Statistical Methods Links, 5/17/2025
Inquisitive Bird on Assimilation; Christopher J. Ferguson on correlations with mental health; David A. Shaywitz on Super Agers; Steve Stewart-Williams on an IQ heritability puzzle
When it comes to the Age of Mass Migration (1850 to 1920s), it is often ignored that large shares of European immigrants did not “assimilate”—they returned home because they couldn’t assimilate. At least 25-40% of European immigrants returned to their home, but serious estimates as high as 60-75% exist for some decades (Bandiera et al., 2013). Return migration was also negatively selected (Abramitzky et al., 2019). Large shares of migrants could not successfully integrate, and those were typically the ones who returned. What at the surface appears to be widespread successful integration is thus to some extent a manifestation of survivorship bias.
Christopher J. Ferguson writes,
As I wrote about before, the CDC found that, even in essentially bivariate correlations lacking proper statistical controls, relationships between social media and youth mental health were trivial, basically at the level one would expect from statistical noise.
By contrast, another report from the CDC did discover what appears to be related to youth mental health issues: basically, their families. The CDC asked youth a series of questions about adverse events in the household…being abused, neglected, or having parents with mental health problems or who had been incarcerated. Compared to the wispy social media effects, relationships between these adverse childhood events and self-reported mental health outcomes were robust.
Led by the cardiologist Eric Topol, the researchers hoped to identify the genetic factors associated with healthy aging. To their surprise, they found little in the DNA that stood out. They did, however, notice several striking traits. Compared with their peers, the disease-free subjects were generally thinner, exercised more frequently and seemed “remarkably upbeat,” often with rich social lives. These observations encouraged the research team to think about longevity (years of life) and healthspan (years of health) more broadly. In “Super Agers” Dr. Topol shares the results of this intellectual exploration.
The book review never discusses the issue of causation. But my first thought is to ask how you determine this in the absence of a controlled experiment. Suppose that I just happen to have had better health experiences than another 71-year-old. As a result, I exercise more, am thinner, am more upbeat, and have more of a social life. And if I turn out to be a “super ager,” this will not be because I exercise, am thinner, etc. It will be because I just had better health luck at age 71.
Steve Stewart Williams writes,
The heritability of IQ increases with age, while the effect of the shared family environment dwindles. In other words, when it comes to cognitive ability, nature becomes more important as we get older, nurture less.
He offers an explanation for this finding. But I immediately think “measurement error.” Measurement error of IQ is probably higher in young people than in old people. And measurement error reduces correlation. Since “heritability” is measured by correlation, it will be under-estimated for everyone due to measurement error. But the under-estimate for young people will be larger than for old people.
substacks referenced above:
@
@
@
"To their surprise, they found little in the DNA that stood out."
I'm betting it was more 'disappointment' than 'surprise'. A lot of nice things like longevity, IQ, and health seem to go together and we've known for a long time that, for the genetic components feeding into these characteristics, the aggregate is the combination of lots of little contributions across the genome, and a lot of that is correlated to lower levels of error in alleles that are otherwise common in the population, that is "low genetic load".
It's a hobby of mine to look up how long famous intellectuals lived and it's stunning how many got past 90, with it being a frequent surprise to learn that someone born in the Depression era is still alive. As a random example, I recently heard an old song from the late 1950's from Tom Lehrer (famous satirical musician, music and math prodigy in his youth, invented the Jello Shot while working for NSA). Dead? Nope, he just had his 97th birthday.
Just before the paywalled part, the WSJ notes that the single biggest factor in aging is exercise, “the single most effective medical intervention that we know.” While it might be disappointing to Arnold that the article doesn't repeat the oft-repeated Correlation does not prove Causation, in his comment he also seems unwilling to note the understated truism that All Causation results in Correlation.
Anybody looking for causation should start with correlations. Not yet proving that some correlation is a causal one, doesn't mean it isn't. Lots of true things aren't proven truth. I do wish there were more college graduate alumni studies -- and even think Fed student loans should require collection of more annual info as taxes are filed yearly. Including weight.
All human results are based on genetic & epigenetic influences, environmental influences, individual behavior choices, and luck, or everything else. Studying the survivors after 80 won't show many of the genetic mistakes that cause/ influence earlier death, because those folks didn't survive. Similarly bad environment, and even bad individual behavior. In this paradigm, what does it even meant to say "better health luck at age 71"? Exercising at Israeli dancing, or not, is not an issue of luck.
For me, I was skiing a couple years ago, for exercise. Then I paused on the slope and a young woman barreled into me, damaging my eye bone, teeth, face, back, & knees. I think of that as bad luck -- good decision to ski for exercise, but bad outcome. Now I exercise (almost) every day, for my back, and usually also a 40 minute walk or so.
Plus I read Arnold Kling & his links a lot to keep up my good spirits, and to learn new things.
You can't teach an old dog to do new things ... so if you're learning new things, you're not an old dog.