Tag Archives: science

Work, Drink, Sleep, Repeat

Slate’s Jordan Weissman reports on a new paper regarding the drinking habits of professional who work long hours:

Science has finally confirmed what anybody who has ever met an i-banker, lawyer, or journalist already knew: People who work exhaustingly long hours like to drink themselves insensate at the end of the week.

To be specific, an analysis published in the British Medical Journal found that working more than 48 hours a week was associated with a slightly higher probability of “risky” alcohol consumption. The authors reached their main conclusions by analyzing unpublished data from 27 studies conducted in the United States, Europe, and Australia. They also looked at the findings of 36 previously released papers, many of which were also from Japan—where after-work binge drinking is basically a cherished part of its office culture.

I’ll confirm this from personal anecdotal experience, and I think we’re all intuitively on board with the finding. Weissmann, however, backs the wrong conclusion:

the data doesn’t give any clear answers on correlation vs. causation. Maybe those who tend to pull endless hours at the office are the same personality types who tend to imbibe heavily. Or maybe working long hours drives people to drink. Personally, I’d bet on the latter.

While no doubt some people drink to cope with the long hours, it is far more likely that the same factor causes both long work hours and heavier drinking. I’ve had this hypothesis for a while and this is as good a time as any to mention it: the brain’s reward system  is the same for all rewards. Your dopamine level doesn’t know if you’re raising it with work, sex, alcohol, or cocaine. My hypothesis is that the people whose reward center is more sensitive are more likely to exhibit addictive behavior in different domains; that is, they’re more likely to feel “rewarded” working long hours and similarly “rewarded” drinking that 8th whiskey. I’d bet that this common factor explains a lot of the correlation found in the paper.

Of Mice And Men

Science improved a little bit recently when researchers in Canada found that laboratory mice react differently to scientists in a blatantly sexist way:

Both when the mice were left on their own and when they had any of four female researchers in the room with them, the injections induced visible distress. But when any of four male researchers was there, the animals showed significantly fewer signs of pain. The same was true when the team did the experiment on rats, and also when they used a different pain stimulus, formalin.

Nor was the actual presence of a human required. When the team substituted the four men with T-shirts that those men had slept in the night before, the rodents responded in the same way. Women’s T-shirts, like women themselves, did not reduce apparent pain—indeed, when placed in the room next to a man’s shirt, a woman’s shirt was able to abolish its effect. Nor was the phenomenon caused only by the scent of human males. Bedding slept in by male dogs, cats, rats and even guinea pigs all produced a similar response. …

Further experiments, which measured levels of corticosterone, a stress hormone, in the animals’ blood, showed that they were indeed stressed by the mere smell of a man. And examination of gene activity in their pain-producing nerve cells confirmed that these cells temporarily shut up shop at the same time. Simply put, the animals were being scared painless. (A significant increase in faecal pellets suggested they were scared shitless as well.)

So what does this tell us? Well, for one, science is difficult. Two, gender matters.* Three, there are probably many other variables we haven’t even considered that affect human and animal research, so good luck to us.

*Also confirmed by a recent NIH request for more research on female animals, who are usually avoided by researchers because their (the animals’) hormonal cycle can mess with data.

Finally, if this means a lot more jobs for female scientists at the expense of males, that’s fine. Reality is allowed to be sexist. Just remember this when reality swings the other way.

Question Authority – Specifically, Their Methodology

Whenever you see any study reported in the popular press, as a first approximation, you should not believe it. The more salacious the headline, the less you should believe it. I say this because I posted yesterday on an experiment that received some play, and it presents a good opportunity to show the many reasons you need to discount any particularly interesting finding. This is not to pick on this paper in particular – no matter what you believed before you read it, you should be slightly more likely to believe its findings than you were before.

0. Publication bias. Before even thinking about the topic, remember that only studies that get a result get published, and even at high confidence levels, if you do enough studies, you’ll get fluke results. See, generally, XKCD. We have no way of knowing whether this is the only experiment in 10 that yielded any results.

1. Sample size. Always check how big the sample was. 20-30 on each side might be enough. Might not. Flukes happen in small groups.

2. Experimental method. Was it a randomized controlled trial, the gold standard in social science? In the above study, it’s actually pretty close – all things equal except race, and no reason to believe that the two races were assigned to partners anything but randomly. Perhaps you’d want it to be double-blind, so that senders don’t know what they’re sending to which partner, but that’s a nitpick. A bigger issue, perhaps, is that the partners involved were not particularly motivated to do well, since there was little to be gained from a good review. Actual work product that might be used in court would probably get a lot more partner attention than a review of strangers’ writing samples.

3. Incentives. Are study participants properly motivated? Would partners who are reviewing work product that would go to a court or a client miss as many errors as they seem to have done here?

4. Suggestion. It’s really easy to get people to focus on something when you bring it to their attention. In this case, the reviewing partners knew the writer’s name, class year, law school, and race. I doubt I’m the only one who finds race to be a clear outlier in that group  – I’ve never been handed a piece of work product with a note that says “a black secretary made these copies.” By making race salient, researches can make it play an exaggerated role. Perhaps this is the intent – mentioning it probably does elicit subconscious biases – but it also undermines the experiment’s validity in real life. If a partner in real life wouldn’t care about the associate’s race, but does so in an experiment because it’s brought to his attention, the experiment misidentifies the real-life relationship.

5. Alternative interpretations. This is the tough part: trying to find other possible interpretations of the data you get. I remember a simple example from a college class: China has low inequality and poor beaches, while Brazil has high inequality and nice beaches. Thus, beaches cause inequality. It could be that nice beachfront property causes inequality because only some people can have it; more likely, a history of colonialism and demographic heterogeneity are more important causal factors.

In the study above, there are multiple possible explanations:

  • conscious, intentional, coordinated racial bias: partners don’t like black people and want to make them look worse than whites.
  • subconscious, unintentional racial bias: partners just expect worse work product from black attorneys and spend more energy looking for and finding mistakes. (This seems to be the authors’ preferred reading.)
  • statistical discrimination: partners have historically received worse work product from black attorneys and have learned to scrutinize their work more carefully. This is basically a version of subconscious bias, though the root causes are very different.
  • affirmative action bias: partners, perhaps falsely, expect a black student at NYU Law to have been admitted for reasons other than merit, and thus spend more time scrutinizing his work for errors.
  • other things I haven’t thought of.

If I had to guess, I’d say 2 probably dominates the explanation, with 3 and 4 perhaps playing a role and perhaps not.

So what does this all mean?

Not much. Raising questions about a study doesn’t invalidate it. You can’t dispose of a conclusion you dislike or dispose of objections to conclusions you like with this process. The reason this sort of methodological analysis is important is that facts are important. Raising doubts until we know the true facts with reasonable certainty is essential to making sure we don’t base decisions on falsities. Those tend to work out poorly.

An Obvious Insight About Basketball Players That Surprises Way Too Many People

Both my Facebook and Twitter feeds lit up with references to a Seth Stephens-Davidowitz article in the New York Times about the origins of pro basketball players. SSD’s study tells us that, contrary to the thug stereotype, middle- and upper-class players are much more likely to make it to the NBA:

AS the N.B.A. season gets under way, there is no doubt that the league’s best player is 6-foot-8 LeBron James, of the Miami Heat. Mr. James was born poor to a 16-year-old single mother in Akron, Ohio. The conventional wisdom is that his background is typical for an N.B.A. player. A majority of Americans, Google consumer survey data show, think that the N.B.A. is composed mostly of men like Mr. James. But it isn’t.

I recently calculated the probability of reaching the N.B.A., by race, in every county in the United States. I got data on births from the Centers for Disease Control and Prevention; data on basketball players from basketball-reference.com; and per capita income from the census. The results? Growing up in a wealthier neighborhood is a major, positive predictor of reaching the N.B.A. for both black and white men. Is this driven by sons of N.B.A. players like the Warriors’ brilliant Stephen Curry? Nope. Take them out and the result is similar.

Most of the people posting this stuff noted either how the “conventional wisdom” has been disproved or how things are even more stacked against the lower classes.  I’m probably an outlier here, but I never believed the conventional wisdom – I didn’t even know it was conventional wisdom. I have less knowledge about basketball than I do football, but in football lots of talent never makes it out of high school for academic or legal reasons, including questions about character, work ethic, and “makeup.” I’ve always expected it to be similar in basketball, though natural talent plays a larger role there. As far as opportunity for the lower classes, go, SSD opines:

These results push back against the stereotype of a basketball player driven by an intense desire to escape poverty. In “The Last Shot,” Darcy Frey quotes a college coach questioning whether a suburban player was “hungry enough” to compete against black kids from the ghetto. But the data suggest that on average any motivational edge in hungriness is far outweighed by the advantages of kids from higher socioeconomic classes.

What are these advantages? The first is in developing what economists call noncognitive skills like persistence, self-regulation and trust.

One unstated fact underlying SSD’s study is, once again, genetics. SSD doesn’t say it, but middle/upper-class children inherit, partially genetically, the abilities of their parents, which includes the above-listed persistence, self-regulation, and trust. Those abilities are less widely available, both via nature and nurture, in lower classes. Since these abilities are important to a successful career – in fact, as natural talent across the compresses in diversity,* they become even more important – it’s understandable that most players come from families where these skills are both inherited and taught.

*SSD addressed one aspect of talent compression: height. As nutrition in the world improves, there are more taller people, so the average height of the professional player goes up, and the range of heights compresses. I imagine similar developments are occurring in nutrition, training, and fundamentals, as younger and younger players are scouted and developed.

SSD even gives us a glimpse as to how middle- and upper-class black players tend to differ from their lower class brethren:

The economists Roland G. Fryer and Steven D. Levitt famously studied four decades of birth certificates in California. They found that African-American kids from different classes are named differently. Black kids born to lower-income parents are given unique names more often. Based on searches on ancestry.com, I counted black N.B.A. players born in California in the 1970s and 1980s who had unique first names. There were a few, like Torraye Braggs and Etdrick Bohannon. But black N.B.A. players were about half as likely to have a unique name as the average black male.

Considering how competitive professional sports is, it would be surprising if teams discriminated against the lower classes. During the Moneyball revolution in baseball, we learned how teams found undervalued players by using statistical methods. Similarly, any NBA team could find undervalued players from poorer communities and win more efficiently. The fact that this happens only sporadically probably means that there aren’t that many players who are capable of making the jump. Obviously NBA teams aren’t perfect, and there are only 30 of them, so clearly there is talent out there that isn’t being properly developed or discovered, but considering the multi-billion basketball industry and the rise of cell phone video and YouTube clips, I expect that scouting has been pretty good for a while and will only improve.  I’d be surprised if the results turn out vastly different from what they have been (that is, relatively few players from poor origins), especially, as I noted above, talent levels out and peripheral skills like work ethic and coachability rise in important.

 

Tabarrok And Hard Social Science Fiction

I’ve been pondering at length (see parts one, two, three, four, five, six, seven, and eight) the nature of good and bad narrative writing (in the broader sense that includes TV and movies) and what an exacting audience member such as myself likes and dislikes in such writing. Alex Tabarrok at Marginal Revolution weighs in and offers a useful distinction:

Hard science-fiction is science fiction that respects the findings and constraints of contemporary science. By analogy, I deem hard social science fiction* to be science fiction that respects the findings and constraints of contemporary social science especially economics but also politics, sociology and other fields. Absent specific technology device such as a worm-hole, hard science fiction rejects faster than light travel as little more than fantasy. I consider Eden-like future communist societies similarly fantastical. Nothing wrong with fantasy as entertainment, of course, just so long as you don’t try to implement it here on earth.

Tabarrok’s take is more policy-oriented, but since I share many of his econo-political views, I don’t mind. More importantly, he’s offered me excellent terminology in the distinction between hard and soft science and social science fiction. Many of my complaints in the posts link above seem to be complaints about pieces of fiction that are soft social sci-fi but claim to be hard social sci-fi. I’ve previously referenced movies that pretend to extend current trends into an inevitably dystopian future as a warning against continuing such trends – recently, the topic is inevitably wealth inequality. I’ve criticized these movies for their lack of verisimilitude, and I think it’s their pretense of realism that bothered me so. Tabarrok’s terminology captures my issues with such writing.

NDT On The Critique of Science In Film

I recently quoted blog-favorite Neil deGrasse Tyson’s Twitter feed and called him a fellow member of the  “difficult audience,” the group of people (of which I am one) that nitpick movies/TV shows/books for scientific and social-scientific accuracy. His tweets got quite a bit of media attention, and the man himself responded. In the interest of full disclosure, here’s an excerpt, but the whole thing is short and worth reading:

What few people recognize is that science experts don’t line up to critique Cloudy with a Chance of Meatballs or Man of Steel or Transformers or The Avengers.  These films offer no premise of portraying a physical reality.  Imagine the absurdity of me critiquing the Lion King:  “Lions can’t talk.  And if they could, they wouldn’t be speaking English.  And Simba would have simply eaten Pumba early in the film.”

The converse is also true.  If a film happens to portray an awesome bit of science when there’s otherwise no premise of scientific accuracy, then I’m first in line to notice.

 

 

 

The Truth About Mobility The Media Doesn’t Want You To Know

A recent study about income mobility in the United States received attention in the media (here is the NY Times on it) as it showed large differences in upward mobility in different US locations. Here’s the summary:

We find substantial variation in the economic outcomes of children from low income families across areas of the United States. Some areas have rates of upward mobility comparable to the most mobile countries in the world while others have lower rates of mobility than any developed country for which data are currently available. These geographical differences are modestly correlated with variation in tax expenditure policies across areas. But much variation in children’s success across areas remains to be explained, potentially by factors such as income segregation, school quality, or social capital.

Much of the commentary either sought to vindicate tax expenditures after all (consistent with the writers’ political leanings), or, more commonly, identify the other factors mentioned by the authors. However, it’s notable that no journalist or academic has dared point to the real cause of income mobility. The study’s home page identifies the cities with the highest income mobility:

1. Salt Lake City, UT
2. San Jose, CA
3. San Francisco, CA*

Salt Lake City seems different from San Jose and San Francisco in many ways, but the two areas (San Jose/San Francisco being basically one large region) share one important set of views. In Utah, evolution has its third-lowest level of acceptance in the United States. Meanwhile, in San Francisco, vaccination rates are the lowest, by far, in the US, despite all scientific evidence pointing to their safety.

What follows from these two facts?

Income mobility depends on the rejection of sound science. But you won’t hear the so-called mainstream media admit that.

*(Alternately, since #4 on the list is Seattle, perhaps mobility depends on being in cities starting with an S.)