Subjective life expectancy is a measure that quantifies the perceived extent of one’s remaining life time. It is derived from respondent’s estimates of either the length of their whole life or the number of remaining years. At least three different questions have been used in empirical studies. First, they have been asked a direct question ‘To what age do you expect to live?’ (Mirowsky
1999). Second, respondents have been asked to estimate their chances (0–100%) of living to a given age/x more years (Hurd and McGarry
1995). Third, respondents have been asked to indicate on a 5-point scale whether they thought it likely that they would live another 10 year (Van Doorn and Kasl
1998) or till age 75/80 (Popham and Mitchell
2007; Van Solinge and Henkens
2010). As such, SLE may be conceptually related to other measures that capture the individual’s perceived life horizon or survival, such as nearness to death (Kotter-Grühn et al.
2010), subjective age (Bergland et al.
2014) and self-perceptions of ageing (Levy et al.
2002). Our measure for SLE taps the perceived likelihood of living to a specified age.
Explaining subjective life expectancy and actual mortality
Existing research on SLE has so far mainly focused on the clarification of its correlates. The bottom-up approach is dominant: models include a variety of measures for health conditions and health behaviour, and they control for actuarial correlates of general life expectancy (e.g. Adams et al.
2014; Hurd and McGarry
2002). These studies show that subjective life expectancy systematically varies across individuals in accordance with known risk factors for mortality, such as age and gender, poor health conditions, diagnosed diseases and health habits (Griffin et al.
2013; Hurd and McGarry
1995). Moreover, there is evidence that individuals take genetic information, such as family longevity into account (Van Doorn and Kasl
1998; Van Solinge and Henkens
2010; Zick et al.
2014) and adapt subjective life expectancy in response to new information, such as health change and the onset of diseases (Hurd and McGarry
2002). So far, only one study (Griffin et al.
2013) has explicitly studied the impact of top-down factors, that is psychosocial variables, in the development of an individual’s evaluation of their own longevity. Using a subsample of over 2500 older workers in the Australian 45 and up cohort study, Griffin et al. (
2013) found that optimism was significantly associated with SLE in the expected direction. Optimistic individuals reported higher survival probabilities. Given the proven relationship of psychological variables like optimism (Giltay et al.
2004), control beliefs (Bosma et al.
2005), self-efficacy (Kaplan et al.
1994), life satisfaction (Gerstorf et al.
2008) and type D personality (Denollet et al.
1996) with mortality, it is remarkable that so few attention has been paid so far to top-down effects in the formation of subjective life expectancy.
We combine both approaches. Our model for understanding judgemental processes underlying subjective live expectancy assumes that SLE is a result of both bottom-up and top-down processing. In line with the implicit assumptions underlying previous research on this issue, we assume that individuals have a basic understanding of trends in general life expectancy as reflected in actual statistics and that they take their own genetic background, their health and functional status and (behavioural) risk factors into account in their subjective evaluation of life expectancy.
In addition, we assume that global features of personality influence the way a person perceives or evaluates information. We include two psychological variables that are deemed important. In the first place, self-efficacy also referred to as functional optimism (Schwarzer and Jerusalem
1995). Functional optimism pertains to the belief that the future will be positive because one can control it more or less. Second, satisfaction with life (SWL). SWL refers to a person’s evaluation of his/her life as a whole (Diener
1984). There is evidence for the proposition that a person who has a generalized expectancy of good outcomes in life tends also to have positive expectancies when evaluating life as a whole (Scheier and Carver
1985). We argue that this may also apply to survival probabilities. This leads to the following hypothesis:
Evidence is inconsistent regarding the
predictive validity of SLE on actual mortality. Among the existing studies, some (Hurd and McGarry
2002), but not all (Kotter-Grühn et al.
2010; Siegel et al.
2003) found an association of SLE with individual mortality when introduced into a model together with health and socio-economic variables. Some studies found associations in some subpopulations, but not in others (Adams et al.
2014; Van Doorn and Kasl
1998). The heterogeneity of results of existing studies warrants an additional examination of the potential sources of this heterogeneity. Apart from differences in sampling, follow-up period and measurement instruments for SLE, this heterogeneity in findings may result from the fact that models vary in terms of confounding factors that have been taken into account. Existing studies on the predictive validity of SLE on mortality exclusively focused on bottom-up factors. To our knowledge, no research has incorporated top-down factors. Given that psychological factors may play a role in the individual’s evaluation of his or her survival probabilities (Griffin et al.
2013) and in the light of the growing evidence that psychological traits and dispositions predict mortality (Chida and Steptoe
2008; Rasmussen et al.
2009), this is remarkable. We include both bottom-up and top-down factors as confounders in our model explaining the predictive validity of SLE on mortality. Our approach is explorative. We will first examine whether the factors that play a role in the judgemental process underlying SLE predict actual mortality as well. We assume the following:
Next, we will investigate whether or not SLE predicts mortality, crude and adjusted for bottom-up and top-down factors. Following Griffin et al. (
2013), we assume that individuals have a basic understanding of the risk factors for mortality, and that they take this information into account in the subjective evaluation of their own life expectancy. We therefore expect that SLE predicts actual mortality, but that the predictive validity of SLE on mortality will decline when the bottom-up factors are included as potential confounders.
Additionally, we will investigate to what extent SLE reflects psychological traits and dispositions, such as optimism or psychological well-being, that can influence the length of life. People with an optimistic life orientation experience life and life events in a more positive way and expect more positive outcomes than pessimists (Scheier and Carver
1985). A positive life orientation is believed to be beneficial to health, as optimistic individuals appear to have more supportive social networks, use adaptive coping strategies and have different health habits, than pessimistic individuals (Kivimäki et al.
2005). There is indeed evidence that optimism is a predictor of physical health outcomes (including mortality) (Rasmussen et al.
2009) and that positive psychological well-being (happiness, optimism and life satisfaction) has a favourable effect on survival in both healthy and diseased populations (Chida and Steptoe
2008). We therefore expect that SLE predicts actual mortality, but that the predictive validity of SLE on mortality will decline when the top-down factors are included as potential confounders. We included life satisfaction as a measure for psychological well-being. Following Schwarzer and Jerusalem (
1995), we use self-efficacy as an indicator of optimism. We assume the following: