Taking the satisfaction (and the life) out of life satisfaction

Abstract

The science of well-being studies an evaluative kind, well-being, which raises natural worries about the ability of empirical research to deliver. This paper argues that well-being research can provide important information about how people are doing without entangling itself very deeply in controversial normative claims. Most life satisfaction research, for instance, purports only to tell us how people see their lives going relative to what they care about – something most people can agree is important, whatever their theory of well-being. At the same time, such research can mislead if it does not in fact have the significance it seems to have. Life satisfaction measures in fact have this problem: life satisfaction attitudes have deep in-principle limitations in their ability to reflect the subjective quality of people's lives, for instance because they are only weakly constrained by the facts about how people see their lives going. You might, e.g., reasonably be satisfied with your life even if you think it's going badly for you. I propose largely replacing such measures with aggregate life evaluation measures, which dispense with both the global judgment and ‘satisfaction’ elements of life satisfaction.

Keywords:

• well-being
• subjective well-being
• life satisfaction
• psychology
• happiness
• measurement

Acknowledgements

Some of the material in this paper was previously delivered in a commentary on Valerie Tiberius, ‘Well-Being: Psychological Research for Philosophers’, at ‘Well-Being: A Cure-All for the Social Sciences’? – an online conference sponsored by Wiley-Blackwell, November 2010. The author thanks Valerie Tiberius and an anonymous referee for this journal for helpful comments and discussion.

Notes

For surveys of the literature here, see, among many other good sources, Eid and Larsen (2008), Tiberius (2006), and Layard (2005).

Here I use these terms interchangeably to refer to matters of value.

Philosophers commonly distinguish three basic views about well-being: hedonism, desire satisfaction theories, and objectivist or ‘objective list’ theories. Hedonism is the best-known mental state theory. Mental state views are widely rejected, particularly because of ‘experience machine’ type worries: you could have any manner of mental states while stuck in a virtual reality device, as in the Matrix, which does not strike most people as very appealing. Also influential are ‘happy slave’ worries, whereby one might be happy leading a seemingly impoverished life. For a review of the well-being literature, see Crisp (2005).

I will argue that life satisfaction is not in fact very important, but will not dispute that the kind of subjective success life satisfaction is meant to capture does indeed matter. Some may doubt the importance of happiness on any plausible understanding of it, notably given the phenomenon of adaptation (happy slaves and the like). To a great extent such worries reflect problematic views of happiness, such as the life satisfaction theory critiqued here; indeed, my argument against life satisfaction offers a partial explanation of adaptation worries. As well, people can do pretty well even in circumstances that otherwise call for assistance. If you are leading a pleasant, fulfilling life but lack access to decent healthcare, say, you might have a high level of well-being yet still have a fairly urgent need for more resources.

Oswald and Wu (2010).

Could it merely endorse a position in the literature, such as Aristotle's, without trying to advance the philosophical debate? In principle, yes; the trouble is that attempts to operationalize a philosophical theory often require doing philosophy, since the theory needs interpreting: what is the most coherent, charitable interpretation of Aristotle's views? Would measures of subjective meaning and capacity-development, with no reference to objective standards of excellence, adequately assess well-being as Aristotle saw it? Could any cross-sectional study, which disregards the overall shape of a life, do this? These are substantially philosophical questions, and even Aristotle scholars will often disagree on their answers.

‘Can be’ because hedonists about well-being may not accept the possibility of defeaters.

E.g. Diener and Diener (1996). This remains a consensus view among SWB researchers.

For reviews of the recent literature, see Eid and Larsen (2008), especially Pavot (2008).

Haybron (2005, 2008).

Fredrickson and Losada (2005) and Larsen and Prizmic (2008).

I have argued for claims along these lines in Haybron (2005, 2007, 2008, forthcoming). Here I significantly sharpen and extend the critique of life satisfaction, and suggest a positive alternative.

Haybron (2008).

The qualifier, ‘in itself’, is strictly needed since there may be situations where we already know a lot about how people tend to evaluate their lives. E.g. expressions of dissatisfaction among Americans are sufficiently unusual, and strongly enough discouraged in the culture, that they may provide good evidence that the individual is faring poorly.

Or if it does tell us who is happy, then it does not matter very much whether people are happy.

E.g. Cummins (1995).

While this is my fairly strong hunch, there are other possibilities, and in reality there is likely lots of variation across subgroups; I am talking only about an overall tendency. But maybe poor people compensate for disadvantage by cheering themselves up or motivating themselves with a sunnier outlook, while rich people can more easily get away with griping, or have higher expectations which lower their self-reports even when doing equivalently well. In any event, any skewing of self-reports between groups is a problem: for the reliability of life satisfaction measures, it does not matter whether positivity biases are stronger with the rich or the poor; so long as they systematically differ, that is a problem.

This problem does not necessarily afflict all ‘life satisfaction’ instruments. The Cantril ‘ladder of life’ scale does not ask people to set a ‘good enough’ point, only to rate their lives from 1 to 10. So it is not strictly a life satisfaction instrument, and while still vulnerable to the preceding problem, it has psychometric virtues that may owe precisely to the absence of an actual ‘satisfaction’ judgment. Two important recent studies using this measure are Diener et al. (2010) and Kahneman and Deaton (2010).

The problem here is not just having too precise a bar. Even a vague threshold will be arbitrary.

See, e.g., Schwarz and Strack (1991, 1999). For a possible example of life satisfaction rising en masse while life subjectively gets worse, see a study of African-American well-being over the 1980s by Adams (1997).

At least, barring special circumstances, such as a population known to register satisfaction only when things are going exceedingly well for them.

Glatzer (1991).

For other criticisms, see, e.g., Carson (1981), Davis (1981) and Feldman (2010). Sen's critiques of happiness have the most force against life satisfaction theories (e.g. Sen 2009).

Such information is not completely irrelevant, since it may be useful to know that people view their lives in favorable terms, even if this does not tell us whether they see themselves as doing well. But this supports only a drastically reduced role for such measures.

E.g. perhaps, the remarkably positive life satisfaction results found among slum-dwellers in Calcutta (Biswas-Diener and Diener 2001), or, again, the 6–7% of ‘completely satisfied’ who also report being ‘usually unhappy or depressed’ (Glatzer 1991).

Given my earlier complaints about the notion of satisfaction, these terms are not entirely felicitous, but they should serve well enough here. In fact it is possible to view ALE as a kind of ‘life satisfaction’ instrument, as long as we are clear about the differences from the standard notion of life satisfaction.

Following the ‘U-Index’, which assesses the amount of time people's experience is predominantly unpleasant (Kahneman and Krueger 2006).

A referee for this journal suggested that the problems afflicting life satisfaction measures would be just as bad for ALE. But ALE is not an ‘all-in’ evaluation of a person's life, much less whether that life is good enough; it is an index of various localized evaluations about particular items (or domains) in a person's life, in this case just the most important pros or cons. (Perhaps the term ‘aggregate life evaluation’ is a bit misleading, then.) Even if it seems to you a coin toss whether to be satisfied with your life, it should not seem much of a coin toss whether to say that, say, having cancer or being laid-off is a major con in your life, and something you would very strongly like to be different.

Ubel and Loewenstein (2008), Torrance (1976), Riis et al. (2005).

From an economist's perspective, two (incommensurable) numbers are much more difficult to work with than one. Richard Layard, e.g., recently wrote that ‘rational public policy requires a single criterion for comparing the benefits of different types of expenditure’ (review of Seligman, Flourish, The Guardian, 15 May 2011). Granted, it would be possible to combine the D/S-indices (or any other measures) into a single number, so the advantage in practice may prove to be modest. But such an index would at least be pretty transparently artificial.

Additionally, its limited information base – just the most important pros and cons – means that minor pros and cons get left out, and this could sometimes yield significant errors for ALE measures. But offhand, it seems good enough for most purposes just to get the major items.