Subjective well-being across the life course among non-industrialized populations

Abstract

Subjective well-being (SWB) is often described as being U-shaped over adulthood, declining to a midlife slump and then improving thereafter. Improved SWB in later adulthood has been considered a paradox given age-related declines in health and social losses. While SWB has mostly been studied in high-income countries, it remains largely unexplored in rural subsistence populations lacking formal institutions that reliably promote social welfare. Here, we evaluate the age profile of SWB among three small-scale subsistence societies (n = 468; study 1), forest users from 23 low-income countries (n = 6987; study 2), and Tsimane' horticulturalists (n = 1872; study 3). Across multiple specifications, we find variability in SWB age profiles. In some cases, we find no age-related differences in SWB or even inverted U-shapes. Adjusting for confounders reduces observed age effects. Our findings highlight variability in average well-being trajectories over the life course. Ensuring successful aging will require a greater focus on cultural and socioecological determinants of individual trajectories.

Fig. 1.. Subjective life satisfaction scores among Baka, Punan, and Tsimane’.

Loess (dashed blue) and OLS regression (in red) using age and age². The midlife trough is 47.5 years for Baka and 57.6 years for Punan.

Fig. 2.. Subjective life satisfaction in PEN sample of forest-dependent sites in 23 countries (study 2).

Loess smooth (in dashed red) and second-order OLS regression (black) by age.

Fig. 3.. Probability of high life satisfaction in forest-dependent communities across continents (study 2).

Random intercepts are used for country-specific sites. None of the age effects are significant at conventional levels.

Fig. 4.. Subjective life satisfaction in forest-dependent communities from 23 countries (study 2).

Age is treated as a categorical variable in OLS regressions. Four sets of model predictions are shown: (i) age and sex (open circles); (ii) age, sex, and country as a random effect nested within continent (red circles); (iii) age, sex, sociodemographic controls, and similar random effects (gray circles); and (iv) age, sex, full set of controls, and random effects (blue circles). Covariates evaluated at sample means.

Fig. 5.. Depressed affect score by sex for Tsimane’.

Trendlines are from loess and OLS regression including age (b = 0.370, P < 0.0001) and age² (b = −0.002, P = 0.002), R² = 0.16.

Fig. 6.. Tsimane’ depression score and medical diagnoses.

Diagnoses reflect clinical condition categories evaluated by Tsimane Health and Life History Project physicians during routine checkups.