Apr 15, 2016
ssc
Read on (unread)

Links 4/16: They Can’t Link Our Dick

A links roundup examining various scientific studies, social developments, and historical findings, with a particular focus on recent research about confounding variables and replication issues in science. Longer summary
This is an extensive links post covering various scientific studies, news items, and interesting facts. Scott discusses several studies about confounding variables and replication in science, historical findings about ancient civilizations, new research on topics ranging from climate change to criminology, and various social and political developments. The post starts with methodological concerns in science, moves through various research findings and historical tidbits, and ends with news about ongoing projects and obituaries. The tone is analytical but casual, with Scott often adding his own commentary and pointing out interesting implications or contradictions. Shorter summary

[Disclaimer: None of these immediately set off alarms, but I have not double-checked all of them to make sure they are accurate. Please correct me if any are false or misleading]

Zerão is a Brazilian football stadium with the dividing line exactly on the Equator, so that each goal is on a different hemisphere.

Maybe the most important article I’ve read this year: When Confounding Variables Are Out Of Control. A new PLoS paper argues that “controlling for confounders” doesn’t work as well as we’d like. Confounders are always imperfectly measured, so when you control for your measure of a confounder, you’re only getting a portion of the real confounder, and the portion you didn’t get might be more than enough to sustain a significant effect. So many studies that claim to have gotten a result after “controlling for confounders” but which haven’t used complicated statistical techniques that nobody uses are now potentially suspect. I’ve always noticed that correlational studies that control for confounders get confirmed by experiments much less often than I would expect, and now I finally understand (some of) why.

Related: Andrew Gelman: “Let’s just put a bright line down right now. 2016 is year 1. Everything published before 2016 is provisional. Don’t take publication as meaning much of anything, and just cos a paper’s been cited approvingly, that’s not enough either. You have to read each paper on its own. Anything published in 2015 or earlier is part of the “too big to fail” era, it’s potentially a junk bond supported by toxic loans and you shouldn’t rely on it.”

Related: Internal Conceptual Relations Do Not Increase Independent Replication Success. That wouldn’t make sense if the problem was just the normal vagaries of replication, and suggests that “the influence of questionable research practices is at the heart of failures to replicate psychological findings, especially in social psychology”.

Related: A long time ago I blogged about the name preference effect – ie that people are more positively disposed towards things that sound like their name – so I might like science more because Scott and science start with the same two letters. A bunch of very careful studies confirmed this effect even after apparently controlling for everything. Now Uri Simonsohn says – too bad, it’s all spurious. This really bothers me because I remember specifically combing over these studies and finding them believable at the time. Yet another reminder that things are worse than I thought.

Wikipedia: List Of Games That Buddha Would Not Play

The lost medieval City of Benin in Nigeria had streetlights, great art, and was larger than many European capitals. It also boasted “the longest walls in the world”, beating the Great Wall of China. I’m confused why I never heard about this before – not in a “neocolonialist society covers up the greatness of Africa” sense, but in a “even the people complaining about how neocolonialist society covers up the greatness of Africa only ever talk about Zimbabwe and Kilwa which are both way less impressive” sense. Also, how did one small British expedition destroy earthworks longer than the Great Wall of China?

Some very complicated and potentially questionable attempts to ferret out all the different personality traits involved in religiosity tentatively conclude that it is directly related to moral concern and inversely related to analytic thinking, which are inversely related to one another.

Vox: You Can Finally Stop Feeling Guilty For Eating Quinoa. Apparently some people felt guilty because they thought that quinoa-eating Westerners were taking all the quinoa and then Peruvians were starving. But a new study suggests that the increased Western demand for quinoa has increased welfare throughout Peruvian quinoa-farming regions both for farmers and for non-farmers, presumably because the farmers’ increased wealth is trickling down to non-farmers.

Vox: The Most Important Foreign News Story This Week Was About Russian Tax Policy.

Did you know: when the British Empire abolished slavery, it paid 40% of the government’s total annual expenditure as compensation to slaveowners.

The prison phone system is a national disgrace. Predatory companies make deals with the government to get a monopoly on calls to and from specific prisons, then charge inmates trying to call their families rates that are orders of magnitude higher than normal. I have some patients with incarcerated family and they confirm that this is a big problem for them. The FCC has been trying to cap rates, but was recently thwarted by the courts. This seems to me like one of the clearest and most black-and-white political issues around.

Business scientists run a trial to compare promotion-by-merit with promotion-by-seniority and include random promotion as a control group. Now the results are in and random promotion does the best. Even weirder, the result seems to have been replicated. This kind of reminds me the old saying that “anyone who can be elected President shouldn’t be allowed to do the job”. [EDIT: Study uses potentially faulty computer model]

New study in Nature by leading climatologists says that the consensus is now that the global warming hiatus is real. And here are some blog posts (1, 2) explaining the result in more accessible language. Both emphasize this doesn’t mean that global warming has stopped or was never real, only that it seems to be slower now than it was before. Leading theory – complicated ocean cycles working in our favor now may work against us in the next few decades, and we should still be careful.

Thirty-five overweight people were asked to do the same amount of extra exercise. They differed wildly in how much weight they lost. Authors theorize a distinction between “compensators” and “noncompensators” with different metabolic reactions to exercise.

Contrary to conventional wisdom, Gallup poll finds Clinton supporters are more likely to describe themselves as enthusiastic about their candidate than Sanders supporters. [EDIT: Maybe some extra complications. See here]

The largest area of dry land under sea level, by volume, is the Qattara Depression in Egypt. If it were ever connected to the Mediterranean, it would produce loads of hydroelectric power and form a giant lake in the middle of the Sahara. By my calculations it would also reverse one year worth of global-warming induced sea-level rise.

Ideal Conceal is a handgun which can be folded up to look exactly like an ordinary cell phone. Nothing can possibly go wrong. [But these guys think it’s a hoax]

Old studies: Australia’s experience in the 1990s proves gun control worked. New study: Australia’s experience in the 1990s proves gun control didn’t work. I am so past the point of trying to figure this out now.

Did you know: the computer game “Ecco the Dolphin” was partially based off the work of LSD-abusing delphinologist John Lilly, who thought reality was controlled by an alien conspiracy called E.C.C.O.

After a study a few months ago showing that toxoplasma didn’t produce behavioral changes in humans, a new study suggests toxoplasma is no more common in cat owners than anyone else. All of the cool toxoplasma theories are going out the window. [Edit: This is confusing]

Yasmin Nair: Suey Park and the Afterlife of Twitter. Firebrand Twitter activist Suey Park has reinvented herself as a speaker warning about the dangers of firebrand Twitter activism, now says that social justice is a “cult” and that “the violence I have experienced in SJW circles has been greater than that of ‘racist trolls'”. Nair questions the convenience of a pipeline between fame as a Twitter activist and fame as a person speaking out against Twitter activism. But part of me worries that the entire chain – Park engaging in activism, Park speaking out against activism, Nair writing about Park, and now me linking Nair – is part of the problem, in that it promotes paying attention to Twitter activism at all.

Higher amounts of dairy fat markers in the blood associated with less diabetes. Very reminiscent of past studies showing that whole milk drinkers are healthier than nonfat/lowfat drinkers. Unclear if this says something profound or just that milk is a healthier source of calories than a lot of the alternatives. Related: TIME: The Case Against Low-Fat Milk Is Stronger Than Ever.

Reason #6894019 not to mess with Finland.

Previously: a short conversation with someone can decrease prejudice. Later: no, sorry, that study turned out to be fraudulent. Now: okay, the study was fraudulent, but the conclusion was actually true.

Successful charter schools seem to do much better than public schools in educating the most disadvantaged minority children, but critics have scoffed that they must either be selectively admitting the best students or just “teaching to the test”. But one new study finds charter school success cannot be explained by selective admission, and a second finds commensurate success on non-test-related outcomes, including lower teenage pregnancy and lower incarceration rates for charter school students. Educational establishment vows to respond to findings by improving their own performance calling charter schooling racist a lot.

When Queen Elizabeth I wanted to claim the New World, she asked court mathematician/astronomer/historian/angel-summoner John Dee for scholarship relevant to the expansion. Dee hit the books and conveniently discovered that King Arthur had led a vast army to conquer America, Greenland, and the North Pole

Some changes in Italian penal law help us more accurately determine the time discount functions of criminals.

Police, emergency responders, and other professionals try to trace Internet users’ IP addresses to find out where they live. Untraceable IP addresses – a big chunk of the total – all show up as coming from the geographic middle of the United States, which happens to be on a random farm in Kansas. Here’s a look into the life of a random Kansas farmer who has no idea why everyone is after him.

Everything we knew about classical Chinese civilization came from attempts to reconstruct it after the Qin Emperor burnt all the books around 200 BC. Now for the first time archaeologists have found original texts from Confucius’ time, and they’re a lot less orderly than expected.

Study finds violent video games do not increase misogyny, as usual everyone ignores this, seizes on a single doubtful sub-subfinding out of context and reports that the study proves violent video games do increase misogyny. The only thing at all surprising about the whole process is that Science Of Us notices and complains about it.

New study finds Haidtian moral foundations aren’t stable, heritable, or predictive; Haidt says that’s because the study did a terrible job measuring them.

A team including Russian billionaire Yuri Milner, American billionaire Mark Zuckerberg, and scientist Stephen Hawking have announced a plan to build a fleet of interstellar (!!!!) probes. The idea is very clever and has at least partially overcome the obvious skepticism such a plan should warrant. But I can’t help thinking it’s another dead end like the moon landing – something that will make it into the history books but have no broader impact on human activity or the colonization of space. Billionaires are obviously allowed to spend their money on whatever cool stuff they want, and goodness knows this is better than another yacht, but my sympathies are still with the less glamorous projects of Musk, Bezos, and Branson.

The latest step in the sportification of the political process is the release of Decision 2016 Trading Cards.

Women With More Feminine Digit Ratio Have Higher Reproductive Success (p = 0.002) – mediated at least in part by longer reproductive lifespan. The effect is sufficiently strong that we should be really curious why evolution preserved the contrary set of genes – maybe they’re better for men?

The Lexicon of Intentionally Ambiguous Recommendations collects apparently complimentary phrases to use in recommendation letters for people you secretly loathe. For example, “you would be lucky to get him to work for you”, “I recommend this man without qualifications”, “You won’t find many people like her”, “I cannot recommend this person too highly”, and “Nobody would be better than her”. The obvious review for this book is Moses Hadas’ “I have read this book and much like it”.

What happens when people from primitive tribes without mirrors who live in areas without clear water see their reflections for the first time? They freak out. [warning: includes scary picture]

A beautiful theory killed by an inelegant fact: sex offenders have no more testosterone than anyone else. Not in this study, but IIRC violent offenders do have more testosterone than others.

Vox: Everyone says the Libya intervention was a failure. They were wrong.

GiveDirectly will be starting large-scale tests of a universal basic income in Kenya.

RIP The 10,000 Year Explosion co-author and anthropology/evolution blogger Henry Harpending.

If you enjoy this fan website, you can support us over here. Thanks a lot!
Loading...
Enjoying this website? You can donate to support it! You can also check out my Book Translator tool.