Introduction
As a consequence of declining fertility rates and increasing life expectancies, countries around the world are faced with rapidly ageing populations. In Europe, people aged 60 or older are projected to make up more than a third of the population by the year 2050 (United Nations 2017). Against this backdrop, how the growing population of older people constructs their quality of life (QoL) is an important area for research and public policy (compare with Gabriel and Bowling Reference Gabriel and Bowling2004).
The – explicit or implicit – goal of policies is often ultimately to benefit the QoL of (subgroups in) the population (Schuessler and Fisher Reference Schuessler and Fisher1985). This implies that a profound understanding of what QoL entails for the groups targeted is called for when developing or evaluating policies. In preparation of the current study, we assessed how QoL was operationalized in four frequently used measurement instruments developed specifically for older people. When the conceptualizations of QoL in later adulthood underlying such instruments are broadly embraced, they may not only determine how QoL in later adulthood is monitored and how policy measures and interventions are evaluated but also shape what policy makers aim for when designing measures and interventions. Hence, mismatches between how policy makers and middle-aged and older adults define QoL in later adulthood may contribute to policy efforts that do not really benefit the targeted people in a meaningful way. As stated by Bowling and Dieppe (Reference Bowling and Dieppe2005: 1550) ‘[there] is little point in developing policy goals if [older] people do not regard them as relevant’.
Although there is consensus that QoL in later adulthood is a multi-dimensional construct, scholars differ in their perceptions on exactly which dimensions the construct comprises. Content validity, that is, the coverage of all facets of a construct of interest, is a prerequisite for an adequate operational definition of QoL in later adulthood. The current study will shed light on how well several factors that are considered key dimensions of QoL in later adulthood in one or more commonly used instruments correspond with what middle-aged and older laypersons in the Netherlands themselves perceive as constituents of QoL in later adulthood. We fielded a factorial design vignette experiment in which we assessed the extent to which various presumed dimensions of QoL contribute to QoL evaluations of fictitious older persons.
The current study extends the largely exploratory literature on lay perspectives on QoL in later adulthood in multiple ways. Instead of asking respondents to define the components of QoL themselves, as is commonly done (Bowling Reference Bowling1995; Holzhausen et al. Reference Holzhausen, Kuhlmey and Martus2010; Liu Reference Liu2006; Wilhelmson et al. Reference Wilhelmson, Andersson, Waern and Alleback2005), our approach involves presenting them with standardized vignettes. As described in further detail later, these vignettes are descriptions of a fictional third party, where the factors in the descriptions are varied randomly. This method allows for the assessment of the relative impact of the randomized factors on individuals’ responses, independent of respondents’ own characteristics (Auspurg and Hinz Reference Auspurg and Hinz2015; compare with Whitley et al. Reference Whitley, Benzeval and Popham2020). Rather than relying on small non-probability samples, as many exploratory studies of laypersons’ perspectives on QoL in later adulthood have tended to do (Holzhausen et al. Reference Holzhausen, Kuhlmey and Martus2010; Liu Reference Liu2006; Stenner et al. Reference Stenner, Cooper and Skevington2003; Van Leeuwen et al. Reference Van Leeuwen, Van Loon, Van Nes, Bosmans, De Vet, Ket, Widdershoven and Ostelo2019), we moreover draw on a large population-based sample of 2,544 middle-aged and older laypersons who participated in the Longitudinal Internet studies for the Social Sciences (LISS) panel in the Netherlands. Hence, the current study combines the internal validity of an experiment with the external validity of a true population sample (Mutz Reference Mutz2011; compare with Scherpenzeel Reference Scherpenzeel2011).
Background
Commonly used operationalizations of QoL in later adulthood
Several instruments have been developed to operationalize QoL. Each of these instruments is an operational translation of a specific conceptualization of QoL. As a starting point for the current study, we looked at how QoL was conceptualized in four frequently used instruments developed specifically for older persons: the World Health Organization Quality of Life questionnaire for Older adults (WHOQOL-OLD) (Power et al. Reference Power, Quinn and Schmidt2005), the Older People’s Quality of Life questionnaire (OPQOL) (Bowling et al. Reference Bowling, Hankins, Windle, Bilotta and Grant2013; Bowling and Stenner Reference Bowling and Stenner2011), the Control, Autonomy, Self-Realization and Pleasure questionnaire (CASP-19) (Hyde et al. Reference Hyde, Wiggins, Higgs and Blane2003; Wiggins et al. Reference Wiggins, Netuveli, Hyde, Higgs and Blane2008) and the Investigating Choice Experiments for the Preferences of Older People – Capability Index (ICECAP-O) (Coast et al. Reference Coast, Flynn, Natarajan, Sproston, Lewis, Louviere and Peters2008; Grewal et al. Reference Grewal, Lewis, Flynn, Brown, Bond and Coast2006). We selected these scales based on the extent of use rather than, for instance, their foundations in social theory and gerontology, because we intend to evaluate how well the most important instruments of QoL in later adulthood align with the perspectives of the older population they are meant to represent.
The WHOQOL-group of the World Health Organization defines QoL as ‘individuals’ perceptions of their position in life in the context of the culture and value systems in which they live and in relation to their goals, expectations, standards and concerns’ (WHOQOL Group 1998: 551). The WHOQOL-OLD was developed as an add-on module, to be used together with the WHOQOL-100 or WHOQOL-BREF, and covers QoL facets that are particularly salient for older adults (Power et al. Reference Power, Quinn and Schmidt2005). It is, however, also frequently used as a stand-alone measure (e.g. Bilgili and Arpacı Reference Bilgili and Arpacı2014; Henriques et al. Reference Henriques, Silva, Severo, Fraga and Barros2020; Marcos-Pardo et al. Reference Marcos-Pardo, Orquin-Castrillón, Gea-García, Menayo-Antúnez, González-Gálvez, Vale and Martínez-Rodríguez2019; Şahin et al. Reference Şahin, Özer and Yanardağ2019). The six facets of QoL in later adulthood recognized in the WHOQOL-OLD are: sensory abilities; autonomy; past, present and future activities; social participation; death and dying; and intimacy. The WHOQOL-OLD consists of six subscales, each consisting of four items, to capture the extent to which respondents score favourably on each of these six facets of QoL. The scores on the six subscales can subsequently be summed into a total QoL score (Gobbens and van Assen Reference Gobbens and van Assen2016).
The OPQOL finds its origins not in theoretical frameworks of QoL but in the themes of the questionnaire, which were derived from the perspectives of a sample of older adults which were checked against theoretical models (Bowling Reference Bowling2009; Bowling and Stenner Reference Bowling and Stenner2011). The original scale consists of 35 statements covering eight central domains: life overall; health; social relationships/leisure and social activities; independence, control over life and freedom; home and neighbourhood; psychological and emotional wellbeing; financial circumstances; and religion/culture. Favourable evaluations of these domains jointly constitute QoL in later adulthood as conceptualized in the OPQOL (Bowling and Stenner Reference Bowling and Stenner2011). The shorter version of the scale, the OPQOL-brief, consists of 13 items covering all original domains, with the exception of the religion/culture domain (Bowling et al. Reference Bowling, Hankins, Windle, Bilotta and Grant2013).
Building on the seminal work of Maslow (Reference Maslow1968), the scholars who developed the CASP-19 criticized the state of gerontological research in the twentieth century for having an overly strong focus on basic needs, such as financial independence, and too little consideration for the active and reflexive dimensions of being an older person (Hyde et al. Reference Hyde, Wiggins, Higgs and Blane2003). They argued that QoL in later adulthood should be conceptualized as the extent to which four domains of need are satisfied: control, autonomy, self-realization and pleasure (Hyde et al. Reference Hyde, Wiggins, Higgs and Blane2003). The original scale consists of 19 items, but a 12-item version in which the domains control and autonomy are combined also exists (Wiggins et al. Reference Wiggins, Netuveli, Hyde, Higgs and Blane2008).
The ICECAP-O index stands out as a QoL measure by focusing on capability, that is, the opportunities for individuals to pursue what they consider to be important in life (Grewal et al. Reference Grewal, Lewis, Flynn, Brown, Bond and Coast2006). Hence, the emphasis lies on appraising what a person can potentially achieve rather than solely focusing on their actual accomplishments. Specifically, the ICECAP-O defines QoL in later adulthood as the ability to achieve having attachment, role, enjoyment, security and control (Coast et al. Reference Coast, Flynn, Natarajan, Sproston, Lewis, Louviere and Peters2008; Grewal et al. Reference Grewal, Lewis, Flynn, Brown, Bond and Coast2006). For each of the five capabilities considered, respondents can indicate their capability on a four-point scale, and from these responses a total index score based on tariffs computed by Coast et al. (Reference Coast, Flynn, Natarajan, Sproston, Lewis, Louviere and Peters2008) can be calculated.
Dimensions of QoL in later adulthood
Although the various instruments described in the previous section reflect different conceptualizations of QoL, they also have overlapping dimensions. We employed an iterative approach to develop the vignettes for the current study, starting with an analysis of the dimensions from the WHOQOL-OLD, OPQOL, CASP-19 and ICECAP-O. In a back-and-forth process between both authors, the dimensions were continually refined and revisited until a final set of 11 dimensions was arrived at: (1) enjoyment, (2) autonomy, (3) sensory functioning, (4) social relationships, (5) social participation, (6) financial security, (7) secure living environment, (8) absence of fear of death, (9) psychological resilience, (10) self-realization and (11) control. We aim to evaluate the extent to which these 11 presumed constituents of QoL contribute to QoL in later adulthood as perceived by middle-aged and older laypersons in the Netherlands.
Enjoyment matters for ageing as it enhances overall wellbeing and gives meaning to daily life. Engaging in activities that spark joy – such as socializing, hobbies and outdoor experiences – fosters independence, boosts mental and physical health, and helps maintain a positive outlook on ageing (Borglin et al. Reference Borglin, Edberg and Rahm Hallberg2005; Jarosz Reference Jarosz2022; Van Leeuwen et al. Reference Van Leeuwen, Van Loon, Van Nes, Bosmans, De Vet, Ket, Widdershoven and Ostelo2019). The ability to have the enjoyment and pleasure one wants is one of the attributes of QoL recognized in the ICECAP-O (Coast et al. Reference Coast, Flynn, Natarajan, Sproston, Lewis, Louviere and Peters2008; Grewal et al. Reference Grewal, Lewis, Flynn, Brown, Bond and Coast2006). This corresponds closely with the pleasure domain in CASP-19, where pleasure is seen as an active process of being human and involves engaging in activities that bring happiness (Higgs et al. Reference Higgs, Hyde, Wiggins and Blane2003; Hyde et al. Reference Hyde, Wiggins, Higgs and Blane2003). Similarly, Bowling et al. (Reference Bowling, Hankins, Windle, Bilotta and Grant2013; compare with Bowling and Stenner Reference Bowling and Stenner2011) consider enjoyment of life as a key aspect of QoL in the OPQOL, by asking about whether one enjoys life overall and whether one still has things to look forward to.
Autonomy, that is, older people’s ability to maintain independence and make their own decisions, can contribute to a sense of freedom and dignity. People strive to remain in control, as dependency on others can negatively affect self-esteem and wellbeing (Borglin et al. Reference Borglin, Edberg and Rahm Hallberg2005; Van Leeuwen et al. Reference Van Leeuwen, Van Loon, Van Nes, Bosmans, De Vet, Ket, Widdershoven and Ostelo2019). All four QoL instruments highlight the importance of autonomy, albeit using different labels. The attribute labelled control in the ICECAP-O instrument, for instance, ‘involves being independent and able to make one’s own decisions’ (Grewal et al. Reference Grewal, Lewis, Flynn, Brown, Bond and Coast2006: 1897). Similarly, the facet labelled autonomy in WHOQOL-OLD is about independence in later adulthood and refers to the extent to which one is able to live autonomously and take one’s own decisions (Gobbens and van Assen Reference Gobbens and van Assen2016; Power et al. Reference Power, Quinn and Schmidt2005). The domain labelled autonomy in CASP-19 entails freedom from the unwanted interference of others (Hyde et al. Reference Hyde, Wiggins, Higgs and Blane2003).
Impairment or loss of sensory functions becomes more prevalent with age. Aside from the direct consequences, such as difficulty with daily tasks, it may also indirectly impact QoL by contributing to issues like loss of independence and social isolation (Tseng et al. Reference Tseng, Liu, Lou and Huang2018). The link between independence and health is also emphasized by Grewal et al. (Reference Grewal, Lewis, Flynn, Brown, Bond and Coast2006). Relatedly, Bowling et al. (Reference Bowling, Hankins, Windle, Bilotta and Grant2013) highlighted that being healthy enough to have one’s independence and to get out and about were considered the most important in a set of 35 items related to QoL according to the laypersons who participated in the workshops based on which the brief version of the OPQOL scale was developed. Hyde et al. (Reference Hyde, Wiggins, Higgs and Blane2003) also included an item in the CASP-19 about health as a potential barrier to autonomy. The WHOQOL-OLD furthermore considers sensory functioning to be a separate facet of QoL, distinct from the autonomy facet (Power et al. Reference Power, Quinn and Schmidt2005).
Meaningful connections and social interactions can help reduce loneliness, improve mental and physical health, and provide emotional support (Bergland and Narum Reference Bergland and Narum2007; Krause Reference Krause, Binstock, George, Cutler and Hendricks2006; Luong et al. Reference Luong, Charles and Fingerman2011). The ICECAP-O, OPQOL and the WHOQOL-OLD all recognize various aspects of satisfactory social relationships as constituents of QoL. The ICECAP-O attribute ‘attachment’ ‘incorporates feelings of love, friendship, affection and companionship’ (Grewal et al. Reference Grewal, Lewis, Flynn, Brown, Bond and Coast2006: 1897). The OPQOL contains items related to having social contact and social support, as well as to receiving love and affection, under the domain ‘social relationships/leisure and social activities’ (Bowling Reference Bowling2009; Bowling et al. Reference Bowling, Hankins, Windle, Bilotta and Grant2013; Bowling and Stenner Reference Bowling and Stenner2011). The intimacy facet included in the WHOQOL-OLD refers to ‘opportunities for companionship and love’ (Peel et al. Reference Peel, Bartlett and Marshall2007: 163).
Like social relationships, active social participation matters in the ageing process as it provides opportunities for engagement and a sense of purpose. This can help in combatting challenges associated with ageing, such as the loss of connections, and enhance mental health as well as physical health (Bergland and Narum Reference Bergland and Narum2007; Levasseur et al. Reference Levasseur, St-Cyr Tribble and Desrosiers2009). The OPQOL and the WHOQOL-OLD explicitly highlight social participation as an element of QoL. Social participation is one of the facets of QoL included in the WHOQOL-OLD and it ‘addresses satisfaction with level of activity and opportunity for community engagement’ (Peel et al. Reference Peel, Bartlett and Marshall2007: 163). Similarly, the OPQOL domain ‘social relationships/leisure and social activities’ contains items that focus on participating in social activities, hobbies, and paid or voluntary activities that give one a role in life (Bowling et al. Reference Bowling, Hankins, Windle, Bilotta and Grant2013; Bowling and Stenner Reference Bowling and Stenner2011). This, in turn, is closely linked to the ICECAP-O attribute ‘role’, which is focused on the ability to do activities that make one feel valued (Coast et al. Reference Coast, Flynn, Natarajan, Sproston, Lewis, Louviere and Peters2008).
Also shared among multiple conceptualizations of QoL are various aspects of security. Feeling secure is important for reducing stress and maintaining overall wellbeing as people age. However, studies have shown that worry tends to increase with age, and that these worries might decrease life satisfaction and can hinder a fulfilling later life (Brock et al. Reference Brock, Clemson, Cant, Ke, Cumming, Kendig and Mathews2011; Graham Reference Graham2003). Security is one of the five attributes of QoL in the ICECAP-O instrument, with Grewal et al. (Reference Grewal, Lewis, Flynn, Brown, Bond and Coast2006: 1897) defining it as ‘[incorporating] ideas of feeling safe and secure, not having to worry and not feeling vulnerable’. They furthermore highlighted people’s financial situation as a key factor shaping security. Adequate financial circumstances is also one of the domains of QoL considered in the OPQOL (Bowling Reference Bowling2009; Bowling and Stenner Reference Bowling and Stenner2011). In addition to financial security, the OPQOL recognizes feeling safe in one’s living environment as an aspect of QoL (Bowling Reference Bowling2009; Bowling and Stenner Reference Bowling and Stenner2011). Being free of concerns, worries and fears about death and dying is considered an important element of QoL in the WHOQOL-OLD (Power et al. Reference Power, Quinn and Schmidt2005).
Of the four QoL instruments considered, the OPQOL is the only one that conceptualizes QoL in later adulthood as also entailing psychological resilience. Psychological resilience, that is, the ability to bounce back from adversity, may be important for older adults’ QoL because they frequently encounter age-related adversities (Cosco et al. Reference Cosco, Howse and Brayne2017). The OPQOL questionnaire captures resilience within the dimension ‘psychological and emotional well-being’ by examining respondents’ tendency to embrace challenges with optimism and resourcefulness (Bowling Reference Bowling2009; Bowling and Stenner Reference Bowling and Stenner2011). Through items such as ‘If my health limits social/leisure activities, then I will compensate and find something else I can do’, it measures the capacity to remain optimistic and stay flexible in the face of challenges.
Self-realization may be important in later adulthood because it can provide people with a renewed sense of purpose beyond work and family responsibilities. As people grow older, pursuing new experiences, striving for personal growth and finding meaningful goals helps them stay engaged and fulfilled (Lulle and King Reference Lulle, King, Torres and King2023; Reichstadt et al. Reference Reichstadt, Sengupta, Depp, Palinkas and Jeste2010). Both the WHOQOL-OLD and CASP-19 consider self-realization to be a constituent of QoL. In CASP-19, self-realization is described as the active and reflexive process of being human (Hyde et al. Reference Hyde, Wiggins, Higgs and Blane2003). Within this domain, CASP-19 emphasizes personal satisfaction and accomplishment. Self-realization involves being open to developing new interests, as well as being open to new experiences. It entails recognizing and striving to reach one’s potential and feeling fulfilled in personal growth and development (Higgs et al. Reference Higgs, Hyde, Wiggins and Blane2003). Within the ‘self-realization’ domain, there is an item about experiencing that life is still full of opportunities. Relatedly, the WHOQOL-OLD domain ‘past, present and future activities’ contains an item about how satisfied one is with the amount of opportunities one gets (Power et al. Reference Power, Quinn and Schmidt2005). Both questionnaires thus consider the importance of engaging in new opportunities and feeling fulfilled in personal development.
Lastly, as people age, the ability to maintain control over their lives – whether through managing daily activities or through making decisions about their care – can become a key factor in preventing feelings of helplessness and a diminished sense of self-worth (Haak et al. Reference Haak, Fänge, Iwarsson and Dahlin Ivanoff2007; Van Leeuwen et al. Reference Van Leeuwen, Van Loon, Van Nes, Bosmans, De Vet, Ket, Widdershoven and Ostelo2019). Control is considered an element of QoL in CASP-19 and OPQOL. In CASP-19, it is described as the ability to actively intervene in one’s environment (Hyde et al. Reference Hyde, Wiggins, Higgs and Blane2003), whereas in OPQOL it is described as having control over the important things in life (Bowling Reference Bowling2009; Bowling and Stenner Reference Bowling and Stenner2011). Both domains refer to older people feeling that they still have control over the things that happen to them and to them maintaining the ability to do what they want.
Quality of life as a dynamic construct
As emphasized by Whitley et al. (Reference Whitley, Benzeval and Popham2020), who explored what laypeople in the United Kingdom considered to be important for ageing well, it is important to recognize that older people’s attitudes may continue to change as they age. Similar to the related concept of successful ageing (Badache et al. Reference Badache, Hachem and Mäki-Torkko2023), QoL is a dynamic construct, which means that people can change the standards by which they assess it (Allison et al. Reference Allison, Locker and Feine1997). Consequently, definitions of what constitutes QoL may not be stable over the lifecourse (Carr et al. Reference Carr, Gibson and Robinson2001; compare with Whitley et al. Reference Whitley, Benzeval and Popham2020).
Deeg (Reference Deeg, Mollenkopf and Walker2007) argued that the uncomfortable experience of mental incongruence may manifest itself when older persons lose aspects of life that they continue to consider to be important dimensions of QoL. She argued that this, in turn, calls for cognitive adjustment to new situations that cannot be changed, which is likely to come in the form of downgrading the importance assigned to the aspect of life that is lost (compare with Carr et al. Reference Carr, Gibson and Robinson2001) and assigning increased meaning to other aspects of life. In line with this reasoning, her analyses of within-person changes over a six-year period in the importance assigned to various aspects of life indicated that the importance that older persons assigned to physical health decreased with increasing age, whereas aspects like meaningful spending of time and having an adequate income and good housing were increasing deemed to be important as people aged (Deeg Reference Deeg, Mollenkopf and Walker2007). Qualitative work furthermore suggested that age-related health declines may lead to a growing recognition of joy and satisfactory family and social relationships, as constituents of QoL (Prieto-Flores et al. Reference Prieto-Flores, Fernandez-Mayoralas, Rosenberg and Rojo-Perez2010). In contrast to these findings, Whitley et al. (Reference Whitley, Benzeval and Popham2020) found, however, that, compared to their younger counterparts, older people assigned more, rather than less, importance to physical function for ageing well. The authors suggested that impactful problems that are relatively likely to occur in older age groups are sometimes trivialized by people in younger age groups in which these problems are less common.
In the current study, we aim to assess the extent to which each of the 11 presumed dimensions of QoL that have been discussed in this subsection contributes to QoL in the perspective of laypersons aged 50+ in the Netherlands. We acknowledge that people’s perceptions of what constitutes QoL may differ between age groups.
Data and methods
Participants
The current study’s vignette experiment was fielded among respondents aged 50+ of the Dutch LISS panel (Scherpenzeel Reference Scherpenzeel2011). Statistics Netherlands drew a random sample of independent, private households in the Netherlands, whereby people living in institutions and other forms of collective households were excluded. The LISS panel has been maintained by Centerdata at Tilburg University since 2007. The panel consists of 5,000 households in the Netherlands, comprising approximately 7,500 individuals aged 16+. For households that would otherwise be unable to take part, a basic computer and internet access are provided. Every month, panel members fill out online questionnaires and receive a financial incentive for each questionnaire completed. Since the initiation of the panel, Statistics Netherlands has drawn seven refreshment samples for the LISS panel, most recently in 2023, that were added to the panel over the years to account for attrition and sample ageing and maintain optimal representativeness.
Our vignette experiment was presented to LISS panel members in the autumn of 2023, as part of a question battery that also included a list experiment on loneliness in mid-life and later adulthood (cf. Van den Broek et al. Reference Van den Broek, Lam and Potente2024). A random sample of 2,848 panel members aged 50+ was drawn and these members were invited to participate in the module containing our vignette experiment. The arbitrary cut-off at age 50 for defining middle-aged and older panel members is in line with prior large research endeavours in gerontology, such as the Survey of Health, Ageing and Retirement in Europe (Börsch-Supan et al. Reference Börsch-Supan, Brandt, Hunkler, Kneip, Korbmacher, Malter, Schaan, Stuck and Zuber2013). The number of LISS respondents who completed the vignette experiment was 2,544 (participation rate: 89.3%). Although we did not ourselves collect socio-demographic information from our survey participants, such information was supplied by Centerdata from the LISS dataset. An overview of key socio-demographic background characteristics of the participants in the current study is presented in Table 1. As explained in further detail later, some of the analyses presented here will be stratified by age group. Therefore, age group specific descriptive statistics are also presented in Table 1, in addition to descriptives for the full analytical sample.
Sample characteristics
Notes: Data are from LISS panel; N: number of respondents.
Outcome variable and vignette dimensions
In the current study’s vignette, we presented a short interview with a fictitious woman in her early 70s. Several aspects of the interview were manipulated experimentally. After being presented with the fictitious interview, respondents were asked how they rated the QoL of the fictitious interviewee on a scale ranging from 1 (very poor) to 10 (excellent). This rating is our outcome variable.
The presented fictitious interviews that respondents were asked to base the QoL ratings upon consisted of nine questions. On every question, there were multiple response options that reflected either favourable or unfavourable positions on one or more of the 11 dimensions that we identified in the literature as potentially important dimensions of QoL. Per question, it was assigned randomly which response was presented. Table 2 provides an overview of the operational definitions of the 11 dimensions captured in the vignette, with each dimension having two levels (favourable versus unfavourable). In addition to the responses to the questions in the fictitious interview, the name and age of the fictitious interviewee was also randomly assigned, with the options being ‘Anna (74 years old)’ and ‘Lia (73 years old)’. Figure 1 provides an example of a fictitious interview such as those presented to the respondents.
Example of a vignette presented to respondents.
Analytical strategy
We presented two vignettes to each respondent whereby we constrained the second vignette so as to not have any overlap with the first vignette, to make the fictitious interviews as realistic as possible. Given that most questions in the fictitious interview had two response options (see Table 2), presenting more than two vignettes would have resulted in overlaps between vignettes, which plausibly would have undermined the extent to which the respondents empathized with the fictitious interviewees. It is not uncommon to ask respondents to rate multiple vignettes in order to maximize statistical power (e.g. Fleischmann and Koster Reference Fleischmann and Koster2018; Karpinska et al. Reference Karpinska, Henkens and Schippers2011; Oude Mulders et al. Reference Oude Mulders, Henkens, Liu, Schippers and Wang2018; Whitley et al. Reference Whitley, Benzeval and Popham2020), but asking respondents to rate two fictitious interviewees meant that the observations in the current study were not independent (Wallander Reference Wallander2009). Following prior work (e.g. McDonald Reference McDonald2020; Oesch et al. Reference Oesch, Lipps and McDonald2017; Van Houdt et al. Reference Van Houdt, Kalmijn and Ivanova2018), we adopted an individual fixed-effects approach to account for this issue. We analyzed within-person differences in the QoL ratings assigned to the vignette scenarios with a model that can be denoted as follows:
\begin{equation}{\ddot{y}_{ij}} = {\beta _1}{\ddot{x}_{1ij}} + \ldots + {\beta _k}{\ddot{x}_{kij}} + {\varepsilon _{ij}}\end{equation}Vignette dimensions
a Multiple dimensions were covered simultaneously by this question (i.e., financial security, secure living environment, and absence of fear of death). The favourable response (‘No, I rarely worry about things’) was used for all three of these dimensions. Unfavourable responses on one particular dimension were considered as favorable response on the other domains, as no explicit worries in these domains were mentioned.
where
\begin{equation}{\ddot{y}_{ij}} = {y_{ij}} - {\bar y_i}\end{equation}and
\begin{equation}{\ddot{x}_{kij}} = {x_{kij}} - {\bar x_{ki}}\end{equation} Equations 2 and 3 indicate that for both the outcome variable y (QoL rating) and the explanatory variables xi–xk (the vignette dimensions), we deducted the mean score for individual i over the two vignette scenarios presented to this respondent from the score for individual i on vignette number j. This procedure implies that in Equation 1 we regressed intra-individual variation in QoL ratings on intra-individual variation on the k dimensions of the presented vignette scenarios. Consequently, all characteristics of respondents i were accounted for, regardless of whether these characteristics could be observed (Allison Reference Allison2009). The observation-specific error term is denoted with 
${\varepsilon _{ij}}$. All models were estimated with cluster robust standard errors to account for the nested nature of our data (White Reference White1980). Formal tests of differences between the coefficient estimates 
${\beta _1} \ldots {\beta _k}$were subsequently performed using the 
$\texttt{lincom}$ command in Stata 18.0.
To assess whether the estimated QoL contributions of the dimensions considered varied as a function of the age of respondents, we re-estimated our fixed-effects models stratified by age group (50–59 years old; 60–69 years old; 70+ years old). Although arbitrary, this age group classification in ten-year bandwidths is in line with prior studies in gerontology (e.g. Harling et al. Reference Harling, Morris, Manderson, Perkins and Berkman2020; Ryan et al. Reference Ryan, Murphy, Boland, Galvin and Smith2018). An additional pooled fixed-effects model was estimated in which every term was interacted with age category to determine whether the age group differences in the coefficient estimates were statistically significant (compare with Van den Broek and Fleischmann Reference Van den Broek and Fleischmann2022).
The current study was assessed and approved by the Research Ethics Review Committee of the Erasmus School of Health Policy and Management at Erasmus University Rotterdam (reference: ETH2324-0130). A replication package with instructions on how to access the data, as well as annotated Stata code for the data preparation and all analyses presented here, is available on the Open Science Framework at https://osf.io/abx7r.
Results
Results of the fixed-effects regression analyses are presented in Table 3. The first model presented in Table 3 was estimated on the full analytical sample. With the exception of the name of the interviewee in the fictitious scenario, for which we did not expect any effect, all experimentally manipulated dimensions in the vignette were associated with statistically significant changes in the expected direction in the QoL evaluations. Substantial differences between dimensions in the magnitude of the contribution to QoL could, however, be noted. Figure 2 gives an overview of the contribution to the QoL ratings of the various dimensions ranked by estimated magnitude.
Estimated contribution to quality of life by dimension.
Results of fixed-effects regression analyses of quality-of-life ratings
Notes: Data are from LISS panel; models estimated with robust standard errors; Coeff.: coefficient; CI: 95% confidence interval.
a The coefficient estimate differs significantly (p < 0.05) from the corresponding coefficient estimate in the model for the age 50–59 subsample.
b The coefficient estimate differs significantly (p < 0.05) from the corresponding coefficient estimate in the model for the age 60–69 subsample.
c The coefficient estimate differs significantly (p < 0.05) from the corresponding coefficient estimate in the model for the age 70+ subsample.
* p < 0.05, ** p < 0.01, *** p < 0.001.
We performed formal tests of differences in the coefficient estimates for all combinations of dimensions (for full results, see Appendix A in the supplemental material). Most importantly, the results showed that enjoyment and social participation had a significantly stronger impact on the QoL ratings than any of the other nine dimensions considered. Enjoyment furthermore had a significantly stronger estimated QoL impact than social participation. The dimensions with the smallest estimated QoL impact were sensory functioning and having a secure living environment. Both of these dimensions had a significantly smaller QoL impact than any of the nine other dimensions considered in the model.
As indicated in the subsection on the analytical strategy, we re-estimated the fixed-effects model stratified by age group. The results of the analyses for the age 50–59, age 60–69 and age 70+ subgroups are presented in the second, third and fourth models in Table 3, respectively. A few systematic differences between the three age groups could be noted in the magnitude of the estimated effects. Specifically, self-realization contributed more to the QoL ratings among respondents aged 70+ than among their counterparts who were 50–59 years old (Δb: −0.194; 95% CI: −0.337, −0.050; p < 0.01) or 60–69 years old (Δb: −0.186; 95% CI: −0.321, −0.050; p < 0.01). Respondents aged 60–69 (Δb: 0.252; 95% CI: 0.105, 0.400; p < 0.01) and 70+ (Δb: 0.171; 95% CI: 0.028, 0.314; p < 0.05) rated sensory abilities as more important for QoL than did respondents who were 50–59 years old. Finally, enjoyment (Δb: −0.154; 95% CI: −0.290, −0.019; p < 0.05) and psychological resilience (Δb: −0.252; 95% CI: −0.481, −0.024; p < 0.05) were less important for QoL in the model for people aged 70+ than in the model for their counterparts who were 60–69 years old.
Regardless of these differences, it is important to note that, overall, the results were remarkably similar across the three age groups considered. This is clearly illustrated in Figure 3. The strong degree of similarity across age groups in the coefficients for the various dimensions suggests a high level of consensus in what middle-aged and older people perceive as important dimensions of QoL in later adulthood.
Estimated age-group-specific contribution to quality of life by dimension.
Discussion
For the current study, we identified 11 dimensions that were deemed constituents of QoL in at least one of four commonly used instruments of QoL in later adulthood (WHOQOL-OLD; CASP-19; OPQOL; ICECAP-O) and fielded a factorial design vignette experiment to assess the extent to which these dimensions shaped how laypersons aged 50+ rated the QoL of fictitious older women. Our results may shed light on the content validity of the four QoL instruments considered, that is, on the extent to which these instruments capture all facets of QoL in later adulthood. Generally, the dimensions covered by common QoL instruments align well with what our respondents intuitively associated with better QoL in later adulthood, as all 11 dimensions were, to a varying extent, significantly associated with changes in the QoL ratings in the expected direction. However, none of the four instruments considered cover all 11 dimensions included in our vignette experiment. We will elaborate on some notable omissions next.
Enjoyment was the dimension with the largest contribution to the QoL ratings in our vignette experiment. This may be because enjoyment is the broader experience of engaging in meaningful and pleasurable possibilities in life, which can overlap or even transcend other dimensions. Jarosz (Reference Jarosz2022) found that enjoyable daily activities can enhance overall wellbeing and reduce stress. She concluded that enjoyment is not about specific activities but about finding joy in various possibilities in life and the amount of pleasure one experiences. As such, the ability to find joy in everyday moments may help compensate for losses in other areas, making enjoyment a fundamental driver of QoL in later adulthood. However, in contrast to CASP-19, OPQOL and ICECAP-O, the WHOQOL-OLD does not include any items capturing enjoyment. Although the WHOQOL-OLD is frequently used as a stand-alone instrument (e.g. Bilgili and Arpacı Reference Bilgili and Arpacı2014; Henriques et al. Reference Henriques, Silva, Severo, Fraga and Barros2020; Marcos-Pardo et al. Reference Marcos-Pardo, Orquin-Castrillón, Gea-García, Menayo-Antúnez, González-Gálvez, Vale and Martínez-Rodríguez2019; Şahin et al. Reference Şahin, Özer and Yanardağ2019), it should be noted that it was developed as an add-on module to be used together with either the WHOQOL-100 instrument or its shorter version, WHOQOL-BREF (Power et al. Reference Power, Quinn and Schmidt2005). The WHOQOL-100 and WHOQOL-BREF were developed to measure QoL in the general population and both measures do include items about enjoying life (WHOQOL Group 1998). Our findings suggest that caution is called for when using the WHOQOL-OLD as a stand-alone measure because it may have suboptimal content validity when not used in combination with the WHOQOL-100 or WHOQOL-BREF.
In line with prior work (Farquhar Reference Farquhar1995; Van Leeuwen et al. Reference Van Leeuwen, Van Loon, Van Nes, Bosmans, De Vet, Ket, Widdershoven and Ostelo2019; Whitley et al. Reference Whitley, Benzeval and Popham2020), satisfactory social relationships and social participation both had a strong positive impact on the QoL evaluations reported here. Plausibly, social participation was one of the strongest contributors to QoL in later adulthood owing to its broad impact. Social participation allows older adults to stay engaged, maintain meaningful interactions and find new social roles (Jarosz Reference Jarosz2022), and it can foster resilience, psychological security and a sense of purpose (Dahan-Oliel et al. Reference Dahan-Oliel, Gelinas and Mazer2008; Jarosz Reference Jarosz2022; Li et al. Reference Li, Ge, Dong and Jiang2023). Yet, CASP-19 does not include any items that correspond with the dimensions of social participation or social relationships. The ICECAP-O, in turn, lacks items for other dimensions with substantial contributions to the QoL ratings, such as financial security.
The omission in CASP-19 and the ICECAP-O of dimensions that, in our analyses, are strong contributors to QoL evaluations is plausibly related to the rationale underlying these measures of QoL. Both the ICECAP-O and CASP-19 aim to make a sharp distinction between constituents and antecedents of QoL (Coast et al. Reference Coast, Flynn, Natarajan, Sproston, Lewis, Louviere and Peters2008; Grewal et al. Reference Grewal, Lewis, Flynn, Brown, Bond and Coast2006; Higgs et al. Reference Higgs, Hyde, Wiggins and Blane2003; Hyde et al. Reference Hyde, Wiggins, Higgs and Blane2003). For example, the scholars who developed the ICECAP-O consider the ability to feel secure, safe and free from the need to worry an attribute of QoL, whereas the conditions that may shape this ability are explicitly not considered to be aspects of QoL (Grewal et al. Reference Grewal, Lewis, Flynn, Brown, Bond and Coast2006). Hence, the reason not to include items about, for instance, financial strain in the ICECAP-O plausibly was that, in the conceptualization of the scholars who developed the instrument, the inability to stay free from worry is an attribute of suboptimal QoL, but the reasons for worrying, such as concerns about one’s financial situation or fear of death or dying, are irrelevant. It is worth pointing out, however, that the results presented here show that worries about the ability to pay bills were significantly more detrimental to ratings of QoL in later adulthood than were worries about death and dying or about the security of one’s living environment. When determining the 11 dimensions for our models, we focused primarily on condensing the main elements of the four instruments considered, rather than on making a sharp distinction between constituents and antecedents. This choice was also made because this distinction can be somewhat arbitrary. This is illustrated by the fact that attachment, that is, having satisfactory social relationships, is considered a central attribute of QoL in the ICECAP-O (Coast et al. Reference Coast, Flynn, Natarajan, Sproston, Lewis, Louviere and Peters2008; Grewal et al. Reference Grewal, Lewis, Flynn, Brown, Bond and Coast2006), whereas the scholars who developed the CASP-19 consider it to be an antecedent, rather than a constituent, of QoL (Wiggins et al. Reference Wiggins, Higgs, Hyde and Blane2004).
By estimating age-stratified models, the current study also shed light on differences between age groups in the perceived constituents of QoL in later adulthood. By doing this, we acknowledged that QoL arguably is a dynamic construct (Allison et al. Reference Allison, Locker and Feine1997; Farquhar Reference Farquhar1995). Results showed some differences in the extent to which evaluated dimensions were considered to be constituents of QoL in later adulthood. For instance, respondents in the two oldest age groups associated sensory functioning more strongly with QoL in later adulthood than did their counterparts in the youngest age group. Possibly, this may reflect that people in the oldest age groups are more likely to have experienced declines in sensory abilities (Roth et al. Reference Roth, Hanebuth and Probst2011) and that they reconsidered the value they ascribed to sensory functioning in response to these lived experiences (compare with Whitley et al. Reference Whitley, Benzeval and Popham2020). On the other hand, the finding that self-realization contributed more to the QoL ratings among respondents aged 70+ than among their counterparts in the younger age groups may, in line with the mental incongruence mechanism described by Deeg (Reference Deeg, Mollenkopf and Walker2007), reflect that people may, as they age, increasingly value aspects like meaningful spending of time in response to ageing-related physical declines. Overall, however, the results of the three age-stratified models were remarkably similar. This strong degree of similarity in the coefficients for the various dimensions suggests a high level of consensus across age groups in what middle-aged and older people perceive as being important dimensions of QoL in later adulthood.
Several limitations of the current study should be considered. Although the LISS panel is based on a true population sample, selective non-response and panel attrition resulted in a sample with characteristics that differ somewhat from the population of the Netherlands on several relevant markers. Most notably, women and people with high educational attainment are overrepresented somewhat in the LISS panel (Klein Kranenburg and Zandvliet Reference Klein Kranenburg and Zandvliet2022). However, additional analyses showed few differences by gender and by level of education in the coefficient estimates for the 11 dimensions of interest (for results, see Appendix B and Appendix C in the online supplemental material). We thus expect that the impact of the selectivity of our sample on the results reported here is very limited.
It should also be considered that we presented interviews with two fictional women who both had a name that was very common in the 1950s and who were approximately the same age. The rationale is that the respondents perceived the two fictional interviewees as different persons, but that the only truly meaningful differences were the experimentally manipulated dimensions of our theoretical interest. The drawback of this choice is that our results may not correspond fully with what are considered important dimensions of QoL in later adulthood for men. An interesting avenue for future research may thus be to assess potential variation by gender of the fictional interviewee in the importance assigned to the QoL dimensions included in vignette studies such as ours. In factorial design vignette experiments such as those fielded for the current study, respondents are furthermore asked to evaluate fictitious scenarios under the assumption that their evaluations capture what would be the real-life implications of the attributes in the scenario (compare with Wallander Reference Wallander2009). This assumption may not hold when there is a large gap between the respondents’ actual experiences and the situations or characters they are asked to imagine (Hughes and Huby Reference Hughes and Huby2002). If respondents find it difficult to relate to or accurately assume the situations of vignette characters, their responses may not genuinely reflect their true attitudes. To minimize this risk, maximizing relatability and comprehensibility were among the main criteria for us when designing the experiment. Moreover, the experiment was pre-tested prior to being fielded.
A third limitation is that the process that led to the selection of the 11 dimensions included in the current study’s vignettes forced us to make choices that could be considered arbitrary. First, we restricted our focus to four frequently used QoL instruments developed specifically for older adults, and we did not consider measures of the related concept health-related quality of life (HRQoL). Measures of HRQoL tend to have a narrow focus and Karimi and Brazier (Reference Karimi and Brazier2016) argued that these measures often effectively capture little more than respondents’ self-rated health status. Our point of departure was that important constituents of QoL may well be situated in life domains other than health, and this premise corresponded better with the four QoL instruments considered here. Second, we had to collapse the dimensions of QoL considered into a smaller set of main dimensions. Maintaining each dimension of all four instruments considered was unfeasible; it would have led to overly lengthy vignette scenarios that increased the risk for respondent fatigue and consequently may have undermined the reliability of the findings (compare with Wallander Reference Wallander2009). As described in the section on data and methods, we distilled the main dimensions that the current study focused on in a joint iterative process. The fact that we found that all 11 dimensions were significantly associated with changes in the QoL ratings in the expected direction strengthens us in our view that we indeed identified key constituents of QoL in later adulthood. Nevertheless, we did not consult with a broader team of scholarly experts who might have provided valuable insights, and we do not claim to be comprehensive with regard to the dimensions covered. Van Leeuwen et al. (Reference Van Leeuwen, Van Loon, Van Nes, Bosmans, De Vet, Ket, Widdershoven and Ostelo2019), for instance, suggested that feeling attached to and experiencing faith and self-development from beliefs, rituals and inner reflection is an important aspect of QoL for specific subgroups of older people, and the full – but not the shortened – version of the OPQOL scale accordingly contains items about culture and religion (Bowling Reference Bowling2009; Bowling et al. Reference Bowling, Hankins, Windle, Bilotta and Grant2013). Scholars may therefore want to build on the current study and field experiments in which the importance of cultural and spiritual dimensions or other presumed constituents of QoL not considered here is assessed.
The estimated QoL contributions of the various dimensions included in our vignettes might moreover have been influenced by how well the operationalizations captured them, and by the perceived severity of the elements included in these operational definitions (compare with Whitley et al. Reference Whitley, Benzeval and Popham2020). For example, we operationalized a suboptimal position on the sensory functioning dimension as having hearing loss, rather than, for instance, blindness. The relatively small coefficient of the sensory function dimension may reflect the choice to operationalize it with this arguably less severe sensory limitation. It should be noted, however, that, as much as possible, we based our operationalizations on items included in the four scales of QoL in later adulthood considered in the background section. Furthermore, it is worth highlighting that our results are generally consistent with prior exploratory work on what laypersons consider most important for QoL in later adulthood or ageing well. Van Leeuwen et al.’s (Reference Van Leeuwen, Van Loon, Van Nes, Bosmans, De Vet, Ket, Widdershoven and Ostelo2019) review of the qualitative literature on older persons’ perspectives on the meaning of QoL suggested that what people considered to be aspects of QoL in later adulthood could be grouped under eight themes: (1) looking on the bright side of life; (2) feeling at peace; (3) feeling healthy and not limited by one’s physical condition; (4) being able to manage autonomously, retaining dignity and not feeling like a burden; (5) spending time doing activities that bring a sense of value, joy and involvement; (6) having close relationships that make one feel supported and enable one to mean something for others; (7) feeling secure at home and living in a pleasant and accessible neighbourhood; (8) not feeling restricted by one’s financial situation; and (9) the aforementioned theme about spirituality and inner reflection. With the exception of the last theme, these themes align well with the dimensions covered in the current study. The empirical evidence that we obtained from a large population-based sample and presented here substantially strengthens the case that these dimensions are indeed important constituents of QoL in later adulthood.
Although we presented the results of models stratified by age group, we did not test whether the contributions to QoL of the constituents we considered varied by other potentially relevant demographic markers. Importantly, our sample size did not permit us to estimate models separately for native Dutch people without a migration background and for specific migrant groups. This is unfortunate because the rapidly growing group of non-Western older migrants in the Netherlands (Conkova and Van den Broek Reference Conkova and Van den Broek2024) has different views on what ageing entails than their native Dutch counterparts (Nieboer et al. Reference Nieboer, Van den Broek and Cramm2021), and qualitative studies suggest that they may also have somewhat different priorities regarding constituents of QoL in later adulthood (Çayci and Van den Broek Reference Çayci and Van den Broek2022; Conkova and Lindenberg Reference Conkova and Lindenberg2020). Filling this gap in the literature calls for fielding survey experimental studies whereby people with a migration background are oversampled.
Despite these limitations, the insights gained from this study into what middle-aged and older people consider to be important dimensions of QoL in later adulthood are highly relevant, particularly against the backdrop of population ageing. By highlighting the aspects that older adults prioritize, the findings presented here can guide the creation and evaluation of policy interventions in ways that correspond closely with older persons’ preferences and needs. Our findings emphasize the importance of enjoyment and social participation for QoL in later adulthood. To align with older adults’ needs and preferences, interventions could prioritize fostering enjoyable and socially engaging activities. Prior research suggests that this could entail providing accessible volunteer opportunities and social leisure activities, to promote a sense of connection (Ten Bruggecate et al. Reference Ten Bruggecate, Luijkx and Sturm2018). Additionally, encouraging both in-person and digital social activities, such as exercise groups or online games, could further enhance engagement and participation (Devereux-Fitzgerald et al. Reference Devereux-Fitzgerald, Powell, Dewhurst and French2016; Michèle et al. Reference Michèle, Guillaume, Alain, Nathalie, Claude and Kamel2019). By focusing on older adults prioritizing enjoyment and social participation, future policies and interventions can be better tailored to older adults’ preferences, and accordingly improve their overall QoL.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0144686X25100135.
Author contributions
TvdB and MH were jointly responsible for the conceptualization and writing of the original draft. TvdB was also responsible for the data curation, methodology, formal analysis and visualization.
Financial support
This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 895537 (SAMBa).
Competing interests
The authors declare that they do not have competing interests.
Ethical standards
This study was assessed and approved by the Research Ethics Review Committee of the Erasmus School of Health Policy and Management at Erasmus University Rotterdam (reference: ETH2324-0130).
Open access
