Category: Science
A modicum of sanity on choice
I have not had time to read the original study but this rings true to me:
But now Benjamin Scheibehenne and colleagues have waded into the topic with the claim that the "too-much-choice effect" has in fact failed to appear in many experiments, and with the real-life observation that shops that offer more consumer choice tend to be more successful.
In a series of experiments, Scheibehenne's team tested 598 participants who were asked to choose from among restaurants, charities and music downloads. Throughout, they varied factors that they hoped might explain why the too-much-choice effect sometimes occurs and sometimes doesn't.
Examples of these factors included the need to justify one's choice; the perceived variety of choice, as opposed to actual amount of choice; the mean attractiveness of a range of choices; cultural differences (they tested German and US students); and individual differences such as people's tendency to maximise – that is, their consistent desire to find the perfect option.
For most of the experiments, the too-much-choice effect wasn't actually observed and when it did, the only relevant factor which increased the effect was the need to justify one's choice.
"The fact that most of the variables that we tested were not sufficient to elicit choice overload suggests that the too-much-choice effect is less robust than previously thought," the researchers said.
Repeat this fragment after me: "…the real-life observation that shops that offer more consumer choice tend to be more successful."
Why do I choose that you should repeat that fragment and not some other? I'm not going to tell you.
The Case Against Breast Feeding
Hanna Rosin's article on breastfeeding in the latest Atlantic is excellent and would make a topical and accessible introduction to causality studies in an econometrics or statistics class. (And lest that sound damning it's also a great read.)
The general point will be familiar to the audience at Marginal Revolution. The studies that show breastfeeding leads to lower weight, fewer ear infections, less allergies, less stomach illnesses and so forth are almost all observational studies.
An ideal study would randomly divide a group of mothers, tell one half to breast-feed and the other not to, and then measure the outcomes. But researchers cannot ethically tell mothers what to feed their babies. Instead they have to settle for “observational” studies. These simply look for differences in two populations, one breast-fed and one not. The problem is, breast-fed infants are typically brought up in very different families from those raised on the bottle. In the U.S., breast-feeding is on the rise–69 percent of mothers initiate the practice at the hospital, and 17 percent nurse exclusively for at least six months. But the numbers are much higher among women who are white, older, and educated; a woman who attended college, for instance, is roughly twice as likely to nurse for six months.
Moreover, the better we control for other factors that might account for differences in child outcomes between mothers who breastfeed and those who do not, the less evidence there is for breastfeeding's benefits. Even looking at children within the same family (still far from the gold standard of randomization), shows many fewer benefits from breastfeeding than studies that look across families. Some modest evidence suggests a gain in IQ and better evidence suggests minor improvements in avoiding some diarrhea. Rosin does not discount these benefits (so the title of her piece is unnecessarily sensationalistic) but she very appropriately does point to opportunity cost.
The debate about breast-feeding takes place without any reference to its actual context in women’s lives. Breast-feeding exclusively is not like taking a prenatal vitamin. It is a serious time commitment that pretty much guarantees that you will not work in any meaningful way. Let’s say a baby feeds seven times a day and then a couple more times at night. That’s nine times for about a half hour each, which adds up to more than half of a working day, every day, for at least six months. This is why, when people say that breast-feeding is “free,” I want to hit them with a two-by-four. It’s only free if a woman’s time is worth nothing.
One final point, Rosin's article is also usefully read as a study in propaganda and social psychology.
The Genial Gene
That's the new book by Joan Roughgarden and the subtitle is Deconstructing Darwinian Selfishness. I'm not sure how true this book is, but if you're looking for a new popular book on evolutionary biology which is engaging, this is the first one in some time.
The book rejects the "Red Queen" hypothesis for why there is sex (e.g., outracing parasites by frequently rolling the genetic dice) and presents a "portfolio diversification" view:
The explanation for why asexual species keep popping up and quickly dying compared with sexual species would seem to be completely explained by thinking of asexual species as genetic versions of get-rich-schemes and of sexual populations as genetic versions of long-term mutual funds, without any need to invoke cost-of-meiosis considerations.
In other words, sex brings a genetic diversity which protects against rapidly changing environmental conditions and thus favors parental genes.
The author also argues against signaling theories of the peacock's tail and against sexual selection more generally (especially on that latter topic I was not convinced but the discussion of sexual dimorphism and why it doesn't always hold is nonetheless interesting). She presents "social selection" as an alternative and if you turn to pp.237-8 you will see an excellent page-and-a-half summary of what the book is about. Male promiscuity, for instance, is viewed as a genetic "tactic of last resort."
Recommended, but with caution. It is a must for anyone who reads about evolutionary biology and by the end of the book I was less skeptical than when I started it.
Oh Ye of little faith
Mark Davenport reports to me (the link I added):
I noticed your recent blog entry about Rejecta Mathematica. A couple of us on the editorial staff are semi-regular readers of your blog, so we were happy and excited to see your posting. I wanted to let you know that we are actually in the later stages of releasing our first issue – just finalizing a few things with the small crop of brave authors who have responded to our invitations. If you would like, I will let you know when our inaugural issue is available.
Only in England, part II
Spud was found wandering in a garden and taken to Tiggywinkles Wildlife
Hospital in Aylesbury, Bucks, last August. Experts are baffled by his
condition, which has caused his skin to dry and his spines to fall out.
The hospital is asking practitioners of alternative medicine to suggest
possible remedies.
Here is a picture of Spud, a hedgehog with no spines.
I thank Chug for the pointer.
Patents versus Markets
Long ago Jack Hirshleifer pointed out that markets can reward innovative activity even in the absence of patents (H.'s point was actually that markets could over-reward such activity but the point was clear). If an inventor discovers a new source of energy that requires the use of palladium, for example, he can buy palladium futures, announce his discovery and wait for the price of palladium to increase. Of course, this only works if the discovery is credible so betting (contra Tyler) is an important way to test the credibility of a theory (e.g. here and here and of course Hanson's key paper Could Gambling Save Science).
All this is by way of introduction to a new paper in Science, Promoting Intellectual Discovery: Patents Versus Markets (press release here). Bossaerts et al. compare a patent system with a market reward system in an interesting experimental setting. The innovation is the solution to a combinatorial problem called the knapsack problem. In the knapsack problem there are Z items each with a certain value. You must choose which items to put into the knapsack in order to maximize it's value but each item also has a weight and you cannot go over a fixed weight which is set such that you can't carry all the items. The solution to a knapsack problem is not obvious since it's not always best to include the most valuable items. The authors argue that solving the knapsack problem is like combining ideas to create a new innovation. The authors, of course, know the optimal solution to each knapsack problem.
Rewards for creating the innovation are offered in two ways, in the patent method the first person to produce the optimal solution gets the entire reward. In the market system each participant is initially given an equal number of shares in each item. The item-shares trade on a market. After the markets close a $1 dividend is paid to each item-share if the item is in the optimal solution, other shares expire worthless. Thus, the price of the item-shares can be thought of as the probability that the item is in the optimal solution. (i.e. is palladium in the optimal solution to the energy problem? If so, it will have a high price.) Dividends are set such that the total reward is about the same in the two treatments. Proposed solutions were also collected in the market setting although the solutions per se were not the basis of any reward.
Important findings are that the problem was solved just as often in the market setting as in the patent setting. Indeed, in the market setting more people solved the problem on average. There are two possible explanations. First, the winner-take-all nature of the patent system may have deterred some of the weaker participants from exerting effort. Second, and more interesting, is that the prices in the market system did in fact incorporate information about the optimal solution – thus market prices may have given people hints about the optimal solution, much like seeing a partial solution to a jigsaw puzzle.
Problems are that the market system can work only if there are rents to be had from market prices. A new computer chip design, for example, won't change the price of silicon (although even here side-bets may be possible, the inventor knows the manufacturer to whom he sells the invention for example). Also, the price of an input, like palladium, can be influenced by many things other than the innovation so the market system will typically often involve more risk. Still this is an interesting experimental approach to a deep problem.
Thanks to Monique van Hoek for the pointer.
Why aren’t non-parametric statistics more popular in economics?
Abel, a loyal MR reader from Valencia, asks:
I've recently been reading an introductory book on nonparametric models (Nonparametric and Semiparametric Models: an Introduction – Härdle et al) and the apparent flexibility of the approach makes me wonder why aren't those models more used in empirical economics.
¿Are their drawbacks too big? (The so-called "curse of dimension")
¿Perhaps it's just economist's community lack of knowledge or willingness to learn?
¿Are they perceived as a threat to conventional or more established estimation methods?
I would be really glad to hear your opinion and also the feedback in the comments.
He could have added "signalling" to the list, since many non-parametric methods are relatively easy and thus do not demonstrate the skill of the researcher. But the fundamental reason I think has to do with the nature of economics: non-parametric methods are most likely when you don't have a well-defined, formal structural model in mind. But many MR readers know more about this than I do, so please offer us your opinions…
A new theory of suicide
This morning I read this:
In essence, Joiner proposed that people who kill themselves must
meet two sets of conditions on top of feeling depressed and hopeless.
First, they must have a serious desire to die…Second, and most important, people who succeed in killing themselves must be capable of doing the deed.
Maybe it’s the fault of the press coverage (remember when Modigliani won the Nobel Prize?…”people save for their old age”) but then I thought of this.
The time is now 1234567890.
That is all.
Keith Stanovich and what IQ is good for
The always-interesting-and-still-underrated Michelle Dawson points me to this batch of work. Here is one of the papers, by Keith E. Stanovich and Richard F. West:
In 7 different studies, the authors observed that a large number of thinking biases are uncorrelated with cognitive ability. These thinking biases include some of the most classic and well-studied biases in the heuristics and biases literature, including the conjunction effect, framing effects, anchoring effects, outcome bias, base-rate neglect, “less is more” effects, affect biases, omission bias, myside bias, sunk-cost effect, and certainty effects that violate the axioms of expected utility theory. In a further experiment, the authors nonetheless showed that cognitive ability does correlate with the tendency to avoid some rational thinking biases, specifically the tendency to display denominator neglect, probability matching rather than maximizing, belief bias, and matching bias on the 4-card selection task. The authors present a framework for predicting when cognitive ability will and will not correlate with a rational thinking tendency.
Even more interesting, in my view, is that higher-IQ people are more likely to behave rationally when they are told that a rationality issue is on the table, but less so otherwise.
If you are interested in issues of IQ, or for that matter overcoming bias, you should read Stanovich's work. As noted above, higher-IQ people seem to be just as guilty of "myside bias."
Stanovich has a new book summarizing some of the results, namely What Intelligence Tests Miss: The Psychology of Rational Thought. It is more idiosyncratic than the articles (he overcommits to one very particular model of the mind; cognitive laziness, without regard for margin) but recommended nonetheless. For those who care about these issues, a must.
The Physics of BS
Here is Frank Tipler on macroeconomics:
accurate predictions means that they do not know what they are talking about. We
non-economists should realize this also, and realize that our leaders, who are
being advised by macroeconomists, haven’t got a clue where they are leading us.
Well ok I have some problems with macroeconomics too but considering many of Tipler's writings his criticisms of macroeconomics are rather amusing. e.g.
Singularity–God–is like. The laws of physics tell us that our universe
began in an initial singularity, and it will end in a final
singularity. The laws also tell us that ours is but one of an infinite
number of universes, all of which begin and end in a singularity. If we
look carefully at the collection of all the universes–this collection
is called the multiverse–we see that there is a third
singularity, at which the multiverse began. But physics shows us that
these three apparently distinct singularities are actually one
singularity. The Three are One.
There is one religion which
claims that God is a Trinity: Christianity. According to Christianity,
God consists of Three Persons: God the Father (the First Person), God
the Son (the Second Person), and God the Holy Ghost (the Third Person).
But there are not three Gods, only one God. Using physics to study the
structure of the Cosmological Singularity, we can see that indeed the
three “parts” of the Singularity can be distinguished by employing the
idea of personhood. In particular, physics can be used to show how it
is possible for a man–Jesus, according to Christianity–to actually be
the part of the Singularity that connects the Initial and Final
Singularities. So the Incarnation makes perfectly good sense from the
point of view of physics.
Salsa Dancing Into the Social Sciences
The author is Kristin Luker and the subtitle is Research in an Age of Info-Glut. I enjoyed this book very much and I thought it was one of the best books on the philosophy of the social sciences I have read, ever. In part it is good because it ignores philosophy of science (and Continental philosophy gobbledy-gook) and focuses on the anthropology of how research is actually done. Here is the author's summary of her message
Let's review the state of play. I've told you that "methods" in the social sciences are historically, socially, and politically located in both time and place. I've also told you that the methods most commonly taught (canonical social science, "normal science") grew out of a particular time and place, namely postwar America. I've tried to convince you that in this new postmodern, globalizing world, those old methods don't work as well as they used to, at least not for the kinds of problems that most of us are interested in these days. Finally, I have argued that a whole set of "practices," that is, taken-for-granted ways of doing things that aren't even at the level of consciousness most of the time, grew out of those old methods and now must be rethought by those of us whose contributions will consist of making connections across boundaries, rather than following the normal-science way of making incremental contributions to a deep but narrow part of our fold.
There's much more to the book than that quotation indicates. Recommended.
How financial economics should evolve, from this point onwards
I read this in Temple Grandin's new (and often quite interesting) Animals Make Us Human: Creating the Best Life for Animals:
She [Jane Pruetz] spent four years just habituating the chimpanzees to her presence before she could study them. Then she spent three summers observing their lives. She discovered that some of the chimpanzees make spears out of tree branches and use them to spear bush babies inside hollow trees. Bush babies are small furry animals. The chimpanzee breaks a branch off the tree, strips off the leaves, and sharpens one end to a point with its teeth. Then it stabs the spear violently inside the hollowed trunk to kill any bush baby that might be inside. This discovery is so revolutionary that it has caused a big controversy in the field of primate research, because it is the first documentation of an animal using a tool as a weapon for hunting.
But alas we are told:
Animal research is getting more and more what I call "abstractified." Instead of people studying the real animals in their natural habitats, researchers use fancy statistical software to construct statistical models, and then they study the models.
The 10,000 Year Explosion
The subtitle is How Civilization Accelerated Human Evolution and the authors are Gregory Cochran and Henry Harpending. I do think that such topics should receive open debate but, as with Greg Clark's book, I'm not convinced. There is plenty on dog breeding, lactose intolerance, Genghis Khan and his children, the difficulties of settling the Andean Highlands, and just-so stories about medieval Ashkenazi Jews. What's missing is a sense of what the hypothesis does not explain, what its limitations are, and also what exactly is being claimed beyond the particular cited examples. The stories of "lots of recent change overall" and "current groups differ" are jammed together but of course they are very different. Epigenetics don't receive much attention, even critically, and the lower levels of Ashkenazi social achievement before 1800 are dismissed quickly. It's fine and indeed correct to claim they were oppressed but that opens up many doors to explain many other observed correlations. The authors report that we have Neanderthal genes even though this seems to fly in the face of recent discoveries and more importantly the evidence that such interbreeding (if it occurred) mattered is extremely speculative. Perhaps the authors are right but the reader is not given the tools to see why their understanding is a superior one.
Razib liked this book (see the first Amazon review) and I suppose it is a good introduction to this point of view, but overall I didn't come away feeling I obtained a superior understanding of the issues.
Don’t touch when you are shopping, or the new endowment effect
Be careful how you reach out:
A new study suggests that just fingering an item on a store shelf can create an attachment that makes you willing to pay more for it.
Previous studies have shown that many people begin to feel ownership of an item – that it "is theirs" – before they even buy it. But this study, conducted by researchers at Ohio State University, is the first to show "mine, mine, mine" feelings can begin in as little as 30 seconds after first touching an object.
Here is the full story. I thank Deron Bauman for the pointer.