Edge really mixes it up this year. Instead of “who’s science do you like” it is “who’s science do you not like” or “please knock around your favorite strawman.”
While this question encourages combativeness, what I like about what Edge does, is that it gets everyone’s often-hidden biases and beliefs out on the table (and in writing). These are types of beliefs that come out in conversation or at a conference presentation, but are often obscured in professional publications by careful constructed collegiality and jargon.
Many of the 174 responses have to do with the normal topics of this blog: evolution and social evolution, especially of humans. Here are a subset of that subset . Most have to do with sussing out the relative contributions to behavior of genes, environment, culture, maternal effects, epigenetics… I agree to these to various extents. There are at least two that I just don’t understand, even after a few re-readings. But I include them anyway to contrast with the others.
Are you, or do you know, an undergraduate who has good math, biology and/or computer programming chops wanting to spend some of their summer getting paid to do research in these areas?
My research institute, NIMBioS, is now accepting applications from undergraduates and for its Summer Research Experience program. NIMBioS provides the housing and a stipend and two mentors who will guide students through a research project. The 2014 projects are pretty diverse and include one that I will co-mentor with fellow post-doc Keenan Mack on the Argentine ant supercolony.
As described in this Radiolab story, the massive scale of Argentine ant cooperation is astounding. But, as described in this (gated) commentary by David Queller, probably unstable. Our project, in short, is to model this instability.
If you know an undergraduate with a pretty good grasp of entomology, evolutionary game theory, and/or agent-based modelling skills, please encourage them to apply. If you happen to be an undergraduate who regularly reads this blog, chances are you are the type of person who should apply as well.
Having served on admissions committees for my former graduate program, having these types of research experiences (and strong associated recommendation letters) are a good way to convince admission committees and potential major professors that you know what actual science (as opposed to classroom science) looks like.
For reasons unrelated to my usual research, I have been looking into medical literature concerning diabetes in “normal weight” people. If you read anything about treating diabetes the first steps are to eat better, exercise and lose weight. But what if you already eat healthy, exercise a lot and are at a normal or ideal weight? Well, current medical thinking is that if you are at normal weight with diabetes you are more likely to die than someone who is overweight. This seems counter-intuitive which is why it is called an “obesity paradox.”
Much of this comes from a paper published last year in JAMA (a top medical journal) finding that “[a]dults who were normal weight at the time of incident diabetes had higher mortality than adults who are overweight or obese.” This counter-intuitive finding was reported in the New York Times, the NYT health blog, CNN, Rueters, CBS, and many other places.
The study included 2,625 individuals pooled from five observational datasets. The individuals “were classified as normal weight if their BMI was 18.5 to 24.99 or overweight/obese if BMI was 25 or greater.” BMI, or Body Mass Index, is calculated as one’s weight in kilograms divided by one’s height in meters squared.
BMI is the measurement the media uses when reporting, for example, America’s increasing obesity epidemic. BMI is also often used by doctors as an individual measure of health (even though that was never its intended purpose and it has many well-known limitations). One thing to notice is that BMI does not take body composition into account, just weight – one kilogram of muscle counts the same as a kilogram of fat. A common criticism of using BMI for assessing individual health is that a very muscular person is considered as “obese” as a very fat person.
So do diabetic people with normal weight have higher mortality than obese people with diabetes? The results, as quoted, from the JAMA paper:
After adjustment for demographic characteristics and blood pressure, lipid levels, waist circumference, and smoking status, hazard ratios comparing normal-weight participants with overweight/obese participants for total, cardiovascular, and noncardiovascular mortality were 2.08 (95% CI, 1.52-2.85), 1.52 (95% CI, 0.89-2.58), and 2.32 (95% CI, 1.55-3.48), respectively.
Wait a second! They controlled for WAIST CIRCUMFERENCE?!? (I think this may have been the only time I’ve literally done a double-take while reading a scientific paper.)
To see why this might be a big problem, please consider two fictional characters, Zangief* and E. Honda, from the hit 1987 arcade game Street Fighter II. These gentlemen have about the same BMI (using the stats from the Street Fighter wiki). However, their waist circumferences are radically different.
Who would you consider more obese? Using BMI alone they are equally obese, but I am also going to go out on a limb and say that E. Honda is closer to what most of us would consider “obese” given that a higher percentage of his mass comes from body fat.** But what happens when we control for waist circumference?
Essentially, we are asking “what is the estimated effect of BMI, discounting the estimated effect of abdominal circumference.” I conservatively estimate that Zangief’s waist is about half the circumference of E. Honda’s. So “controlling” for waist circumference discounts E. Honda’s BMI more and thus we effectively count Zangief as more obesethan E. Honda. This is because we have abstracted away the additional abdominal fat that contributes both to E. Honda’s circumference and BMI. The result is that if even if someone built like E. Honda has a higher rate of mortality than someone built like Zangief, we would conclude that less obese people have a higher mortality rate. This “obesity paradox,” though, would just be a statistical artifact of controlling for something we should not have.
But these are video game characters. Would this hold up for real-world data? Let’s see.
R (the statistical software) has an easily downloadable dataset for bodyfat and other body measurements for 253 males. (The documentation suggests that five observations have errors, so I removed them, leaving us with data for 247 males. This is much smaller than the sample in the paper, but good enough to illustrate my point.) For those playing at home, click below for the code to snag the dataset:
#Installs the mfp package - only do this once
#Load bodyfat dataset
#Copy bodyfat dataset to bf_data
bf_data <- bodyfat
#Add a column of BMIs calculated from weight and height data
#Remove observations documentation suggests are errors
bf_data <- bf_data[-c(42, 48, 76, 96, 182),]
This data contains body fat percentage (called “brozek”), height and weight (from which we can calculate BMI), and waist circumference. What we are going to do next is a thought experiment where we assume that the common wisdom is exactly true. That is we are going to assume that body fat is an exact predictor for mortality (however measured). In real life this is obviously not true, there will be noise in the data. But this assumption will hurt us in that it should be the hardest case for falsely finding that overweight people have lower mortality than normal weight people. In other words, even assuming that the intuitive result is exactly true, can we erroneously find the counter-intuitive result by controlling for abdominal circumference?
First, lets run a regression of BMI on body fat percentage. If the coefficient of BMI comes out positive, this indicates that there is a positive relationship between BMI and body fat. And, since we assumed that body fat was a perfect predictor of mortality, BMI would have a positive relationship with mortality.
# Fit the regression model (brozek is a measure of body fat percentage)
BMI_fit <- lm(bf_data$brozek ~ bf_data$BMI, data=bf_data)
#Report the coefficients
coefficients(BMI_fit) # model coefficients
#Plot the data and regression line
plot(bf_data$BMI, bf_data$brozek, xlab="BMI", ylab="Body Fat Percentage",main="BMI vs Body Fat")
Below are the results. The coefficient for BMI (1.51) is positive, indicating a positive relationship between BMI and mortality. This is consistent with the conventional wisdom.
The same can be done with waist circumference.
# Fit the regression model (brozek is a measure of body fat percentage)
WC_fit <- lm(bf_data$brozek ~ bf_data$abdomen, data=bf_data)
#Report the coefficients
coefficients(WC_fit) # model coefficients
#Plot the data and regression line
plot(bf_data$abdomen, bf_data$brozek, xlab="Waist", ylab="Body Fat Percentage",main="Waist vs Body Fat")
coefficients(WC_fit) # model coefficients
Again, the coefficient for waist circumference (0.58) is positive, indicating a positive relationship between waist circumference and mortality. This is, again, consistent with the conventional wisdom.
Finally, lets see what happens when we run the regression for BMI “controlling” for waist circumference.
# Fit the regression model (brozek is a measure of body fat percentage)
BMI_and_WC_fit <- lm(bf_data$brozek ~ bf_data$BMI + bf_data$abdomen, data=bf_data)
#Report the coefficients
coefficients(BMI_and_WC_fit) # model coefficients
The coefficient on BMI came out negative! To a naive observer this would look against the conventional wisdom. BMI correlates negatively with mortality (and remember that this is after we assumed that body fat was a perfect predictor of mortality). This would seem to indicate that obese people are less likely to die than normal people. An “obesity paradox.”
Why does this happen? It happens because when controlling for waist circumference, you are essentially making fat cost less than muscle in your BMI calculations – the same as with E. Honda and Zangief above. In other words, you are counting people who are in better shape as more obese than they really are.
How does this simple thought experiment jibe with the original paper? Pretty well I think.
One thing about the paper is that they actually ran one model where they did not “control” for waist circumference. They did not report finding an “obesity paradox” for that model – which is consistent with my thought experiment. However, they suspiciously did not report any results for that model in the text of the paper -which leads me to suspect that the results did not fit well with their story. [But they did in a Table, see important UPDATE below.] They also found that abdominal circumference was associated with mortality which is consistent with my thought experiment.
So there you go, it seems like people with more muscle relative to fat live longer. Not exactly as counter-intuitive as the study (and press about the study) might make us think.***
I get really mad about bad statistics in medical research. It is one thing to erroneously claim that beautiful people have more daughters. It is not going to literally kill anyone. But medical doctors (most without much statistical training) rely on published medical research to treat patients. When they rely on bad research, it can kill people.
Medical stats people. Am I missing something important here? If not, is this problem worth pointing out in a more formal way? Obviously my thought experiment is simpler than the model in the paper (I consider this a feature) and my dataset is smaller (easily corrected, I think). Send me an email or comment below.
[UPDATE: Thank you for all of the comments thus far, both here and through social media. One of my friends pointed out that Table 2 of the original paper contains the results for the model that did not control for waist circumference that were not reported in the text. He said that this was standard practice in reporting medical research, so it was unfair of me to say this was suspicious.
From the way I read the table normal weight is still associated with higher overall mortality (but in both models not cardiovascular mortality), but difference between normal weight and overweight/obese is much less than in the model adjusting for waist circumference. The confidence interval for normal BMI total mortality does not quite overlap with the mortality for overweight/obese BMI individuals. So outside of something else going on, indications are that controlling for WC may increase the magnitude of the “paradox” but is not the sole explanation.
The table reports the findings in relative risk – which is fairly uninformative. Since the baserate for death in the overweight group was about 1.5%, we are talking about an increased absolute risk of on the order of 0.5-1.5%. Disagreement?]
*- In the Disney movie, Wreck-It-Ralph, Zangief is portrayed as a “bad guy.” However, in the video game he was not a bad guy. He was just Russian. Just because you are Russian doesn’t necessarily mean you are a bad guy.
** – Please note that I don’t want to pick on E. Honda. Despite his size, he is pretty spry. I almost always picked him when challenging my middle school friends (and enemies) at the arcade.
*** – Relatedly, a lot of people tell me that belly fat is an especially bad kind of fat for diabetes. But this oft cited study on the subject does not support that conclusion. It doesn’t compare belly fat to other kinds of fat. In fact, it finds them very highly correlated. What is going on here?
Randy Schekma, upon winning the Nobel Prize in medicine says he will not publish in Science, Nature or Cell (top science journals). Interesting commentary on this move by Michael Eisen and Jon Wilkins.
Why Biology Belongs in the Study of Politics by John Hibbing guest-posting at the Monkey Cage. I find this whole discussion fairly frustrating. Mostly because biology ≠ genetics. Maybe I’ll write a post about this someday.
What’s the deal with inclusive fitness theory? Ben Allen blogs about his new paper with Martin Nowak and E.O. Wilson. I find little to disagree with here (other than calling my very first blog post “heated.”) Models make assumptions. Models are used for different purposes. These assumptions are often wrong, but sometimes useful. Sometimes assumptions useful for one purpose are not useful for another. I think there is still confusion here between inclusive fitness accounting and inclusive fitness theory, so my opinion of the older Nowak/Wilson paper is unchanged. I also like Jon Wilkins’s bear video.
The lack of information flow between disciplines can hardly be underestimated. A brilliant example is the sunk cost fallacy... Hundreds of papers were written in economics and psychology on the sunk cost fallacy, and hundreds of papers were written in evolutionary biology (by some of the most eminent biologists) on the Concorde fallacy — which is the same fallacy. There is not a single cross reference in these hundreds of papers, nor any awareness that both fields came to opposite conclusions: in economics and psychology, it is taken for granted that humans often commit the sunk cost fallacy, in animal biology,
no conclusive evidence has been found that a single animal species would commit the sunk cost fallacy (Arkes and Ayton 1999).
When I have talked to colleagues in math and physics, I was jealous of the arXiv preprint system model. It always seemed a lot closer to the way someone would design institutions for scientific publication for 21st-century (instead of 17th-century) technology. Now biologists have their own place to play.
On Andrew Gelman’s blog he often reprints emails he receives and his responses when he thinks they may be interesting to his readers. This post will be my first attempt at that. However, I once surprised a friend when congratulating him on his email being answered on Gelman’s blog. So, unlike Gelman, I plan to always ask permission first.
Ed LeGrand writes:
At the 2010 Human Evolution & Behavior Society meeting I met Paul Bingham, who discussed his ideas in his book, Death from a Distance. Intrigued, I bought and read it and have since wondered why I'm not hearing discussion of it in, since it sounds so reasonable.
Bingham's premise is that the evolution of weaponry goes hand-in-hand with the evolution of social size and the size of political systems (of course, there are hundreds of correlates!). As you mentioned, chimps need about a 4:1 advantage before attacking with a reasonable degree of safety. Bingham would point out that because of spatial limitations, not many more than 3-4 individuals can be involved in hand-to-hand combat against an opponent at a time. Because the victim may be large or skilled, there's a rather high risk that one of the attackers will get injured. Bingham views this attack from the standpoint of group members disciplining/punishing a transgressor. Once a method of killing from a distance (throwing rocks) developed, the odds changed dramatically. It now becomes much safer to punish transgressors. Rather than 3-5 group members trying to punish one person and suffering the real risk of physical injury (a 4:1 advantage for the attackers), they can administer punishment at a distance. Not only is it physical safer, but now there are 4 people throwing rocks at one person, who can throw one rock at 4 people (a 4^2:1 advantage). The farther away one can be from the transgressor, the more physical space there can be for discipliners. With the development of new weaponry, a group of 100 people can easily simultaneously send 100 arrows at a transgressor (or victim) while only 1 arrow gets sent back at the 100 people (a 100^2:1 safety factor for the attackers). Bingham's book follows the development of weapons (spears, arrows, firearms, etc.) and notes that the advantage is greatly magnified. Bingham suggests that the size of the group/hierarchy is causally associated with the ability to control transgressors (or the ability of a dictator to control his subjects). In warfare, this ability to deliver death from a distance greatly increases the advantage of having the new technology. As a defense, the opponents without the most advanced weaponry have no choice but to join a larger group, and so on (I'm condensing a couple of hundred pages of argument and examples). The epitome of this is the advent of nuclear weapons where punishment of a whole country can be administered by the push of a button at great distance. Bingham pointed out that this has pushed the world into a huge single society. Drones aren't mentioned, but that's like the technology of a whole society being directed at specific transgressors from the safety of being on the other side of the globe.
Anyway, you may have heard of the conceptual model. If not, it's one to at least be aware of. I'd think the advantage of weaponry for easier and safer ability to effect punishment or to extend one's will would lend itself to interesting mathematical modeling.
I had not heard of this specific thesis, but it reminds me of some of the ideas in Boehm’s Hierarchy in the Forest. Boehm’s story is actually somewhat opposite. He posits that human groups was once fairly similar to chimpanzee groups – physically dominant males were at the top of a social hierarchy and maintained their social status by, basically, brute strength against all challengers. Weaponry allowed less dominant individuals to effectively gang up on any would-be dominant male and invert the hierarchy – creating fairly egalitarian groups. In his story, death from a distance creates the conditions for less social control.
It seems like what should be important, in addition to total killing power, is how killing power is distributed. With Boehm’s chimpanzees and with the nuclear weapons in your example, power is concentrated in the elites (or there are elites because power is concentrated). In Boehm’s egalitarian groups and in the bow-and-arrow example from your email, power is fairly widely distributed. I hadn’t really thought about it in terms of weaponry, but I think this is probably important for understanding the size of hierarchies. I’ll check out Bingham’s book.
The Festival of Bad Ad Hoc Hypotheses (BAH!) is over, but the videos will be posted soon. It was organized at MIT by cartoonist Zach Weinersmith and (my fellow UC Davis Graduate Group in Ecology 2007 cohort member) parasite ecologist Kelly Weinersmith. (They are interviewed here.) The festival included distinguished judges, sponsors and speakers who presented “well-argued and thoroughly researched but completely incorrect evolutionary theory.” The prize for best talk? “A sculpture of Darwin shrugging skeptically.”
Abstract: Research in many societies shows that ethnic diversity correlates with a decline in cooperation at the community level. This literature neglects cases in which ethnic heterogeneity is hierarchically structured. Power and status differences between ethnic groups, or ethnic dominance, may play an important role in determining cooperative outcomes. We test this hypothesis using public goods experiments with caste groups in India in which we manipulate the caste composition of experimental groups. Conservative estimates show that ethnic dominance between high and low ranking castes has a much larger negative effect on contributions in the public goods experiment than does caste diversity. We argue that ethnic dominance interactions such as ethnic discrimination constitute a type of antisocial punishment between groups. We also find that conditional cooperation is limited to within ethnic groups, revealing ethnocentric cooperation preferences. These results confirm the importance of group structure in human cooperative patterns, and help bridge the gap between evolutionary theory and cooperation dynamics in multi-ethnic real world settings.
Abstract: Humans seemingly have a greater capacity for altruistic cooperation with non-relatives than any other species. For example, in experiments humans cooperate more than one would expect either from cross-species comparisons or by calculating payoff-maximizing behavior. Two main hypotheses have sought to explain this – the mismatch hypothesis, where humans evolved genetically-inherited rules for cooperation in small kin groups and misapply them in modern contexts, and the norm psychology hypothesis, where humans evolved the capacity for to learn cooperative norms. I show that a model developed to support the former hypothesis actually better supports the latter. Similarly, in warfare humans take large risks to benefit group members who are mostly non-relativ
es. Quite a few hypotheses have been proposed to explain this, however these hypotheses are fundamentally incomplete when they do not account for both cultural inheritance and group-structure. Finally, explaining cooperation in larger-scale complex human societies has been difficult since many of the institutions that work in small-scale societies become less effect ive as group size increases. Hierarchical organization seems a potential solution to increasing group size, though the basic theory has not been established. I present a preliminary model of hierarchy’s origins that I will develop as a NIMBioS post-doc.
I just took this dialect quiz that purports to tell “where in the continental United States do they speak like you?”
The quiz is based on the Harvard Dialect Survey and I was particularly interested in it because, as a military brat and military veteran, I grew up and lived all over the country and figured I might break the thing. (My father is from the Midwest and my mother is from western Pennsylvainia, so I expected I’d test for a standard Midwest dialect.)
For reference, here are all the places I’ve lived for more than a year and my approximate ages:
Colorado Springs, Colorado (0 – 3)
North Pole, Alaska (4-6)
Woodbridge, Virginia (7 – 10)
Flower Mound, Texas (11 – 13)
Amherst, New York (14 – 18)
Ithaca, New York (19 – 22)
Abilene, Texas (23 – 25)
North Pole, Alaska (26 – 28) – yes, again.
Davis, California (29 – 34)
Knoxville, Tennessee (34 – )
So did I break the thing? 140 questions later, here are my results:
The first thing to notice is that the map is not super informative (there appears to be a band of similarity going from Virginia, through the Midwest, to Texas).
The tables are more interesting. The numbers specify the probability that a randomly selected person from those cities would answer a randomly selected question the same way I would.
First, there does not seem to be a lot of variation in my scores and they all seem quite low to me (though I do not know if this is common). I also appear to not share a common dialect with New Englanders (which is probably very common).
But what is really interesting to me is my “most similar city,” Denton, Texas. It is interesting because Denton is only 15 miles away from Flower Mound, Texas, where I spent my middle school years (and is also, incidentally, the home town of one of my favorite persons from grad school). The map below shows my middle school suspiciously close to Denton. (My middle school was, again incidentally, named after the great-great grandfather of another one of my favorite persons from grad school.)
Many of the people who read this blog have also heard me speak and have even likely had conversations with me. Do I sound like I’m from East Texas to you?
Granted, when I moved to New York my nick-name on the 9th-grade football team was “Tex” based largely on my accent. But by the next year, I lost (most of?) the drawl and that nick-name was forgotten (read: replaced by “Zim”).
But does some of that old middle-school dialect remain? Or is the above result just some sort of coincidence?
There seems to be some evidence for the former. According to this book, I lived near Denton during a key stage of peer-based dialect formation:
Labov (1970)* proposed that, up to the age of 5, children acquire the basic grammar and lexicon of their language, normally under the influence of parents. Between the ages of 5 and 12, children learn the dialect of their peer group… By the age of 14 or 15, adolescents start to move away from the peer group dialect and toward the more prestigious form of speech, especially in formal situations.
Might this peer group dialect still persist 21 years later to the extent it is detectable by the above dialect survey?
* – Labov, W. (1970) Stages in the acquisition of standard English. In R. Shuy (ed.), Social Dialects and Language Learning. Champaign, Ill.: National Council of Teachers of English.
UPDATE: I just want to point out that this “probability that you would answer a randomly selected question the same as someone from a randomly selected city” measure is likely not the best way to pinpoint someone’s area of origin. Some questions are more likely to provide more information to others, and some answers to those questions are likely to provide more information to than others. For example, one of the questions asked about the word used for a light-rail train. One of the possible answers was “BART.” BART stands for Bay Area Rapid Transit and I’m pretty sure someone using that as the generic word for light rail would likely have close ties to Northern California. But even if it is a crude measure, it at least seemed to preform well in my (anecdotal) case.