I don’t get the comparison. It would be like saying it’s okay if an excel formula gives me different outcomes everytime with the same arguments, sometimes right, but mostly wrong.
It would be in the early 2000s. Let me try AI. Found it. What an age we live in.
You Can Count on Me - 2000
Sammy is a single mother who is extremely protective of her 8-year old son. She is satisfied with living in the small town she grew up in and working in a local bank. When her brother Terry visits he fits the void in the life of both her and her son. Temporarily free of the constraints of single motherhood she begins to break free of her normal routine. In a string of traumatic events Sammy is torn between helping her brother and her maternal instinct to protect her son from getting hurt.
From the paper: For example, the Centres for Disease Control generated an independent estimate of average longevity across the USA: they found that Loma Linda, a Blue Zone supposedly characterised by a ‘remarkable’ average lifespan 10 years above the national average, instead has an unremarkable average lifespan29 (27th-75th percentile; Fig S6).
The CDC looked at average life expectancy in Loma Linda across all demographics. Purely geographical and on average.
The blue zones focused on the greater longevity specifically of Adventists in Loma Linda.
It wasn't a question of whether living inside the municipal boundaries of Loma Linda automatically conferred some special health benefits -- clearly it doesn't.
It was, "Why is there an unusually high concentration of outliers living here, and what behaviors cause them to live longer than average?"
There are ~9000 Adventists in Loma Linda. This is two categories you can split people into, then intersect.
There are 330,000,000 million Americans. There are likely millions of categories people can be split into. Just for fun let's say counties (6000+) then any of a zillion other cross items (left handed, blue eyed, above average height, smells like butter, etc., etc.,Etc.) Say we find 10,000 of these categories.
Life expectancy is decently modeled as a gaussian with std deviation 8 years. A 10 year excess is a z-score of 1.25, and 10% of samples will be at this point.
The odds of TONS of subsets of size 9000 of the 330,000,000 people that can be found in the same pair of county+trait from the 600,000,000 pairs is nearly 1.
Thus the Adventists in Loma Linda are far more likely to be one of these many blips that have zero causal power than they are to have special life sauce. Finding them is merely an artifact of being able to filter data, not a special power of the objects.
Or a simpler way: pick two binary traits, split the 330m Americans into 33,000 chunks of size 10,000 where each group has all in one of the four pairs of traits, and you would expect (more or less - there is some more math to do here) that 10% of these groups has average lifespans over 10 years, i.e., 3,300 of the groups are the same as the Loma Linda Adventists.
If "no magic is needed", then why don't you - or someone else - name, say, 5 more such groups/chunks with their exact characteristics? It seems that it is not that easy to find them... and yet someone found such a group in Loma Linda...
It's much easier to find a group of Adventists that have an above average lifespan because Adventists form a community. People with blue eyes or people who are left handed who live in the same county don't all know each other and discuss their statistically insignificant longevity
Doesn’t county or town/cities (doesn’t know the diff in US) counts for "communities", and aren’t those separated in groups while doing national stats? The dice rolling groups are obviously here and have probably been surveyed many time, didn’t they?
I just gave you the math showing such groups are common, with no need for anything special. It's simply math. It as simple as: if I flip a coin long enough, I can find a run of 10 heads, or 100 hears, or a trillion heads.
The number of Americans and the number of ways to organize them is large enough that, just by chance, there will be many that have a 10+ years lifespan for no other reason than we simply have zillions of ways to split people into groups.
The math I presented give you the direction to compute such things. Learn enough math to solve the expected number of such groups, and you will be surprised.
To show one such group is anything other than statistical chance takes far more science and study and analysis than just saying "Look group has desired thing Y all we have is to repeat what the group did!"
> It seems that it is not that easy to find them
It's trivial to find such groups - medicine finds them all the time. Pick any medical result X that is expected to add Y years to life, pick some population center, pick those in the center with the habits/genetics in the study, and voila, you get yet another mystical group with magical life properties.
Except it's not magic. And it will happen with certainly without there being any underlying cause simply due to statistics. Medicine tries to remove the pure randomness of the result and demonstrate a causual relationship, but that is hard and not always done. They do this extra work because they know that stuff like the above happens so often purely randomly.
Simple example: [1] claims (I have not dug into the study, but it is likely well done) that 8 habits (eat healthy, exercise, good body weight, not too much alcohol, not smoking) would add 20ish years to life expectancy. So, go to a big city, find those in this group, and you'll get likely several thousand of them.
And now woo hoo! 24 years!
And for special effect, pick the subset that intersects yet another silly variable, say has red hair, or was bullied as a child, and now you too can get headlines that will spread like this one: "The 8 traits that make readheads live 20 more years!" "Bullied kids can do this one simple trick and outlive their tormentors!"
But this is simply nonsense. There is science, there is causality, and there is statistics, and not being able to disentangle them leads lots of people to post voodoo as if it's not simply random chance.
I don't think that is right. In the Blue Zones marketing material, they characterise Loma Linda's 9000 Adventists, who make up 40% of the population, as living a decade longer on average. That is the claim being investigated.
This claim is hard to reconcile with the CDC's official numbers which show a typical life expectancy for the entire area, unless living next to Adventists somehow lowers the life expectancy for the remaining 60% of the population, which would be far more interesting.
Sure Buettner does focus on the older people of the community by interviewing them, but that does not generalise to the claim of the book (or the website to this day) that this community has a high life expectancy, which is shown to be false by the corrected statistics. This is known as a "population fallacy".
By focusing on the older people only in such a small population, he is introducing selection bias and survivorship bias. Moreover, he did not control or compare studies. I believe there are more than one Adventist community in the US, yet those are not Blue Zones somehow?
> It was, "Why is there an unusually high concentration of outliers living here, and what behaviors cause them to live longer than average?"
Blue Zones LLC also provided a set of answers to that question, and one of those answers (“drinking 1-2 glasses of wine per day”) is clearly not true in this case.
And honestly, it’s just Bayesian statistics—if they present 5 data points, and 4 of those data points are floating somewhere between data errors and fraud, then odds are, that last data point is flawed somehow as well. Certainly they would need to do some extra work to prove that it isn’t.
So first it was Sardinia, Okinawa, Ikaria, Loma Linda. Then it's not even Loma Linda but specifically Loma Linda Adventists. That looks like XKCD-level p-hacking
Yeah, if the point is really about Adventists, I think it's better made with statistics on them. Ditto teetotalers or vegetarians (Adventists are often both). Or if it's about studying individuals with long lifespans, then great, let's do that.
Who's claiming that living inside the boundaries of such zones would confer health benefits?
The paper is pointing out that if you actually look at the data there is nothing remarkable about the region's average lifespan (actually lower than the entire country of Japan), which is what's being discussed here.
> The paper is pointing out that if you actually look at the data there is nothing remarkable about the region's average lifespan
That's my point -- the region's average lifespan is irrelevant. It's only relevant given the misconception that Loma Linda itself has some special properties of rejuvenation.
But that doesn't mean it's not a longevity hotspot. Even if the average lifespan there were lower than normal -- say a large number of unhealthy people lived there -- it still wouldn't negate that, if an abnormally high number of healthy centenarians also live there.
Loma Linda residents do have some of the highest lifespans in the world. Not on average -- but that wasn't Buettner's point. His point was that there's an unusual number of long-living outliers there.
Big enough forest - say, eight billion or so trees - there'll absolutely be 100 weird trees in a spot somewhere.
If Adventists have cracked the code for longevity, you'd find their other congregations with similar benefits. Barring that, we're just p-hacking our way to a spurious conclusion.
reply