (1%)
      1 Japanese (0.5%)
      1 Mexican (0.5%)
      1 Native American (0.5%)
      1 Unknown (0.5%)

  16. ironrailsironweights says:

    Though I would have to wonder whether birth records for blacks in the South during the 19th Century were truly accurate.

    • harpend says:

      The data about US Blacks are from the census, that is current deaths and cohort sizes. Manton and Stallard was published in 1984 so maybe from the 1980 census?? Still the counts are dubious, especially of your males. Demographers go through hoops trying to “correct” the numbers, and the corrections in Manton and Stallard for young Black males pushed 10%. No one, to my knowledge, has published about 19th century US Black rates.

  17. JayMan says:

    Somewhat related, here’s something you may find interesting. Apparently, there is a strong inverse relationship globally between a country’s historic and present pathogen load and age-adjusted Alzheimer’s rates:

    Hygiene and the world distribution of Alzheimer’s Disease

    The paper blames “hygiene”, but I don’t buy that.

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