Thursday, January 24, 2013

economics of happiness

I can't decide whether I hate or love this article more. Her criticisms are related to valid criticisms of happiness research, but she infuriatingly portrays happiness researchers as guileless soothsayers earnestly extrapolating from their immature results. (She is herself an economist, for the record, so I'm at least tempted to forgive the exaggeration as well-founded professional-insider criticism.)

It's true that the 1-2-3 happiness scale is widely used, but not because anyone honestly believes it captures happiness. It's an easily measured proxy variable for something related to happiness (we're not exactly sure what yet). Studies keep using the same proxy variables across many studies not because they have concluded that it is in fact the best measure or that it measures what they actually want to measure, but because this allows you to merge and compare studies coherently. Many other such proxy variables are used in tandem. Happiness research is currently on a simultaneous quest to understand what, exactly, these things measure, and how to measure more relevant things better, and to learn about happiness itself.

But it's definitely true that statements such as "happiness has not risen since the ’50s in the U.S. or Britain or (over a shorter period) in western Germany" make happiness researchers sound like idiots and mislead layreaders. Measurements of some happiness-related proxy variable that we don't fully understand may have not risen over that time, yet I doubt many people would choose to live in those earlier times, so you just can't make that statement with a straight face, caveat emptor. The oft-mentioned finding that "having children makes you less happy" is equally ridiculous for obvious reasons: that billions of people want children, enjoy having children, and don't regret having children, either at the time or especially after raising them, is overwhelming evidence that these results say more about the metric than the thing we hope it proxies for.

Wednesday, January 23, 2013

the beauty of economics

Apparently it takes a rabbi to express the beauty of economics! This is completely wonderful:


Economics writers would do well to become less like dismal scientists and more like this guy. Almost no one emphasizes that economists are just as much seeking to maximize human well-being as anyone else, but the methods that they know will work are frustratingly passive and often counterintuitive.

(The skill of making economics inspirational is also one of the reasons I love David Brooks. Milton Friedman was also often excellent at it.)

[Stolen from MR]

Friday, January 11, 2013

books

Cat's Cradle, by Kurt Vonnegut - I could not put this book down. And I never say that about fiction. It flies by in a whirlwind of gripping, thought-provoking, never-mundane action. Several of Vonnegut's (excellent) tips for writing fiction have to do with conciseness, and he executes fantastically.

The Honest Truth About Dishonesty, by Dan Ariely - I've read too many books that pertain to this subject lately so I can't give a fresh opinion. Not bad. But it felt too much like a conglomeration of studies and observations wrapped in a book cover, rather than a cohesive argument. I liked Liars and Outliers better - it's thinner on content, but he beats you over the head with a well-formed point.

What the Dog Saw, by Malcolm Gladwell - Finally read some Gladwell. This book was more thought-provoking than anything else I've read in awhile, but in the unintentional way, where the existence of the book itself and the industry and society that led to it being a bestseller was the train of thought, rather than any of his theses. But, if I start ranting about it, I'll never stop. Another day, another post. Punchline (and I know I'm not the first to say so): he's an amazing writer. Don't trust anything he says.

...

(In the last six months or so, I've been commuting to San Francisco by BART several times a week, which provides time when the only thing I can do without getting carsick is listen to my kindle with text-to-speech. Heavy nonfiction is hard to listen to because the stubbornly non-telepathic device refuses to pause when I stop to ponder a sentence. Fiction that is engrossing enough to listen to is exceedingly rare. Hence, a deluge of light pop-social-sci...)

Thursday, January 10, 2013

placebo effect

Very cool article on research on the placebo effect itself as a treatment.* (Via Kottke)

It makes me wonder two things in particular. (But read the article; most of it is about other things :)

First, what have drug companies already worked out about the procedural aspects of the placebo effect, in the course of designing RCTs in a way to maximize the chance of an outcome favorable to the drug?** This may not have been considered knowledge of direct interest in the past, but now it clearly is, and I bet they have a wealth of data/local knowledge of value. The article inadvertently makes it clear what great interest they have in placebo effects:
That study ... showed that patients with a certain variation of a gene linked to the release of dopamine were more likely to respond to sham acupuncture than patients with a different variation—findings that could change the way pharmaceutical companies conduct drug trials...Companies spend millions of dollars and often decades testing drugs; every drug must outperform placebos if it is to be marketed. "If we can identify people who have a low predisposition for placebo response, drug companies can preselect for them," says Winkler. "This could seriously reduce the size, cost, and duration of clinical trials…bringing cheaper drugs to the market years earlier than before."

Second, have they looked at Hawthorne / experimenter demand effects? Patients' reports of their symptoms and side effects are quite likely biased, and if they've been told that they should expect something, their reports are likely biased to pay more attention / exaggerate those things. What if they are told (at the time of the report) that they should be as objective and comprehensive as possible in their reports, for the sake of science, because they may or may not have been given the real drug and the doctors need accurate assessments for evaluation? Are reports standardized in such a way to maximize objectivity and comprehensiveness? Are multiple reports made over time to control for individual variation (as the article notes that one of these placebo studies was unique in doing)? Are all reports accompanied by objective physiological measures? The article suggests these things are important, but then implies that the cause is a discrepancy between subjective and objective experience, rather than false subjective reports. Both possibilities should be investigated.***
The researchers had hoped to find improved lung function with both the real and sham treatments; what they found instead was that only the real treatment yielded results—the others showed no significant improvement. Yet when Kaptchuk’s team measured patients’ own assessments of improvement, the researchers found no difference reported between the real and sham treatments: the patients’subjective responses directly contradicted their own objective physical measures...This discrepancy between objective and subjective results is precisely where the danger lies. "Asthma can be fatal. If the patient’s lung function is getting worse but a placebo makes them feel better, they might delay treatment until it is too late."

*I've idly wondered about this previously, but this article suggests that maybe honest expectations, created by directly lying to the patient, are not even necessary to trigger a placebo response after all, contrary to my intuition. Pretty cool.

**Not to necessarily accuse drug companies of anything unethical; you could just as easily describe that as "what situational knowledge, derived from years of experience, have drug companies acquired on how to keep doctor interactions / medical procedures as neutral as possible in order to get the cleanest measure of the impact of a drug"...

***Of course, I'm sure they have been to some degree; I'm commenting on a journalistic treatment, and we all know how reliable/thorough those tend to be...

Saturday, January 5, 2013

Happy birthday to Bayes' rule

On the 250th anniversary of the posthumous publication of Bayes' rule, this is appropriate. Good introduction to the frequentist/Bayesian debate.

I've been pretty confused by that debate - what exactly is the controversy, as parodied by e.g. xkcd - so this was nice to read since it put methods I'm familiar with in the context of the debate and doesn't focus on settings in which one side or the other is a straw man. Having read this, I still can't say I see what the big deal is. Obviously if you have a reasonable prior, you should use it, and obviously if you don't, some additional assumptions will be required to draw any kind of useful conclusion, and whatever kinds of assumptions you allow yourself will make a difference... both camps make assumptions about forms of distributions, models, etc, so I think the author is overly sympathetic to the frequentists when he says they claim for themselves the 'high ground of scientific objectivity'.