If your parents died early, will you die early too?

But in fact the correlation of longevity between individual parents and children is very low.  For the people dying in England in the period 1858-2012 with the rare surnames used in chapter 4, we can measure the correlation of longevity between fathers and sons for more than four thousand sons surviving until at least age 21.  That correlation is only 0.13.  If we take the average of both parents’ ages at death, that correlation increases to 0.26.  But it is still low.  In reality, your age at death is not strongly predictable from your parents’ age at death.  All those saving more for retirement simply because both parents are fit, healthy, and in their nineties should stop immediately.  Your expected additional longevity relative to the average is only three years.

That is from Greg Clark’s new and noteworthy The Son Also Rises: Surnames and the History of Social Mobility.  Here is Kevin Drum on the book.


I wonder if that is just a product of the British industrial revolution. The British comic Alexei Sayle did a joke about this once where he said you were mostly likely to die of the same cause as your Grandparents. Which was great news as he would live forever given Britain didn't have any Cossacks.

I would think that if your Grandparents were killed by Cossacks, or Red Guards, or the SS, or the Khmer Rouge, or some random arm chopping gang of young men, the chances are pretty good that you would die a lot earlier than said grandparents. As you might well be killed too.

Before the industrial revolution the population was probably clearly divided between rural populations that lived a long time and urban populations that died young. More so in the early IR.

If he picked different dates and looked at different countries, he must get a different result?

This is probably too long interval to say anything in the biological sense. There were two world wars in between, and a pandemics of cholera.

I would say that *once you count in quirks of history, the correlation between length of life of parents and children is low*, but it does not really confirm or deny heritability of lifespan.

Try Iceland or Sweden, they haven't had big wars for centuries.

There were quite a few generations who did not lose a lot of people to world wars during that time, but I certainly agree that mortality patterns have changed over time and that this is an important point. Incidentally I don't have intergenerational mortality data from Sweden, but I do have data on related stuff which I thought I might as well add here in case anyone is interested (the stuff below is data from civil register data, 19th century Sweden - 1824-1896 - included in Bobbi Low's Why Sex Matters):

"Women married very late in Gullholmen [a Swedish parish] (26,1 years), and 58 percent of adult women failed to marry at all while in the parish. Women who married earlier had more children than those who married later. Women almost never remarried [...] [in Locknevi parish] 74 percent of [male] agricultural workers and servants living their entire life in the parish failed to marry, compared to 20 percent of lower-middle-class men [...] when landholdings were uneven, a man’s occupational status predicted the number of his children who would survive to age ten; wealthy men did best [...] [in Tuna] Landowners were almost certain to marry (95 percent), in stark contrast to other men (35 percent); they married women about 2.5 years younger than other men, and had about one to 1.5 more children. In sum, landowners had larger families no matter what the times, and their families were not only larger, but less variable in size than those of nonlandowners."

Occupational status was related in all parishes to a man’s likelihood of marrying, although the relationship was only marginally significant in Tuna, where land ownership was crucial."

I'm wondering to which extent people think these patterns still apply to Swedes living today (I don't know to which extent they do, but I'm a bit skeptical that results have held up over time into the 21st century). Using historical pre-modern medicine mortality patterns to predict longevity today might not be as problematic as using historical income data to predict fertility patterns (it may be, I don't know), but pointing out that things have changed a bit over the last 100 years might be an understatement. In 1815 on in four deaths in England was due to tuberculosis, and that disease still killed one in six people in France a century later (1918) - see the wiki. Infectious disease used to kill a lot of people, and now that they no longer do (in rich countries anyway) it's an open question how related the historical factors leading great-grandparents to living longer lives (and having (/more?) children) are to current mortality patterns. One hypothesis trying to explain the black-white gap in mortality rates in the US posits that higher survival rates among African slaves genetically predisposed to conserve salt and water may be part of the explanation why this subgroup have higher rates of hypertension and cardiovascular disease today (see e.g. Wilson & Grim, 1991). For some variables there may be a relevant 'historical connection', but in a lot of cases there'll not be any. The impact of various types of partly inherited cancers on expected longevity of an individual is much higher now than it would have been for that individual 150 years ago, because a lot more people now live long enough to get that cancer. One needs be very careful when drawing conclusions here. That said, I agree with Doug that it's not really a low correlation considering the number of environmental confounds.

This, before antibiotics the leading causes of death were tuberculosis, influenza and gastrointestinal infections. Is there a genetic resistance to these illness? Yes, to a certain degree. But diet variety and overall sanitary conditions have a greater impact on the mortality of these 3 killers.

So, what are the numbers after antibiotics were widely available?

Sorry, last comment is just trash.

Thought again about the whole issue and the question is why some correlation/causality is expected between parents and sons age of death? There are other interesting correlations between age of death and level of education and income.

The interesting question would be the correlation between parent/son age of death assuming both are educated and at least middle class income.

0.26 is not a very low correlation. Particularly because in many ways life expectancy is basically a sample size of 1. You can have very good genes for living long, but get hit by a bus or have a random mutation that turns cancerous. Realized life expectancy standard deviation is about 16.5 years. Having parents in the top 5% for life expectancy means that you'll expect to live an extra 8.5 years over the median. That trumps the life expectancy impact of smoking, obesity, lack of exercise and even the female-male gap. In short being born to long-lived parents is certainly the best thing you can do for your health.

I am suspicious that you have your numbers wrong. What is the standard deviation in life expectancy among people who lived long enough to be parents?

I expect that this correlation has been increasing over time.

It doesn't answer the question. Factors like early deaths, unhealthy lifestyle, social advancment etc. could distrupt this correlation.

This is why statistics gets a bad reputation. Each generation in England has found life expectancy increasing, medical science advancing and societal violence declining. If you looked at the lifespan of parents compared to their contemporaries versus that of their children compared to their contemporaries, you get a more useful comparison. Then you should net out war, misadventure and misfortune. Men killed in war or hit by buses don't tell us much about the heritability of lifespan.

Regardless, real science tells us that a great many diseases that limit lifespan are genetic and heritable.

Your description of what would be a more useful comparison is exactly what a correlation tells us.

"we can measure the correlation of longevity between fathers and sons for more than four thousand sons surviving until at least age 21."

How about both fathers and sons who lived to at least age 40?

Or whose cause of death was not misadventure or suicide?

Aren't those still relevant if the tendency towards suicide and misadventure are both potentially heritable - which I am pretty sure they are?

I don't know. I think you are right that violent and reckless behavior is heritable. Suicide could be, but I don't know. It really does come down to what you seek to measure. I think it would be useful to know if lifespan, correcting for non-heritable factors (that we know of), correlates across generations. Advances in medical technology like curing a disease have nothing to do with heritable traits, but would throw off a general study like the one in the article. My ancestors may have been haters thus dying early from madness. I'm not a hater and we no longer let people inhale mercury. Correcting for this seems obvious to me.

But I'm open to correction.

Hatter, right? Not hater.

Suicide is highly correlated with mental illness, and mental illness is generally heritable. A large fraction of suicides have had mental illness diagnoses before any suicide attempt. I don't know if people have looked at the specific kinds of mental illness that lead to suicide and the heritability of those problems.

I imagine you'd have a confound with violence insofar as some areas have had a lot more violent death over the last 150 years than others. For example Germany compared to just about anywhere else. Maybe that's because of heritable factors in the population, and maybe it's caused by the geography and politics.

@Finch: Yeah, but I thought it was funnier the other way so I left it. I'll do the same thing with guerrilla and gorilla sometimes. I don't know why, but it makes me laugh.

Your suicide bit makes sense. I did not think of it in the context of mental illness, but that makes perfect sense.

Teasing out violence that is serendipity from that which is cultural would be a problem. Central Europe is a perfect case. In the 17th century they were slaughtering one another. Today not so much.

> Today not so much.

Give it time.

You can also correct for inheritable conditions you didn't inherit. Just because something is inheritable doesn't mean it has been.

Feels like a lot more noise than signal to me. People nowadays wondering about how long they might live generally aren't wondering about survival to 40, or 50 even.

Like other traits with noisy data sometimes it's more informative for answering certain questions by first looking at an extreme instead of averages. (Though in the above you're writing about the general population, of course - so a different question addressed.) Thomas Perls (et als) has been looking at the offspring of centenarians, for example -- From 2002, "Cardiovascular Advantages Among the Offspring of Centenarians" in the Journal of Gerontology. More recently from Nov. 2012, "The Genetics of Extreme Longevity: Lessons from the New England Centenarian Study" in Frontiers in Genetics. 1. http://biomedgerontology.oxfordjournals.org/content/58/5/M425.short .. 2. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3510428/ ..

Correlation is the weakest way to look at this. There are familial traits but also individual ones. Genes may offer a longer life but addictive behavior can cut it short. Cause of death is important because bad behaviors have the most significant effect on shorting life spans, but these can be readily excluded by the assessor. If your parents lived to 90, and you don't smoke, aren't an alcoholic, drug addict, adrenline junkie, consummate risk taker, have an impairment, Go ahead and save because that is your best estimate.

Can't tell much from the statistics presented in one paragraph, though they are suspect as others describe here. But the general thinking capabilities of the author are more of a concern due to this tell, "All those saving more for retirement simply because both parents are fit, healthy, and in their nineties should stop immediately." Immediately? Really? Even if he's joking that's a bit rediculous.

Comments for this post are closed