The Importance of the Baby Boom Cohort and the Great Recession in Understanding Age, Period, and Cohort Patterns in Happiness

Abstract

Twenge, Sherman, and Lyubomirsky (TSL) claim that long-term cultural changes have increased young adults' happiness while reducing mature adults' happiness. To establish their conclusion, TSL use trend analyses, as well as more sophisticated mixed-effects models, but their analyses are problematic. In particular, TSL's trend analyses ignore a crucial cohort effect: well-known lower happiness among baby boomers. Furthermore, their data aggregation obscures the ephemerality of a recent period effect: the Great Recession. Finally, TSL overlook a key finding of their mixed-effects models that both pre- and post-Boomer cohorts became happier as they aged from young to mature adults. Our reanalyses of the data establish that the Baby Boomer cohort, the short-lived Great Recession, and unfortunate data aggregation account for TSL's results. The well-established, long-term relationship between age and happiness remains as it has been for decades despite any cultural shifts that may have occurred disfavoring mature adults.

Keywords: Baby Boomers; Great Recession; age–period–cohort; aging; happiness; subjective well-being.

Figure 1

Average levels of happiness among 18- to 29-year-old adults and adults aged 30 and over by time period. The 1982 and 1987 Black oversamples were removed, and the WTSSALL weight variable was used; Boomers include those born between 1945 and 1964, which includes an additional year (i.e., 1945) of birth than normally recognized by demographers, but this year range was selected to match TSL’s 5-year birth cohorts used in mixed regression analyses; see Table 1 for 95% confidence bounds.

Figure 2

Mature adults’ average level of happiness during the Great Recession era using single-year General Social Survey (GSS) waves. The 1982 and 1987 Black oversamples were removed, and the WTSSALL weight variable was used; Boomers include those born between 1945 and 1964, which includes an additional year (i.e., 1945) of birth than normally recognized by demographers, but this year range was selected to match TSL’s 5-year birth cohorts used in mixed regression analyses; see Table 1 for 95% confidence bounds.

Figure 3

Trends in the correlation between age and happiness across years: scatterplots and linear regression lines. The 1982 and 1987 Black oversamples were removed, and the WTSSALL weight variable was used; circle size reflects relative sample size.

Figure 4

(a) Cohort pattern in happiness net of age and period patterns: TSL’s age–period–cohort analysis (APC) model replicated. (b) Period pattern in happiness net of age and cohort patterns: TSL’s APC model replicated. (c) Period pattern in happiness net of age and cohort patterns: APC model with data disaggregated by single year.

Figure 5

Cohort and period patterns in happiness for persons ages 18–29 years and 30 years and over.