site banner

Troubles with life expectancy gender gap

The gap

This gap between the average male and female life expectancy of given population group is alternately labeled as Life Expectancy Gender Gap (LEGG) or Gender Gap in Life Expectancy (GGLE). Going forward I will be using the LEGG acronym.

The expectancy

There are basically two types of life expectancy. In historical context, we usually refer to the Cohort Life Expectancy. We track a group of people born in a particular year, many decades ago, and observe the exact date in which each one of them died. Then we can calculate this cohort’s life expectancy by simply calculating the average of the ages of all members when they died.

It is of course not possible to know this metric before all members of the cohort have died. That is why when talking about the present and the future we use Period Life Expectancy. This is an estimate of the average length of life for a hypothetical cohort assumed to be exposed, from birth through death, to the mortality rates observed at one particular period – commonly the previous year. Estimates of life expectancy of the current generation, which are also used in the calculations of indexes like the Human Development Index or Gender Development Index, are of this Period Life Expectancy type.

There is a corollary: because we are judging the existing generation based on mortality rates of the previous year only, whatever happened to that generation before that year is not taken into account. So let's say. if there was a recent deadly pandemic that affected one gender disproportionally, and this pandemic ended and is not affecting the mortality rates of the previous year, there will be little evidence of said gender disproportionally in the current life expectancy estimates.

The lost years

The first important thing to know about the LEGG is that its impact is, without an exaggeration, enormous. Let's take for example the US, with a LEGG of 5.8 years at the average predicted age for men and women 73.5 and 79.3 years respectively. Do you see the enormity? You don't, do you.

Ok, let's put things into perspective - how do you measure an impact of early death? With Years of Potential Life Lost (YPLL). This is an estimate of the average years a person would have lived if they had not died "prematurely". It is usually reported in years per 100,000 people and the reference, "mature" age should correspond roughly to the life expectancy of the population and is now usually given as 75 years.

Now, men and women in the US lose some 8,265 and 4,862 potential years per of life per 100,000. Given the population as 332 millions, men lose some 5,648,980 more years of potential life than women. Do you see the enormity now? Not yet?

During the roughly 3.5 years of WW2 the US lost 407,300 military and 12,100 civilian lives. With an average life expectancy back then 68 years, and a guestimated age at the time of death 21 years, every killed American lost some 47 years and the US as a whole lost some 5,640,000 potential years of life every year of the war. Do you see the enormity of the LEGG now? I think you do.

The causes

The second important think to know about the LEGG is that nobody seem to care. Biologists, statisticians, politicians, Wikipedians - not even men's rights activists - nobody seems to be franticly looking for the causes or proposing policies to stop this haemorrhage of men's lives. Let me paraphrase what Wikipedia has to says about it:

It is the life style, men drink more and smoke more and eat crap. And it is also the biology, men lack the double X chromosome, we see this across all mammalian species, plus male babies and boys dies of diseases much more than girls.

Speaking of Wikipedia, it has dedicated pages for many things, including the Orgasm gender gap, but it does not have a dedicated page for the LEGG.

To my surprise I have not been able to find any further information, neither on biology forums, nor on Google Scholar. Studies usually focus on one cause or divide the mechanisms into social and biological but there our knowledge seem to end.

At this point I was so intrigued that I decided to do some "research" myself. My first observation was that there is a great variance between developed countries with similar GDP and life expectancy, suggesting that a large part of the gap is not biological. Example:

  • 2021 Norway - LE: 83.16 years, LEGG: 3,0 years
  • 2021 France - LE: 82.32 years, LEGG: 6,2 years

Next, I knew where to find Eurostat data on causes of death - unfortunately only from 2010 - and I filtered out everything mechanical: suicides, assaults, accidents and drug and alcohol overdoses. The LEGG shrunk significantly:

  • 2010 Norway - all LEGG: 4.54 | non-mechanical LEGG: 3,51, decrease by 29.5%
  • 2010 France - all LEGG: 7.14 | non-mechanical LEGG: 6.19, decrease by 15.3%

Then I was curious how much of the LEGG is caused by mortality differences of infants and children so I calculated non-mechanical LEGG at 20 years, as opposed to LEGG at birth. The difference is negligible:

  • 2010 Norway - non-mechanical LEGG at birth: 3,51, non-mechanical LEGG at 20: 3.37, difference: 3.8%
  • 2010 France - non-mechanical LEGG at birth: 6.19, non-mechanical LEGG at 20: 6.07, difference: 1.7%

Next, I did one more napkin calculation. Assuming that smoking reduces the life expectancy on average by 10 yers and smoking rate among French men and women are 0.349 and 0.319 and smoking rate among Norwegian men and women are 0.17 and 0.154, I reduced the LEGG further:

  • 2010 Norway - all LEGG: 4.54 | non-mechanical, non-smoking LEGG: 3,35, decrease by 35.7%
  • 2010 France - all LEGG: 7.14 | non-mechanical, non-smoking LEGG: 5.89, decrease by 21.2%

Of course this does not mean the reminder is caused by biological factors. There are drugs and alcohol, there is a meat consumption and overall life style. Man also do more paid work so there is work related stress and exposure. It should not be a rocket science to isolate these factors, actually, it would amount to a very cool paper with plenty of citations. So where is this paper?

Actually, I found one piece of information: Causes of Male Excess Mortality: Insights from Cloistered Populations, the abstract talks about 11,000 Bavarian monks and nuns living in "very nearly identical behavioral and environmental conditions" with nuns having only a "slight advantage" in life expectancy - whatever that means, I can't access the paper itself. This of course only applies to men and women who already survived into their teens or twenties, but as we saw above the contribution of different child mortality to LEGG is negligible.

The bad and the ugly

FYI, about 80% of suicide victims are men and suicide is the second leading cause of death in middle aged men only after car accidents. Also, 90% of workplace accidents are men - constructions, mining, trucking, heavy industry, you know - and even though the total numbers are too small to meaningfully influence the LEGG it does not cover exposure to chemicals, hard labour or health impact of night shifts.

Some social and biological mechanisms out there are causing men to lose life equivalent to WW2 every year. We should be creating policies to reduce this loss but we don't. Why? We know that men are far less likely than women to visit a doctor. Where are the public health campaigns and safe driving campaign targeted at men specifically?

Could it be the case of "you grow what you measure"? UN's Gender Development Index that measures gender disparity in achievements between men and women very quietly removes 5 years from the LEGG in it's calculations, arguing that men living 5 years shorter is necessary biology. The Global Gender Gap Report published annually by the World Economic Forum does something similar, arguing that if women live at least 6% longer than men, parity is assumed - but if it is less than 6% it counts as a gender gap.

As a corollary, women is Norway living only 3 years longer than men is interpreted as oppression.

Jump in the discussion.

No email address required.

Cool post, but

It should not be a rocket science to isolate these factors, actually, it would amount to a very cool paper with plenty of citations. So where is this paper?

There are plenty of papers on the life expectancy gender gap in general. This one for instance looks at why the gap has narrowed in Sweden by looking at the contribution of different causes. They find "External causes of morbidity and mortality" (i.e. accidents, self-harm, assault) explained 15.1% of the gap in in 1997 and 21.2% of the gap in 2014 - conclusions that seem to broadly match yours.

However, I take your point about the lack of such papers. Given the political leanings here, I suppose commenters will point out that this hasn't stopped similar gender pay gap analysis. There is truth to that, so allow me to point out factors that don't boil down to "it isn't fashionable in this zeitgeist."

Such papers on the gender pay gap are almost always just glorified linear regressions. This makes the causal claims questionable, but also makes it clear why such papers are easier to write than a mortality gap paper would be. To do pay analysis, you can just survey 10k people, which will give you about 10k pay datapoints. You ask those people a battery of standardized questions. You plug-and-chug in a linear regression model. Done. If you tried that for mortality, you'd survey 10k people, which will give you ~150 death datapoints after a year. That's not enough to do any kind of definitive regression with lots of controls.

So, for mortality research, you will realistically need hundreds of thousands of people, which means you are realistically using either medical records or government records. If the latter, privacy and lack-of-standardized-questionnaires will make it a struggle. If the latter, privacy and lack of asking-relevant-questions will make it a struggle.

And, so, to do This Paper on the life expectancy gap, you are stuck integrating results from multiple other papers/fields. It's easy to subtract specific causes-of-death (as you did and the paper I linked to above did). But you want more, you want to control for smoking, drugs, alcohol, "meat consumption and overall life style". Here you run into serious correlation≠causation issues. Sure, the link you used claims that smokers live 10 years shorter, but there is no RCT - how much do we trust that as a causal claim? Ditto for the other things. Another issue is that the linear-regression-doer can include interaction terms, while the literature-review-using-various-sources paper cannot.

I'm not saying there is no impact on what-is-popular regarding what is published in academia, but there are serious differences between researching the gender mortality and pay gaps.

I admit that I am not an expert on Google Scholar but I was struggling to find anything useful in those papers. However the Swedish paper you found is super relevant and it is on the first result page so I apparently was not looking hard enough :)

As for your argument that studying the LEGG is difficult, I have my doubts. I am in no way a statistician or sociologist, but I was able to do a decent analysis in my spare time in Excel. A team of trained professionals, who know where to look for existing datasets and what statistical tools to use should be able to crack the problem "in no time".

As usual men are doomed as they are in large amounts, simps or to some extent, endoctrinated self-hating misandrists. Note however that biologically speaking there are reasons for the shorter lifespan of men though, the biggest one probably being height. height is one of the strongest predictor of low life expectancy. Basically anabolism has costs, including some immuno deficits in terms of resource allocation, possibly increased oxdative stress and cancer risk. It might be that short and non-high bmi men live longer than the average woman though? But mens body also has advantages, for example increased brain volume means men are less prone to neurodegenerative diseases, especially for example, 2 times less chance of developing multiple sclerosis (although the specific reason here being that testosterone increase myelin production)

Now as a reminder, you can increase significantly your lifespan via skq1.

However the biggest omission in your blog, and the question that leaves me most curious, is a comparison of women/men not lifespan but healthspan I already talked about dementia but what about sarcopenia (should advantage men too) and what about chronic hospitalizations rate per age? I suppose the gap of ill men (especially cancer) is even bigger than the gap of prematurely dead men.

Also a question no ones asked before, are baby boys more often victims of baby shaken syndrome by their parents?

height is one of the strongest predictor of low life expectancy.

Is it? Source?

Larger animals tend to live longer because metabolism scales slower than size.


I’ll take that bet. No way that squirting some random antioxidant into your eyes has a real effect on aging.

are baby boys shaken more?

Yes. More ER visits, and more money spent on each visit, if I’m reading this right.

For the cloister study, you can go to sci-hub at the Russian TLD and search for to find it.

I had a more substantial comment, but it got eaten. Alas. I'll just point out that the LEGG varies significantly among different subpopulations (e.g. working class vs professional class), and it is in large part driven by maladaptive (at least with respect to longevity) traits exhibited by men in those specific subpopulations.

Any chance you could remember what was in your eaten content?

What do you mean by maladaptive traits? lifestyle choices?

And thanks for the sci hub reference, I found it!

Brief sketch: comparison of the LEGG in humans vs other mammals (takeaway, chimps have a much larger one, at 25%). Different interventions (castration/neutering; being in captivity) substantially decrease the LEGG. Human eunuchs live longer than regular men of same class. Then a bit about how you can't separate biology/society; hemophiliacs have a much lower lifespan than non-hemophiliacs "in the wild" but we have interventions that eliminate that gap, could be done for men too if we prioritized it; different subpopulations of humans have wildly differing LEGGs, with more egalitarian (in the class sense) societies tending to have much smaller gaps. Lastly some speculation that it's not primarily because of either biology or healthcare access but the incentives created by gender roles, which makes it sadly much harder to fix.

Different interventions [...] being in captivity

This suggest the LEGG mechanism is not biological

Human eunuchs live longer than regular men of same class.

My guess would be eunuchs have way fewer accidents than men with testosterone? :)

Then a bit about how you can't separate biology/society

That is of course true

with more egalitarian (in the class sense) societies tending to have much smaller gaps.

Source? I think it depends on what you mean by egalitarian, but I don't think this is correct.

which makes it sadly much harder to fix.

Does it? What evidence makes you think so?

Maybe greater male variability hypothesis could explain some amount of LEGG. It is true that many men die quite early in life, but those who live to an old age enjoy better physical health and cognitive abilities then women in the corresponding age on average.

Greater variability on it's own does not explain the skewing of the life expectancy curve to one side. If it was greater variability on it's own, the long lived men would cancel out the short lived men and the effect of life expectancy would be zero.

I don't think that's true, as there seems to be a cap on human longevity -- so it seems it's more like you have a larger population dying potentially decades earlier, and then 'the rest' living essentially to near the cap, which seems to have a LEGG effect.

I think you are right.

If the distribution of lifespans was normal (or, at least, symmetric) this might be believable, but since it’s not I have to ask: why does increased variation have to keep expected value the same?

For example if I move some people from the 50th percentile to the 60th and an equal number to the 40th, that’s going to change the expected value.

There’s no particular reason to expect symmetrical variability in the underlying factors causing life expectancy to lead to no expected value change in life expectancy — E[f(X + Gaussian noise)] isn’t E[f(X) + Gaussian noise].

I think you are right.

Lot easier to die at 20 than to die at 140.

Not necessarily. As a general rule, our biology is the way it is for a reason, and the great majority of large variations from the mean in biological traits is dysfunctional. As such, a greater variability on basic biological traits can easily lead to a lower average life expectancy. In terms of statistics, you should be careful when trying to estimate the effects of differences in upstream variables on downstream measurements. It's not all nice and linear!

Edit: Though I also disagree with the parent post. Men generally live shorter, and even the extreme cases of old age is dominated by women. Live expectancy is generally not really strongly evolutionary selected, and even less in men than in women.

What would be the evolutionary advantage of longevity among women? Helping with grand children and labour? Is there any research around this hypothesis?

As I said, modern live expectancy is imo generally not strongly selected. That includes women.

If you look at biological differences between men and women, men are clearly optimized for warfare, hunting and recovering from serious injuries, while women are optimized for childbirth and -rearing as well as general health. There are for example some studies showing differences such as men and women actually recovering at different speed from serious injuries, or the other way around, women being sick less seriously and recovering faster from common diseases.

Furthermore, men are much higher variance because a single men can father literal thousands of children, and many real-life examples of those did not actually enjoy a particularly long life expectancy. On the other hand, women can at most have a number of children in the low double-digits and critically, need to look after them for 15+ yrs after their birth. Therefore, as a women there is little benefit in being exceptional and a lot of benefit in just being boring healthy and fertile for as long as possible. Remember, historically adults lived ca 60 yrs, which is only ca 20 yrs after fertility. Healthy lifespan was more in the fifties. So (healthy) longevity significantly past childrearing age wasn't really a thing. Yes their oldest children may have children that the grandparents can care for, but that's not the primary reason for their lifespan; The primary reason is that they still may have children themselves that might need caring. Also keep in mind that evolutionary selection usually works on smooth curves and generalities, not on perfect sharp optimums. Even if the theoretic optimum female lifespan were to, say, live exactly for 15 yrs after the birth of your last child and then drop dead instantly, evolution instead just selects for better general health until you have a 90% chance to survive that long. This may cause you to live 5 extra "unnecessary" years on average, but it's still worth it to avoid dying prematurely.

In my opinion, longevity past 60 is primarily a result of modern healthcare, and as such not evolutionary selected directly. The female-male gap is probably mainly due to better female general health, which itself is partially a result of biological and social differences, which are to different degrees a result of evolutionary selection. For an example of the latter, higher male risk tolerance is technically a social difference, but also an obvious result of evolutionary selection (and most likely rooted in hormonal differences, so partially biological as well).

Btw though, there is also some research around grandparentry being positively selected, though I do not consider it compelling. It's imo similar to the gay uncle hypothesis, superficially plausible but it doesn't really work out once you look into the details. For example, I just looked into an old colleague of mine who showed us a game theoretic model in a meeting ca 7-8 years ago where under certain assumptions longevity significantly past fertility was positively selected. Apparently he did not end up publishing it, and I strongly suspect it's because the assumptions were very questionable.

I like your reasoning. The question is, where are the studies? At minimum, we should be able to compare non-drinking vegans from the same background working the same job.

modern live expectancy is imo generally not strongly selected.

Our cardiac longevity is a pretty noticeable outlier. Most animals get around a billion heart beats per life, and if anything the trend is downward for longer-lived animals, but humans get nearly 3 billion. If we were closer to average it wouldn't just be impacting the ability of grandparents to care for juvenile grandchildren, it would be affecting the ability of parents to care for juvenile children.

On the other hand, humans are also an outlier at endurance running. Perhaps all the selection for cardiac performance came via that, and any longevity benefit was a side effect?

As I wrote, I think life expectancy and fertility was selected in the ancestral environment, which resulted in ca 60 / 40 yrs respectively. It's just the modern longevity way past 60 with no or very little increase in fertile years for women which I think is not strongly selected and so differences are primarily due to the result of other, actually selected attributes such as the immune system, risk tolerance or as your example, cardiac health.

Uh, is that latter part actually true? The list of oldest people in the world is consistently dominated by women. One would assume that this correlates strongly to physical health at advanced age.

This is undoubtedly true and there must be some biological advantage for women, but it is also possible that older women just live longer in poorer health. I read an article in an Austrian newspaper that claimed that men who live to a ripe old age have better health due to their greater mass, including muscle mass and cognitive reserve caused by a larger brain on average. Now, when you asked me, I cannot find the particular study and a quick googling returns non-conclusive articles. This and this articles seem to support my thesis, but this one is in stark contrast. By far, I anecdotally believed that this is true and I feel too incompetent to take a strong position.