Further thoughts on happiness and life satisfaction research

As part of my continuing research project with Grazia and Roberto, I’ve been reading papers on happiness and life satisfaction research. I’ll share with you my thoughts on some of the published work in this area.

Alberto Alesina,, Rafael Di Tella, and Robert MacCulloch published a paper in 2004 called “Inequality and happiness: are Europeans and Americans different?”:

We study the effect of the level of inequality in society on individual well-being using a total of 123,668 answers to a survey question about “happiness.” We find that individuals have a lower tendency to report themselves happy when inequality is high, even after controlling for individual income, a large set of personal characteristics, and year and country (or, in the case of the US, state) dummies. The effect, however, is more precisely defined statistically in Europe than in the US. In addition, we find striking differences across groups. In Europe, the poor and those on the left of the political spectrum are unhappy about inequality; whereas in the US the happiness of the poor and of those on the left is uncorrelated with inequality. Interestingly, in the US, the rich are bothered by inequality. Comparing across continents, we find that left-wingers in Europe are more hurt by inequality than left-wingers in the US. And the poor in Europe are more concerned with inequality than the poor in America, an effect that is large in terms of size but is only significant at the 10% level.

Setting aside the “significant at the 10% level” business, there’s something that disturbs me about this article, interesting and informative as it is. My problem is that their data are aggregate, but their interpretations are at the individual level. They find correlations between the inequality levels of states (within in the U.S.) or countries (in the E.U.) and average happiness levels of people living in these places. But then they describe this as people being “unhappy about inequality” or “hurt by inequality” or “concerned by inequality.” And they present models of individual utility. I just don’t see any justification for them to jump from aggregate statistics to claims of individuals as being “hurt” by inequality rather than any other state- or country-level variables that happen to be correlated with inequality.

This is not to say their analysis is useless; I’m just surprised to see such casual use of causal vocabulary here. I always thought economists were more careful about this sort of thing.

Betsey Stevenson and Justin Wolfers published a paper, “Economic Growth and Subjective Well-Being: Reassessing the Easterlin Paradox,” in which they examined data from many countries over many years and found “a clear positive link between average levels of subjective well-being and GDP per capita across countries, and find no evidence of a satiation point beyond which wealthier countries have no further increases in subjective well-being.” They continue:

We [Stevenson and Wolfers] show that the estimated relationship is consistent across many datasets and is similar to that between subjective well-being and income observed within countries. Finally, examining the relationship between changes in subjective well-being and income over time within countries, we find economic growth associated with rising happiness. Together these findings indicate a clear role for absolute income and a more limited role for relative income comparisons in determining happiness.

This is an interesting paper, and an unusual paper to be written by two economists, in that it’s pure empirical analysis, with no individual-level utility model (or, as they call it in political science, a “formal model”). As a statistician, I find this refreshing to just hear from the authors what the data have to say, without having to interpret it through the often-annoying (to me) “empirical implications of theoretical models” framework.

Also interesting, in different ways, are the discussions. The first discussion is by Gary Becker and Luis Rayo. I’d say this discussion was disappointing, but given that Becker is famous for saying that “most (if not all!) deaths are to some extent ‘suicides'” (a description that happens not to fit the deaths of any of my four grandparents, just for starters), I can’t say that my expectations for his article were high or even moderate. What interested me about Becker and Rayo’s article was their defensiveness; in their words, “although reported happiness and life satisfaction may be related to utility, they are no more measures of utility than are other dimensions of well-being, such as health or consumption of material goods.” This is so bad I hardly know where to start. For one thing, “utility” is not real; it’s something that you can only define based on a model. The idea that they want “a more comprehensive theory of utility maximization” . . . ugh! I mean, sure, they can do this if they want, but please leave me and my grandparents out of it!

Becker and Rayo also write:

From a Darwinian perspective, since happiness has limited, if any, direct fitness value, it can easily lose its priority when the right opportunity arises. For instance, if we were to metaphorically “ask” our genes whether we should accept an unpleasant but high-paying job that would increase our social standing, with the only drawback of making us less happy, they would not hesitate for a moment.

Huh? Asking a question of your genes? Talk about interviewer effects!

Alan Krueger’s discussion is more interesting. Check this out:

People do not think about their life satisfaction or level of happiness in the same way they think about their mailing address or years of schooling. When asked, they construct an answer on the spot. They often use rules of thumb for providing their answers. They are also affected by their current mood and thoughts. In an ingenious experiment to demonstrate the importance of transitory mood on reported life satisfaction, Norbert Schwarz invited subjects to fill out a satisfaction questionnaire. Before answering the questionnaire, however, he asked them to make a photocopy of the questionnaire. For half of the subjects, a dime was planted on the copy machine. Reported life satisfaction was a point higher for those who encountered a dime!

Perhaps this is not a problem with large national surveys, but . . . maybe it is? For example, if you are surveyed at dinnertime and that annoys you, will this lower your reported life satisfaction? I don’t know what to think about this one, although maybe it’s an issue that Stevenson and Wolfers have thought about.

Krueger continues:

When people are asked about their own life satisfaction or happiness, they may reflect on their economic conditions and partly use that as a handle on providing an answer. The mental exercise that well-off respondents go through is probably something like, “I’m a fortunate person. I have a high-paying job. I live in a big house and I have an expensive car. I should report myself as satisfied with my life. If I don’t, I’m not a very responsible person.” This tendency is less likely to affect people’s moment-to-moment mood or affect. It is probably more than a coincidence that the measures Stevenson and Wolfers examine that are more closely related to how people felt yesterday (that is, their affect), as opposed to measures that reflect an evaluative judgment of how they feel about their lives over all, tend to be less related to income. Measures of well-being that are closer to actual feelings are probably less prone to bias from a focusing illusion.

This leads to a potential concern with some of the international data used in the paper. When respondents are asked to place themselves on a ladder of the best possible life in a survey that is represented as a world poll, they may be more prone to a focusing illusion that goes something like, “I live in a rich country with many amenities. I should place myself high on the ladder of life, regardless of how I feel in my own life moment to moment.” A reasonable concern is that the focusing illusion causes the ladder of life to exaggerate the effects of national economic development on people’s self-reported step on the ladder.

P.S. How should I think about the work of Gary Becker etc? On one hand, it’s ridiculous stuff; on the other, he’s a respected economist and said to be brilliant. The right analogy, I think, is to postmodern literary critics and other such intellectuals. Some of them are very smart, expert symbol-manipulators, and are very well respected in their fields. But to outsiders, they just seem to be talking mumbo-jumbo. To the extent their work is valuable, it is because it sparks insights in their readers or because it is commenting on more accessible work done by others. Becker’s work is like that, I think. It’s based on silly models of reality but makes sense within the walls of the academy. And, just as postmodernists can leap from claims such as the death of the text, to start making recommendations about government policy, similarly, Becker and his more or less empirical colleagues can jump from claims about “utility” to political recommendations. Whether it’s coming from postmodern literary critics or from economics, the political recommendations might be useful, but I’ll have to take them on their own terms without taking the theoretical logic too seriously.

7 thoughts on “Further thoughts on happiness and life satisfaction research

  1. Perhaps the comparison with postmodern critics is more apt then you intended. The literary critic Fredric Jameson suggests that there's something very postmodern about Becker's whole intellectual project, given the way it relativizes and flattens out all different kinds of human activities and goals by allowing virtully anything to be the source of some kind of "utility". (Jameson, *Postmodernism*, pp. 267-271.)

  2. But, to some outsiders, quantitative guys "seem to be talking mumbo-jumbo".
    Furthermore, how many times don't we do research based on 'silly models of reality but [that] makes sense within the walls of the academy'?
    Sometimes we assume independence of events when they are clearly dependents. Sometimes we assume that events are identically distributed when they are not.

    I am not saying there are not valuable work among quant people. What I am saying is that it is difficult to not use silly models of reality.

    Ok, maybe Becker's work shouldn't award him a Nobel, and I agree with you on that.

    Anyway, I am paraphrasing what you said the other day commenting about Seth's remarks.

  3. Krueger's point about the extent to which life satisfaction measures are affected by simple manipulations of irrelevant factors (like finding a dime) is important for research on well-being. Examples like the dime study are mentioned frequently, but the evidence for these effects is not as strong as some argue. For instance, the studies are often based on extremely small samples and often never replicated. I have never read the dime study, and I don't think Krueger provided a citation to it in his comment (the citation he does provide is to a review that as far as I can tell does not mention the dime study). Therefore, it is difficult to evaluate the quality of the research or to check whether it has been replicated. What I do know is that other similar studies often do not replicate or replications are never attempted. Another famous effect is that making a life domain salient (for instance by asking a question about that domain) before asking about general life satisfaction will increase the correlation between satisfaction with that domain and satisfaction with life in general. Although there are a couple studies showing huge effects for such a manipulation, Schimmack and Oishi (2005, Journal of Personality and Social Psychology) had difficulty replicating this effect; and after reviewing all studies that tested it, they showed that the average effect is quite small. Another effect that often gets cited is that life satisfaction scores are strongly affected by the weather (based on a 1983 paper by Schwarz and Clore). Although this very-small-sample study has been cited many times, as far as I can tell, the weather part of the study has never been replicated.

    Krueger favors measures that focus on immediate affective reactions (“how are you feeling right now”), presumably because they do not require respondents to remember and aggregate across various life domains. This is a potentially desirable feature, depending on exactly how global judgments are made, and how difficult each step in the process really is. If respondents can't recall and aggregate, then measures that rely on these processes would not be valid. But as it stands, we are far from having a complete understanding of the processes behind well-being judgments, and it is not clear whether processes like those suggested by the dime study have a strong enough impact to affect validity. An alternative way of evaluating the validity of these measures (both the immediate affective reactions and the global judgments) is to examine their correlations with other measures, predictors, and outcomes. The evidence suggests that both types of measures are associated with relevant criteria in meaningful ways, but the correlates are different. Ironically, the fact that life satisfaction measures often behave the way intuitive theories of well-being would predict (e.g., people who are poor or who have worse health report lower life satisfaction) is sometimes used as evidence against their validity. The idea is that if income correlates with life satisfaction, it shows that people are just using their lay theories about what should correlate with happiness to come up with incorrect judgments that don't actually reflect their true levels of well-being. A simpler explanation is that things that suck actually undermine life satisfaction, and survey questions asking about life satisfaction are sensitive to these effects. Affective measures may have some other problems that prevent them from picking up these effects, or they may be assessing a different construct that is not actually affected by these life circumstances.

  4. Good comments about Becker. The day I'll agree with many economists that Becker is brilliant is the day he (a) makes a prediction that most people think is unlikely to be true that turns out to be true or (b) helps someone else do that. To the best of my knowledge he's never done either. I'm not terribly concerned by his odd economics-uber-alles ideas — better to judge people by their best work than their worst.

  5. Can't rule out an hypothesis by the way it was generated – but you need to decide whether or not too spend your time taking it seriously.

    In academe, if some continue to quote you – not completely negatively – you will very likely survive. Just survival of the fit/discourse as virtu (of Machivillie sense)

    K?

  6. Bit late to reply here, but I remember to have loved Beckers work when a student. It really demonstrated to me how you could use 'the economists toolkit' to basicly any problem and even get predictions that fit the real world (way!).

    His novelty was to apply economics to a much broader field than macro, but to down-to-earth stuff as criminal punishment and marriage and family. Perhaps he has gone loco, but he was one of the starters of the whole freakonomics class of studies.

  7. First on Becker: His simple goal is to be as reductionist as possible, buidling a lean theory that infers a lean language and a kind of rigidity on the micro level that is simply intended to explain macro phenomena. His complete disinterest in indiosyncratic preferences, or to put it his terms idiosyncratic sources of utility are irrelevant as remains irrelevant to explain the overall macro phenomena. A good illustration is in Stinchcombe's "Constructing Social Theories" on pages 67-68 (1968).

    Especially in the field of sociology, in which researcher are very sensitive towards language and meaning, Becker makes it really hard for researchers like us that use Rational Choice (RC) to connect or to establish some kind of bonds towards the non RC-folks.

    To Alesina et al.: Several times I have wondered about these connections of the micro and the macro level, I guess they just validate this theorizing on the ground of the mathmatical models in which they combine micro and macro level quiet easily through algebra. Some goes with the new article where they connect family values to labour market regulations: While at first it seams ridiculous to draw a direct connection, their reasoning and mathmatical proofs provides an interesting explanation than can be backed up empirically. It leaves you with tons of interesting ideas, but Alesina and colleagues simultaneously leave you with a bad taste in your mouth.

Comments are closed.