Presentation on theme: "THE SET-POINT THEORY OF WELL- BEING IS IN TROUBLE – ON THE EVE OF A SCIENTIFIC REVOLUTION? Bruce Headey Melbourne Institute & DIW Berlin November 2007."— Presentation transcript:
THE SET-POINT THEORY OF WELL- BEING IS IN TROUBLE – ON THE EVE OF A SCIENTIFIC REVOLUTION? Bruce Headey Melbourne Institute & DIW Berlin November 2007
www.melbourneinstitute.com MOTIVATION The main point I am trying to make in this paper is that the most widely accepted theory of subjective well-being (SWB) – set-point theory – has serious flaws & probably now needs replacing. Set-point theory has dominated the field for 30 years – it has served as a classic Kuhn (1962) research paradigm. The main claim of the theory is that adult SWB does not change. It is supposed to be stable, except for temporary deviations due to major life events.
www.melbourneinstitute.com MOTIVATION Key evidence: German panel data (SOEP): In 1985- 2004 about 20% of this huge national sample recorded large & apparently ‘permanent’ changes in their levels of life satisfaction. Not compatible with set-point theory as currently understood. Need for a new theory? Are we on the eve of what Kuhn calls a ‘scientific revolution’? I think we probably are…. & I will present a few results – also from the SOEP panel data – which suggest lines of inquiry which may contribute to a new theory.
www.melbourneinstitute.com HISTORY OF A PARADIGM – why set-point theory seemed so convincing Set-point theory developed cumulatively over 30 years, successfully accounting for more and more evidence. By the mid-1990s it appeared to successfully link 3 sets of variables – personality traits & other more or less hereditary & stable individual characteristics, major life events, well-being (life satisfaction, positive affects) & ill-being (anxiety, depression, stress). New labels for old bottles: the genuine cumulativeness of theory development was partly disguised by the tendency of researchers to keep giving new labels to extensions of essentially the same theory.
www.melbourneinstitute.com “New labels for old bottles” Adaptation level (AL) theory – Brickman & Campbell (1971). First to observe over-time stability in SWB. Lottery winners and paraplegics. The Easterlin Paradox (1974). In Western countries economic growth does almost nothing for happiness – people just adapt to improved living standards & still want to “keep up with the Jones’s”. Personality theory of SWB (Costa & McCrae, 1980). Extroversion & neuroticism largely account for stable but varying levels of SWB. E and N are about 50% hereditary – so stable – but vary across individuals. Dynamic equilibrium theory (Headey & Wearing, 1989). History repeats itself in people’s lives – the same life events keep happening to the same people. So both events & SWB are partly endogenous. Set-point theory (Lykken & Tellegen, 1996). Twin studies.
www.melbourneinstitute.com Further developments in theory Multiple discrepancies theory (Michalos, 1985) Current levels of SWB can be understood partly in terms of various yardsticks: parents’ perceived SWB, SWB 5 years ago, SWB in 5 years time etc. Homeostatic theory (Cummins, 1995) Average SWB levels are high with low variance. People are highly motivated to keep reverting to these high levels even after adverse events. Peak-end theory (Kahneman, 1999) The ‘remembering self’ & the ‘experiencing self’ are different. SWB research is too exclusively concerned with the remembering self. This self usually only remembers ‘peaks’ & ‘ends’, and so sometimes reports results (e.g. in response to a life satisfaction question) which contradict actual experience. Compare white American & Asian-American students (Oishi, 2002).
www.melbourneinstitute.com DISCORDANT DATA – FIRST CRACKS IN THE THEORY Personal tragedies: As with many scientific theories, there always was some discordant data. It has long been known that some events are so tragic (esp. death of your own child) that people never recover back to their set-points. (Collective tragedies – e.g. defeat or occupation in war may have the same effect). Set-point theorists could perhaps just dismiss these events as ‘tragic exceptions’. Harder challenges: marriage & unemployment. Lucas et al (2003) showed that some people (although of course not most) appear to become permanently happier due to a successful marriage. Clark et al (2004) showed that repeated spells of unemployment have a ‘scarring’ effect & permanently lower SWB set-points. These not very uncommon events are harder to dismiss just as ‘exceptions’ to the theory. Cosmetic surgery works! Evaluations of cosmetic surgery show that it mostly DOES permanently enhance SWB (Wengle, 1986). Cosmetic surgery is not all that uncommon – especially in Los Angeles and Brazil.
www.melbourneinstitute.com THE CRACKS WIDEN – EASTERLIN DOES A U-TURN Richard A. Easterlin (2005) seems to be the first researcher to claim that set-point theory is seriously in error. He now believes that we must sharply distinguish between different domains of life. In the economic domain the Easterlin Paradox holds – people adapt back to set-point even after large economic gains or losses. But in the family domain and the health domain Easterlin finds that complete adaptation does not occur, although partial adaptation does (he cites Lucas et al, 2003 re family life & Mehnert et al, 1990 re health). Diener, Lucas & Scollon (2006) reviewed AL theory as set out by Brickman & Campbell (1971).They make 5 criticisms, some of which apply to set-point theory as currently understood. In particular, they note that evidence of medium & long term change in life satisfaction found in the SOEP panel calls for revision of set-point theory. Which events change satisfaction levels & which don’t?
www.melbourneinstitute.com SET-POINT THEORY CANNOT BE SALVAGED?? I want to take a different tack & suggest that set-point theory probably cannot be salvaged. Scientific theories cannot be partly right and partly wrong – we need a new theory. Why? The SOEP data destroy the central plank of the theory. If one takes 5 year averages of life satisfaction in order to eliminate transitory effects, it transpires that between 1985-89 & 2000-04 about 6% recorded large gains of 2 or more points (close to 1.5 standard deviations) on a 0-10 life satisfaction scale. About 13% recorded large declines in life satisfaction. (Sample=2872 West Germans & Auslaender remaining in the panel from 1985-2005). The country is at peace! These changes have all happened in peacetime & without a major economic slump. It is just not compatible with set-point theory to find that nearly 20% of the population have recorded large, long term changes in life satisfaction in these benign circumstances.
www.melbourneinstitute.com Possible new lines of theory – accounting for long term change So the new challenge for theory development is to account for medium & long term changes in SWB. Actually, we need to account for 2 things - both for the fact that most individuals don’t record large changes, and the fact that substantial minorities do. Two lines of theory which aim to go beyond dismissing long term change as due to ‘exceptional events’: - (1) personality traits linked to upside risk & downside risk: some individuals have personality traits, or combinations of traits, which increase their ‘upside risk’ of gains in SWB, or ‘downside risk’ of losses in SWB. (2) non-zero sum life goals: individuals who persistently pursue non-zero sum goals (family goals, altruistic goals) have a reasonable chance of recording gains in SWB (upside risk). Individuals who persistently pursue zero sum goals (financial gain, career goals) run a downside risk of declines in SWB.
www.melbourneinstitute.com Personality traits, upside risk & downside risk – SOEP 1985-2005 In 2005 SOEP included measures of the Big Five personality traits (N,E,O,A,C) for the first time. Extroversion (E) & neuroticism (N) are of particular interest to SWB researchers. E is positively related to SWB, & N is negatively related. Previous research has shown that people who score high on extroversion (E) repeatedly experience many positive life events & few adverse events (Headey & Wearing, 1989; Magnus et al, 1993). So these people, in a sense, ‘roll the dice’ more often than others – & their dice has a positive bias. They may be the individuals at greatest upside risk/probability of long term gains in SWB.
www.melbourneinstitute.com Personality traits, upside risk & downside risk – SOEP 1985-2005 Similarly, people who score high on neuroticism (N) experience many adverse events & so ‘roll the dice’ more often than others, but with a negative bias.They may be at greatest downside risk of declines in SWB. Openness to experience (O): the trait Openness to Experience (O) is not normally found to be directly related to SWB. But high scorers do repeatedly experience both more positive events and more negative events. So, extending the same logic, we can hypothesise that individuals who score high on both E and O (interaction term = E * O) may be at especially high upside risk of gains in SWB. Individuals who score high on both N and O may be especially prone to the downside risk of losses in SWB (interaction term = N * O).
www.melbourneinstitute.com How to test these hypotheses? Let’s begin by just looking at the effects of traits E and N on change in Life Satisfaction between 1985-89 and 2000-04 – take 5-year averages in order to largely remove the effects of temporary fluctuations. Then later we will look at the effects of trait O. Controls = gender, age & its quadratic, and Life Satisfaction 85-89
www.melbourneinstitute.com Long term change in Life Satisfaction: upside & downside risks Dependent Variable = Life Satisfaction 2000-04 minus Life Satisfaction 1985-89 All Men Women Age 25-59 Extroversion 0.14*** 0.12*** 0.16*** 0.08** Neuroticism -0.24*** -0.25*** -0.23*** -0.22*** Female 0.03 - - 0.06 Age 0.01 0.00 0.02 -0.14 Age squared -0.00 0.00 -0.00 0.02 Life Sat 1985-89 -0.55*** -0.55*** -0.54*** -0.50*** Adj. R squared 28.5% 27.8% 29.0% 23.6% Sample N 2872 1364 1508 1062 Source: SOEP balanced panel 1985-2005 ***sig. at 0.001 **sig. at 0.01
www.melbourneinstitute.com Interpretation The hypotheses relating to E and N receive some support. High E scorers ARE more likely than others to record gains in Life Satisfaction (controlling for where they were 20 years ago). High N scorers ARE more likely to record losses in Life Satisfaction. So it appears that long term changes in Life Satisfaction are systematically related to the sort of person you are…. to personality traits.
www.melbourneinstitute.com A minor setback – no sensible results yet relating to Openness (O) The logic of this approach suggested that trait O should also be implicated. BUT adding interaction terms (E*O) and (N*O) to the equations just given produced only nonsense. This may have been due to multicollinearity (r E,E*O = 0.82). But dummy variable approaches did not work either.
www.melbourneinstitute.com Do Life Goals Matter To Happiness? Now a second line of theory – life goals and ‘happiness’ There is increasing evidence that life goals matter to SWB. This is perhaps the most damaging evidence so far against set-point theory. The idea (if correct) that sensible choice of life goals can make a difference to SWB is plainly incompatible with the basis of set-point theory, which emphasises heredity.
www.melbourneinstitute.com Life Goals: zero sum & non-zero sum goals The idea that choice of life goals might help to account for long term changes in SWB comes partly from thinking about the implications of Easterlin’s (2005) review of the literature on the family life & health domains versus the economic domain. Perhaps one key point is that the economic domain is mainly zero sum (one person’s gain is another person’s loss), whereas some other domains, including family life, health and community activities are mainly non-zero sum or even positive sum (my gain does not entail your loss & may even bring gains to you).
www.melbourneinstitute.com Life Goals: zero sum & non-zero sum goals Hypotheses: (1) Individuals who give priority to non-zero sum life goals have a significant probability of achieving gains in SWB. (2) Individuals who give priority to zero life goals tend, on average, to record no gains or losses in SWB. Simple rationale: It is easier to achieve non-zero sum goals than zero sum ones, because no-one is competing with you. That is presumably why, in all countries, people have higher average satisfaction scores for domains relating to families & personal relationships than to domains related to work & material standard of living.
www.melbourneinstitute.com Testing hypotheses about life goals The SOEP also has questions about life goals. In several surveys respondents have been asked to rate the importance they attach to: (1) career & success goals (2) goals related to family life (3) altruistic goals related to friendships & community activities Perhaps we can loosely classify (1) as zero sum goals, and (2) & (3) as non-zero sum goals.
www.melbourneinstitute.com Life goals linked to Life Satisfaction (SOEP 2004 data. Sample N=8026) Dependent variable = Life Satisfaction 2004 Importance of: Success goals -0.09* Family goals0.26*** Altruistic goals0.16*** Results are net of gender, age & its quadratic, E, N, internal locus of control, marital status, years of education, household disposable income and disability status. Adj. R squared = 15.8%. ***sig. at 0.001 *sig. at 0.05
www.melbourneinstitute.com Interpretation It appears that choice of life goals may be linked to life satisfaction. Notice that these moderate linkages hold up, controlling for a wide range of other variables. Further, it appears that SOEP respondents who have persisted in giving priority to non-zero sum goals for the last 15 years (they were questioned on four occasions between 1990 & 2004) show small but statistically significant gains in Life Satisfaction, whereas those who have persisted in prioritising zero sum goals show small losses. Corroborative evidence: (1) NMRI research shows that altruistic acts strongly stimulate the pleasure centres of the brain (Rilling et al, 2004). (2) Repeated acts of kindness appear to enhance SWB (Lyubomirsky, 2005).
www.melbourneinstitute.com Concluding points - what might a new theory of SWB look like? First, to repeat an earlier point: a worthwhile new theory of SWB needs to account for the substantial minority who record large changes, as well as the majority who don’t. Secondly, a new theory needs adequate scope: it would now need to account for linkages among at least four sets of variables: stable person characteristics (inc. personality traits), life goals, life events, and measures of both well-being & ill-being. Don’t just focus on WB. It is probably a mistake that, in recent times, SWB researchers (esp. economists) have increasingly specialised in well-being, rather than trying to account for the partly overlapping and partly different correlates & causes of well-being & ill-being. Better to go back to the approach of the mid-West founding fathers of SWB research – Bradburn, Andrews & Withey, Campbell, Converse & Rodgers – who measured both WB and IB. Better measurement? Need to collect on-line or DRM measures of SWB - not rely only on recall measures provided by the ‘remembering self’. Advances in science usually require advances in measurement.
www.melbourneinstitute.com Promising lines (contd) I have tried to indicate what I regard as two promising lines of theory development, relating to personality traits and to life goals. I think another promising avenue is suggested by recent work in personality theory which indicates that adult personality may not be all that fixed (Roberts, Walton & Viechtbauer, 2006). It now appears that people can change quite a lot in their 20s, and even that change in later life is not all that uncommon. It may be that, in SWB research, we will be able to identify positive & negative feedback loops, linking positive (high E?) and negative (high N?) personality traits with gains and losses in SWB, followed by further positive and negative personality changes.
www.melbourneinstitute.com Exciting Times? So these should be exciting times in SWB research. Set-point theory (in my opinion) is not worth crying over. It stunted research by implying that the main task was to keep on looking for variables which usually keep people close to their set-points. (Also ok to investigate ‘events’ which produce transitory changes in SWB). The theory implied that adults (esp. older folk) could not do very much to increase their own SWB, nor could Governments do much. It is not rocket science, but I think we should be pleased to have the opportunity to contribute to our own minor ‘scientific revolution’ by developing a new theory of SWB.