Towards a Theory of Medium Term Life Satisfaction: Similar Results for Australia, Britain and Germany

We analyse the Life Satisfaction trajectories of respondents in three long-running, national panel surveys: the Household, Income and Labour Dynamics Australia Survey (HILDA), the British Household Panel Survey (BHPS) and the German Socio-Economic Panel (SOEP). Previous research has shown that substantial minorities of respondents in all three countries recorded long term changes in LS (Fujita and Diener in J Personal Soc Psychol 88:158–64, 2005; Headey in Soc Indic Res 76:312–317, 2006; Headey et al. in Proc Natl Acad Sci 107:17922–7926, 2010; Headey et al. Soc Indic Res 112:725–48, 2013). In a recent SIR paper based on the German data (Headey and Muffels in Soc Indic Res, 2015. doi:10.​1007/​s11205-015-1146-8), we showed graphs of LS trajectories which suggested—and subsequent statistical analysis confirmed—that respondents typically spend multiple consecutive years above and, in other periods, below their own long term mean level of LS. Here we extend the analysis to Australia and Britain, showing that results replicate in two more Western countries. It appears that most people go through relatively happy periods of life, and relatively unhappy periods. The evidence runs counter to set-point theory which views adult LS as stable, except for short term fluctuations due to life events. In the second half of the paper we try to contribute to a theory of medium term life satisfaction. We estimate structural equation models with two-way causation between LS and variables usually treated as causes of LS, including health, social support, frequency of social activities, and satisfaction with one’s work, partner and family life. In all three countries we find that there are positive feedback loops between these variables and LS, which partly account for extended periods of high or low LS. The two-way causation models are based on a modified concept of ‘Granger-causation’ (Granger in Econometrica 37(3):424–38, 1969). The main intuition behind Granger-causation is that if x can be shown to be statistically significantly related to y in a model which includes multiple lags of y, then it can be inferred that x is one cause of y.