Trying to become happier may be as futile as trying to become taller. ~ Lyubomirsky et al 2005, p. 113
In the Pursuit of happiness: The architecture of sustainable change, Lyubomirsky, Sheldon & Schkade (2005) [LSS] argued that we can attain lasting happiness by doing certain things. This is big news after centuries of skepticism about humans’ ability to make themselves happier by their own efforts. Schopenhauer (early 19th century) doubted that there was such a thing as happiness; Schooler, Ariely, & Loewenstein (2003) thought that its pursuit will backfire; Gilbert (2006) accepted the reality of happiness, but thought you need to stumble into it; and Brickman & Campbell (1971) warned that the pursuit of happiness will turn into a hedonic treadmill which will eventually exhaust a person’s strength and resources.
LSS fire back by pointing to studies showing that certain practices can overcome this pessimism. Many of these practices express virtues such as gratitude, prosocial giving, or mindfulness. With the right timing and the right intensity, these practices can lift the happiness boat. Critically, these actions should be done intentionally, and some of them may be transformed into habits and a way of life.
Statistically, the effects of intentional action can be expressed as correlation coefficients or increments in standard units. Little of that can be found in LSS, however. One piece of evidence, coming from a then unpublished study, linked “positive activity change” to later well-being by a path coefficient of .14, which indicates a small effect assuming that the coefficient is standardized. The types of positive activities change are not specified. It might have been a bundle of activities. Another piece of evidence, also from then-unpublished data, links “acts of kindness” and “counting one’s blessings” to positive changes in well-being. These changes are .4 and .15 respectively. But what do these numbers mean? We do not know because no context is given.
LSS do, however, offer a quantitative sense of hope. With a now well-known pie chart they suggest that 50 percent of individual differences in happiness can be accounted for by genetic differences, as revealed by twin studies. Another 10 percent are accounted for by various circumstantial variables, such as ethnicity, socio-economic status, marital status, and age. LSS (p. 116) conclude that
this leaves as much as 40 percent of the variance for intentional activity, supporting our proposal that volitional efforts offer a promising possible route to longitudinal increases in happiness. In other words, changing one’s intentional activities may provide a happiness-boosting potential that is at least as large as, and probably much larger than, changing one’s circumstances.
Note that the 40 percent allocation to intentional activity is a value obtained by the method of subtraction. If we assume that 50 percent of all variation in happiness is genetic, that 10 percent is circumstantial, that intentional action is the only remaining source of variation, and that the estimates are free of error, then the remaining 40 percent must be attributable to intentional action. If these assumptions are met, the method of subtraction has logical force. But are they met?
Aside from the fact that the estimates for genetics and circumstance are merely rough approximations (as LSS note), they are likely to be underestimates (which they do not note). The percent explained variance can be expressed as a correlation coefficient by taking the square root of the percent variance estimate. Happy genes and circumstances are respectively correlated with happiness at .707 and .316. However, every measurement contains error. If these correlations were attenuated for unreliability, they would be larger and the room for intentional activity would be smaller. In other words, the pie chart exaggerates the room available for intentional action to affect happiness.
Now suppose we had measures of intentional action and we could correlate them with happiness. We could then estimate the percent explained variance and we could see if there is residual error variance after adding up the squared correlations between happiness and the three predictors. We could estimate the unreliability of the intentional-action measures and correct for them, which would drive up the correlation with happiness. We might even find that the sum of the three percentages of explained variance exceeds 100. How would that be possible? How could one account for more variance than there is? The sum of explained variance may exceed 100 if we ignore the possibility that the predictors are not independent. To the extent that predictors are correlated with one another, some of the variance in the outcome of happiness is explained by more than one predictor, thereby creating the impression that more is explained than there is. Methods of multiple regression can separate out such overlap, and the details of how that is done need not detain us here. The point is that the pie chart suggests an independence of the three types of predictor, and this assumption may not hold. It is plausible, for example, that genetic factors underlie people’s willingness and ability to engage in intentional action. If so, the pie chart overestimates the unique role of intentional action in the creation of happiness.
The pie chart identifies genes, circumstances, and intentional action uniquely as causes and happiness uniquely as an outcome. There may be reverse causation, however. Being happy may improve your circumstances, an effect that Lyubomirsky, King & Diener colleagues documented in the same year (2005).
The pie chart suggests that the class of circumstances is represented fully. Perhaps it is, but how do we know? Might there be circumstances beyond the usual demographic suspects? Any omitted circumstantial variable might increase the explained variance in happiness if it were included. If so, the room available for intentional action would shrink. The chart suggests that circumstances and genes are used to predict individual differences in happiness within the same type of population. However, studies in behavioral genetics are usually conducted with national samples (e.g., from a twin registry), whereas one of the major circumstantial effects on happiness is country of residence. Living in Denmark vs. living in Honduras makes a big difference. If the pie chart is meant to represent a national population, it may underestimate the role of circumstances and thereby overestimate the role of intentional activity.
Finally, the pie chart may oddly underestimate the potential causal power of intentional action to raise sustained happiness. Suppose everyone were to engage in intentional actions and be successful. Every person would gain X points on the scale of subjective well-being. If so, there would be no variation in intentional action and the correlation with happiness would be undefined. With intentional action out of the picture as a predictor of happiness, the variance in happiness explained by genes and circumstances would increase. LSS were aware of the possibility of across-the-board increases in action-induced happiness—though not of their implications. They write (p. 114) that
it is worth noting that heritability coefficients describe covariations, not mean levels. Furthermore, even a high heritability coefficient for a particular trait (such as happiness) does not rule out the possibility that the mean level of that trait for a specific population can be raised. Under the right conditions, perhaps anyone can become happier, even if her or his rank ordering relative to others remains stable.”
I suppose it is best to appreciate the pie chart for its most metaphorical message. There may be some room to increase sustained happiness by intentional intervention. That may be enough to motivate people to give it a shot if they so desire.
Return of the epigraph
Suppose you learned that 50% of the variation in body height was due to genes and 10% was due to circumstance (e.g., diet). Would you conclude that 40% is due to intentional action? LSS did, it seems, recognize the limits to the method of subtraction.
Upon reflection, and knowing that LSS no longer use the pie chart even metaphorically, I could have and perhaps should have titled this post Life (and Death) of Pie. As I said before, you can have your pie and eat it 2.
Happiness is a subjective experience. Or is it? Self-reports of happiness are vulnerable to various biases, which raises the specter of inaccuracy. A person might be (un)happy and not know it. If so, the definition of happiness as a subjective experience is false. For example, people do not adapt to noise. They will continue to show physiological signs of stress. However, they will think that they adapted because the noise, if monotone, will drop from consciousness. People will think they are not bothered by something they are not aware of.
According to one view, the physiological evidence for a continued stress response is hard evidence that the person is not happy. Stress and happiness are assumed to be mutually exclusive. According to the view that defines happiness as conscious reportable experience, the physiological stress is exactly that: physiological stress. It is not unhappiness if the person does not have a subjective feeling of unhappiness.
In my opinion, both views are false. If all is staked on a subjective report, one might have people who rationalize or hallucinate themselves into happiness without having the physiological support. Or if the physiology pointed to a different emotion (the person says she feels guilty but is blushing – which suggests shame), it would be dismissed. By contrast, if all is staked on the physiology, one ultimately ends up dismissing subjective experience and the report thereof. One would have to claim that unconscious affect is just as true and valid as conscious affect – perhaps even that it is a purer version of affect. Now, I don’t know about you, but I take pain that I am not aware of over pain that I am aware of any day. And I do not prefer happiness that I am not aware of over happiness that I am aware of.
In my opinion it is foolish to look for one true index of happiness. Usually, subjective experience and physiological state are reasonably well aligned. When they are dissociated, something is amiss. But such dissociations should not be treated as critical experiments to privilege one level of analysis over another.
I end with what I think of as a clever piece of repartee. I asked a colleague how he felt about unconscious affect. He gave me a puzzled look. I said I was hoping he would say “I don’t know.”
Brickman, P., & Campbell, D. T. (1971). Hedonic relativism and planning the good society. In M. H. Appley (Ed.), Adaptation-level theory (pp. 287– 302). New York: Academic Press.
Gilbert, D. (2006). Stumbling on happiness. New York: Random House.
Lyubomirsky, S., King, L., & Diener, E.(2005). The benefits of frequent positive affect: Does happiness lead to success? Psychological Bulletin, 131, 803-855.
Lyubomirsky, S., Sheldon, K. M., & Schkade, D. (2005). Pursuit of happiness: The architecture of sustainable change. Review of General Psychology, 9, 111-131.
Schooler, J. W., Ariely, D., & Loewenstein, G. (2003). The explicit pursuit and assessment of happiness can be self-defeating. In J. Carrillo & I. Brocas (Eds.), Psychology and economics. Oxford, England: Oxford University Press.
Schopenhauer, A. Any edition of The world as will and idea. Google knows where 2 find it.