Skip to content

Teenage sailors descend on Corpus Christi

July 14, 2018

Almost 400 teenage sailors from 86 nations are competing here next week in the Youth Sailing World Championship. The ones I have met so far, from Europe, all are well-mannered and speak good English. The guys are ravenous, which is just part of being an active teenage boy.

Photo shows what young sailors looked like in Henry Matisse’s day.

Advertisements

Making America hate again

July 13, 2018

TrumpMuslimTweetsAndCrimesYesterday, I explained how different scientific studies carry quite different meaning, because of different underlying knowledge, different questions being tested, different experimental set-up. So consider a sociology study (cite) from the University of Warwick, that takes a variety of statistical looks at how Trump’s tweets on Islam relate to anti-Muslim hate crimes in the US. The only reason it makes sense even to do such a study is that we already know there is a strong streak of Islamophobia in US, that it both affects politics and generates hate crimes, and that Trump builds his campaign to appeal to it. Without that, this would just be an exercise in looking for spurious correlation. Because we already know that, there is the interesting question of whether and how communication from their leader affects the behavior of Trump supporters who are Islamophobic.

It might be that his tweets incite his followers’ Islamophobia, causing them to commit more hate crimes. It might be that they serve as a cathartic valve, that given the comfort of a leader who expresses their animosity they don’t need to act themselves. It might be neither. So it’s worth taking a look at the data and seeing what is there. Given the nature of what’s being studied, the fact that the data is entirely observational, and that external events affect both Trump’s tweets and his followers behavior, that statistical relationship only goes so far:

While highly suggestive, a major limitation of this research is that it is correlational not causal. In order to prove that Trump’s tweets truly caused an increase in hate crimes, the researchers would need to run a carefully controlled experiment showing that exposure to such messages leads people to act out in hateful ways. This research also does not prove that Trump’s tweets cause people to hold prejudicial views towards Muslims. In fact, the researchers themselves do not seem to believe this. Instead, they point out that their findings are consistent with the idea that Trump’s presidency has made it more socially acceptable for many people to express prejudicial or hateful views that they already possessed prior to his election.

Some conservatives will hate this research. But they should ask themselves why? Those doing it understand its limits. In the case of Scientific American, the reporter understands its limits. I suspect what they really hate isn’t the research itself, but some of the underlying knowledge that raises the question. They would just as soon pretend that there is no Islamophobia in the US, and that Trump doesn’t campaign on it.

Studies, statistics, Bayes

July 12, 2018

The fact that many scientific studies are wrong now generates regular press. Which might lead the casual reader wondering why they should pay any attention at all to scientists and their research.

What that misses is that no piece of research stands on its own. That research I earlier discussed on coffee and longevity makes sense only in the context of an understanding of biology generally and lifestyle studies specifically. Every individual is biological different. From their own life history as well as from genetic variation. There is no universally optimal diet or lifestyle. Virtually every food or exercise or life practice that is generally beneficial will be problematic for individuals, sometimes because of specific medical issues, sometimes just from individual quirk. Nor is there any optimal diet or lifestyle for an individual. Just about everything that helps you in some ways hurts you in some others. And no one knows what new medical problems their future will bring.

One purpose of randomized trials with controls is to increase the likelihood that the experiment shows an effect of the thing being studied, rather than the effect of biological variability. Which alas then buries any tie between that variability and the thing studied. If 75% of white lab rats have a combination of genes that cause them to live longer when fed caffeine, while 25% die quicker, the randomized trial may show you that there is a “general” benefit to caffeine, while hiding that very connection.

It’s harder to randomize people, who generally will live the way they want. When someone educated in biology reads about the study showing British coffee drinkers live longer and the study showing the French who eat more processed food get more cancer, one of the first doubts that pops to mind is that may merely reflect something peculiar about which Brits drink coffee and which French eat processed foods! Or not so peculiar: those habits likely have some correlation with social class. Medical studies sometimes try to compensate for that, but of necessity resort to rough measures.

All that is quite different from reading a physics study on the latest measurement of mass of the top quark. There, the relevant background knowledge is that a top quark is a top quark is a top quark, rigorously adhering to the laws of quantum mechanics, each characterized only by a few parameters. The statistics relevant are due to the probabilistic nature of QM and due to experimental set up, not because some top quarks spent too many years in hydrogen atoms where they acquired bad habits and gained some unwanted mass.

David Papineau describes how Thomas Bayes’s posthumous paper on probability advanced our understanding of the role of background knowledge. I’m skeptical that what we bring to the reading of any paper can be formalized as a set of prior probabilities. For example, what prior do you give to the top quark adhering to the Schrödinger equation? How many 9s does that have? In contrast, the prior for whether a previously untried surgery will be effective is dicey. But there is no way to assign a precise probability.

Yet, I don’t agree with Papineau that significance tests are useless:

One of the great scandals of modern intellectual life is the way generations of statistics students have been indoctrinated into the farrago of significance testing. Take coins again. In reality you won’t meet a heads-biased coin in a month of Sundays. But if you keep tossing coins five times, and apply the method of significance tests “at the 5 per cent level”, you’ll reject the hypothesis of fairness in favour of heads-biasedness whenever you see five heads, which will be about once every thirty-second coin, simply because fairness implies that five heads is less likely than 5 per cent. This isn’t just an abstract danger. An inevitable result of statistical orthodoxy has been to fill the science journals with bogus results.

The issue lies largely in expectations. Should surgeons publish the positive results of a new surgery? Sure. How else are they going to communicate what they are trying? Should the journal insist at least on some statistal significance test, to weigh against a result that is pure chance? Probably. Should the rest of the medical community then take that as a confirmed finding? Of course not! Does that mean we want journals full of bogus results? In a sense.

If researchers are going to communicate on the edge of experiment, some results they produce will be bogus. No matter how much we improve methodology, there is inherent uncertainty in research. While I have some sympathy for the notion that scientific instituions provide too much incentive for novelty and too little for replication and validation, there is another aspect to the problem. Science journalists are not good at distinguishing different kinds of research and qualifying their results against background knowledge. Published papers are more like people than elementary particles, varying widely along diverse dimensions. Too many science reporters treat them all as the same. They need to be better at helping their readers suss out the differences between different kinds of results. “This is the first result for an experimental medical treatment that may, like many, stumble in future experiment.” Versus: “This is the latest in a long chain of related findings backed by deep theory.” And every variation between and besides.

The painting is a portrait of Richard Price, who hobnobbed with Benjamin Franklin and Joseph Priestly, who mentored Mary Wollstonecraft when she was young, who was an early supporter of American independence, and who published Bayes’s famous paper after his death. And who, unlike Bayes, had a portrait painted.

Our plastic problem

July 11, 2018

If you are a sailor or live near a shore, you likely are well aware of the ever growing amount of plastic debris in the oceans. Big things being big, fishing nets and gear make up 46% of the Great Pacific Garbage Patch. (Cite.) But small things are more numerous, and when you take away the big stuff:

Microplastics accounted for 8% of the total mass but 94% of the estimated 1.8 (1.1–3.6) trillion pieces floating in the area.

It’s hard to know which is more damaging. It is the small stuff that enters the marine food chain:

Microbeads are commonly white or opaque in colour, and research has found microbeads to be commonly mistaken for plankton by many surface feeding fish species. Ingestion of plastics by aquatic organisms is one of the major deleterious environmental impacts in the marine environment. Due to their small size and presence in pelagic and benthic ecosystems, contaminants associated with microplastics are potentially bioavailable for many organisms. Persistent organic pollutants sorbed onto microplastics can accumulate at concentrations several orders of magnitude higher than in ambient seawater. A growing concern related to microplastics is that they can also enter the human food chain through ingestion of fish, shellfish and filter feeders.

Because they are such a tiny part of the problem, a recent Bloomberg editorial dismissed the recent move to decrease the use of plastic straws. Vox is more sympathetic and goes into more detail. I am not as hopeful as that author about a spillover effect.

The real problem is that most plastic doesn’t decompose. The common materials throughout human history — wood and fibers — all came from plants, and all are readily recycled by sun, rain, worm, and fungus. Much to the chagrin of wood boat owners and archaeologists. Plastic replaced much of what formerly was made of those materials, because it is cheap to fabricate. Alas, plastic will be with us forever. What we really need is a material science revolution that provides us decomposable materials that are nearly as cheap to fabricate as plastic.

No right to literacy?

July 10, 2018

Though literacy was lower when our nation was founded and the push for public schools still in its salad days, I suspect those Enlightenment men will be turning over in their graves at the latest court ruling from Michigan denying any minimum standard to public schools there. Texas is in need of such lawsuit, though it would have no more chance of prevailing here. I am reminded of one of the causes that was listed in the Texas Declaration of Independence from Mexico:

It [the Mexican government] has failed to establish any public system of education, although possessed of almost boundless resources, (the public domain,) and although it is an axiom in political science, that unless a people are educated and enlightened, it is idle to expect the continuance of civil liberty, or the capacity for self government.

Coffee and parrots, redux

July 9, 2018

Having just posted on animal intelligence, I now come across a new study on parrot brains, comparing the organization of their brain to the mammalian brain. It hardly is surprising that their intelligence evolved differently yet requires analogous brain structures. (Cite.)

And having just posted on the arterial benefits of caffeine, there now is a a new study suggesting that coffee drinkers live longer. (Cite.) The correlation is unrelated to caffeine, since it holds also for those who drink decaffeinated coffee. The study was prospective, followed a half-million people in Britain, and controlled for a variety of factors. That said, there’s a likelihood that it isn’t the coffee, but something to do with people who drink coffee. See my coming post on Bayes.

On the other hand, parrots like their coffee (youtube). And they have long lives…

Signs of the times

July 6, 2018

One of Britain’s white nationalists is barred from Sweden. Want more? A Trumpista tells the ghostwriter of one of Trump’s books that actually it was Trump who wrote it: