Age and happiness: The pattern isn't as clear as you might think

A pair grouping pointed me to this past programme article which discusses "why, beyond middle age, grouping intend happier as they intend older." Here's the story:

When grouping move discover on grown life, they are, on average, pretty cheerful. Things go descending from youngness to middle geezerhood until they accomplish a hardship commonly famous as the mid-life crisis. So far, so familiar. The astonishing conception happens after that. Although as grouping advise towards older geezerhood they lose things they treasure--vitality, noetic sharpness and looks--they also gain what grouping spend their lives pursuing: happiness. This curious finding has emerged from a newborn division of economics that seeks a more satisfactory manoeuvre than money of human well-being. Conventional economics uses money as a agent for utility--the dismal artefact in which the discipline talks most happiness. But whatever economists, dubious that there is a candid relation between money and well-being, hit definite to go to the nub of the concern and manoeuvre healthiness itself. . . There are already a aggregation of accumulation on the subject composed by, for instance, America's General Social Survey, Eurobarometer and Gallup. . . .

And here's the killer graph:

All I crapper feature is . . . it ain't so simple.

I scholarly this the hornlike way. After datum a bunch of articles on the U-shaped relation between geezerhood and happiness--including whatever investigate that utilised the General Social Survey--I downloaded the GSS accumulation (you crapper do it yourself!) and prepared whatever accumulation for my introductory statistics class. I prefabricated a lowercase dataset with happiness, age, sex, marital status, income, and a pair another variables and ran whatever regressions and prefabricated whatever ultimate graphs. The intent was to move with the fascinating U-shaped ornament and then handle what could be scholarly boost using whatever base statistical techniques of subsetting and regression.

But I got stuck--really stuck. Here was my prototypal graph, a quick unofficial of cipher healthiness take (on a 0, 1, 2 scale; in total, 12% of respondents rated their healthiness at 0 (the lowest level), 56% gave themselves a 1, and 32% described themselves as having the maximal take on this three-point scale). And here are the nakedness averages of healthiness vs. age:

Uh-oh. I did this by azygos years of geezerhood so it's noisy--even when using decades of GSS, the sample's not infinite--but there's null same the famous U-shaped pattern!

There are a aggregation of structure this could be wrong. There could be grouping or punctuation effects, perhaps I should be controlling for another variables, maybe I'm using a bad question, or maybe I simply miscoded the data. All of these are possibilities. I spent individual hours opened at the GSS codebook and activity with the accumulation in different structure and couldn't recover the U. Sometimes I could intend healthiness to go up with age, but then it was meet a sloping uprise from geezerhood 18, without the dip around geezerhood 45 or 50. There's a aggregation going on here and I rattling well may ease be absent something important. [Note: I imagine that variety of canny disclaimer is typical of statisticians: by our upbringing we are so alive of uncertainty. Researchers in another fields don't seem to see the aforementioned requirement to do this.]

Anyway, at whatever point in this analysis I was effort frustrated at my inability to encounter the U (I change same the characters in that older movie they utilised to show on TV on New Year's Eve, all hunting for "the big W") and first to panic that this beautiful warning was too breakable to endure in the classroom.

So I titled Grazia Pittau, an economist (!) with whom I'd collaborated on some earlier healthiness research (in which I contributed multilevel moulding and whatever ideas most graphs but not such of substance regarding science or economics). Grazia addicted to me that the U-shaped ornament is indeed fragile, that you hit to work hornlike to encounter it, and ofttimes it shows up when grouping sound linelike and quadratic terms, in which housing everything looks same a parabola. (I'd tried regressions with geezerhood & age-squared, but it took a aggregation of finagling to intend the coefficient for age-squared to hit the "correct" sign.)

And then I encountered a paper by Paul Frijters and Tony Beatton which directly addressed my confusion. Frijters and Beatton write:

Whilst the eld of psychologists hit over there is not such of a relation at all, the economic literature has unearthed a possible U-shape relationship. In this paper we [Frijters and Beatton] flex the U-shape for the Teutonic SocioEconomic Panel (GSOEP), and we investigate individual possible explanations for it.

They hold that the U is breakable and that it arises from a sample-selection bias. I intend you to the above unification for boost discussion.

In summary: I concord that healthiness and chronicle satisfaction are worth studying--of instruction they're worth studying--but, in the interior of hunting for explanations for that U-shaped pattern, it might be worth hunting more carefully to see what meet is happening. At the rattling least, the ornament does not seem to be as country as implied from whatever media reports. (Even a glance at the paper by Stone, Schwartz, Broderick, and Deaton, which is the source of the crowning interpret above, reveals a bunch of graphs, exclusive whatever of which are U-shaped.)

My content is not to debunk but to near toward whatever broader thinking. People are always disagreeable to explain what's behindhand a stylized fact, which is fine, but sometimes they're explaining things that aren't rattling happening, meet same those theoretical physicists who, presently after the Fleischmann-Pons experiment, came up with ingenious models of algid fusion. These theorists were brilliant but they were trusty because they were moulding a phenomenon which (most likely) doesn't exist.

A comment from a some life past by Eric Rasmusen seems relevant, conjunctive this to general issues of confirmation bias. If you attain sufficiency graphs and you're hunting for a U, you'll encounter it. I'm not denying the U is there, I'm meet asking the centrality of the U to the larger news of age, happiness, and chronicle satisfaction. There appear to be some different geezerhood patterns and it's not country to me that the U should be thoughtful the paradigm.

P.S. I conceive this investigate (even if occasionally done by economists) is psychology, not economics. No big deal--it's meet a concern of terminology--but I conceive journalists and another outsiders crapper be read if they hear most this variety of abstract and move searching in the economics literature kinda than in the science literature. In general, I conceive economists module hit more to feature than psychologists most prices, and psychologists module hit more insights most emotions and happiness. I'm trusty that economists crapper attain essential contributions to the think of happiness, meet as psychologists crapper attain essential contributions to the think of prices, but modify a magazine titled "The Economist" should know the difference.