Skip to main content
NCBI home page
European Journal of Ageing logoLink to European Journal of Ageing
. 2022 Sep 12;19(4):1429–1440. doi: 10.1007/s10433-022-00728-9

A longitudinal study of the effects of well-being and perceived control on preparations for old age: moderation effects of contexts

Jeongsoo Park 1,✉, Thomas M Hess 2, Helene H Fung 3, Anna Kornadt 4, Klaus Rothermund 5
PMCID: PMC9729629  PMID: 36506654

Abstract

Extending research on determinants of preparations for old age across adulthood, we examined the relationship between well-being, perceived control, and preparations for old age over time, along with variation in the strength of these relationships depending on domains of functioning, cultures, and age. We analyzed longitudinal data from the Ageing as Future study assessing changes in well-being, perceived control, and preparations in four different life domains (social relations, finances, work, and health) across a five-year period collected from adults aged from 30 to 85 years in Germany (N = 623), Hong Kong (N = 317), and the USA (N = 315). Positive feelings about one’s current situation predicted greater perceived control five years later, and vice versa. Also, perceived control and preparations were positively associated over time, with only a few exceptions within each domain. For example, high control beliefs were related to subsequent greater preparations and well-being within the domain of social relations and finances, with weak effects of culture. These results suggest that current well-being may promote adaptive behaviors in later life, with the accumulated effects over time. Thus, focusing on how to improve our well-being may promote engagement in preparations for old age. Our results also indicate that these relations in our model vary by contexts, highlighting the importance of variability in age-related processes.

Keywords: Control beliefs, Well-being, Preparations for aging, Culture, Context

Introduction

By 2050, one in six people in the world and one in four people in Europe and Northern America are expected to be over the age of 65 (United Nations 2019). Along with this increase in the aging population, the assumption of a standard retirement age of 65 (OECD 2017) suggests that more individuals will experience an extended period of life after retirement than did previous generations. Given that entering old age is a major life transition involving challenges as well as changes (Wang and Schultz 2010), being prepared for old age is an important developmental goal for most of us. Although societal or familial systems provide support for older adults, they may not be sufficient to meet individuals’ needs, leaving responsibilities for making one’s own provisions for old age. As might be expected, preparing for anticipated changes and needs associated with growing older does indeed have benefits. For example, preparations for old age have been shown to be beneficial to physical health (Topa et al. 2009) and psychological well-being (Earl et al. 2015; Sörensen et al. 2012) in later life. In addition, the challenges associated with preparing for old age may vary across social and cultural contexts based upon variation in the availability of support structures. Even after controlling for socioeconomic and demographic variables, however, the positive effects of preparations for old age on well-being are still observed (Noone et al. 2013; Reitzes and Mutran 2004). Thus, investigation of what personal and psychological factors influence individuals’ preparation for old age is especially important for understanding adjustment in later life. In the present study, we adopt a contextual perspective (e.g., Baltes 1997) to examine preparations for old age. Specifically, we examine control beliefs as a mechanism promoting preparations and well-being across a five-year period along with variations in their influence across contexts (i.e., domains of functioning within cultures) that may differ in the degree of necessary or desired preparations.

Control, well-being, and preparations for old age

A primary assumption associated with preparations for old age is that they have a positive impact on adjustment and well-being, which existing research supports (e.g., Earl et al. 2015; Noone et al. 2013; Reitzes and Mutran 2004; Sörensen et al. 2012; Yeung and Zhou 2017). Given this association, there is much interest in understanding the determinants of preparation. In the present study, we focus on perceived control as a possible determinant, given that perceived control has important implications for adaptive behaviors and functioning in later life. Perceived control refers to beliefs about one’s capacity to bring about a given outcome (Lachman and Weaver 1998; Skinner 1995), and research with older adults has shown that those individuals with high control beliefs are more likely to engage in beneficial behaviors in a wide variety of domains (see Lachman et al. 2011; Robinson and Lachman 2017). In general, greater perceptions of control may facilitate adaptive functioning by increasing the likelihood of pursuing strategies that promote positive developmental outcomes (Robinson and Lachman 2017). Thus, it is not unreasonable to assume that individual’s beliefs in their ability to control age-related change will be positively related to preparations and associated with positive well-being. That is, people who believe that they can control outcomes associated with aging will be more likely to engage in preparations than will people who believe control is less possible. We had previously demonstrated such a relationship using data from the Ageing as Future (AAF) project (Park et al. 2020), but establishment of the directionality of this relationship was limited by the use of cross-sectional data. The present study expands on our previous work by examining cross-lagged relationships using AAF data collected over a 5-year period.

The positive effects of preparations for old age are typically observed in terms of better psychological adjustment, including higher levels of life satisfaction (Earl et al. 2015; Noone et al. 2013), positive attitudes toward retirement (Reitzes and Mutran 2004), and mental health (Sörensen et al. 2012). Control has also been shown to predict well-being in a variety of domains of functioning (e.g., Chhatwani 2022; Gerstorf et al. 2014; Henning et al. 2022). Assuming a positive relationship between control and preparations, psychological well-being might then be considered a distal outcome of control beliefs operating through preparations. That is, preparations may partially mediate the relationship between control and well-being. However, well-being may itself have a positive impact on control beliefs and preparations. Specifically, feeling good about oneself has been shown to serve as a resource that facilitates behaviors promoting long-term goals (e.g., Trope et al. 2006), enhancing future positive affect and well-being (Fredrickson 2004). For example, older adults with positive self-perceptions regarding their functioning are more likely to attend to negative, but important information relating to promoting future outcomes (e.g., Growney and Hess 2019a,b), and are more likely to engage in exercising (Gallant and Dorn 2001) and preventive health care activities (Bennett et al. 2009).

In addition, relationships involving perceived control and preparations may also be bidirectional given that more preparation behaviors before retirement lead to greater personal resources such as perceived control, which in turn contributes to positive changes in well-being (Yeung and Zhou 2017). Similarly, better initial appraisals in the domains of health and social support have been shown to be associated with subsequent increases in perceived control (Gerstorf et al. 2011), which could also have a downstream effect on preparations. Taken together, these findings suggest a recursive relationship between control, preparations, and well-being. Specifically, control has a positive impact on both preparations and well-being, with well-being in turn having a positive impact on control and preparations. The present study focuses on examining the interrelationships between these three variables over time.

Multidimensional preparations for old age and age-related changes

In the context of lifespan development, although individuals continually adjust their goals and proactively respond to different challenges with advancing age (Baltes 1997), most research on preparations for old age has focused on financial planning for retirement in samples of middle-aged/older workers or those who have just retired (e.g., Moffatt and Heaven 2017; Topa and Herrador-Alcaide 2016; van Rooij et al. 2012). However, alongside finances, individuals experience changes with advancing age in many different domains, including their social network, health, and work status. As preparation for old age is a life-long process, adults of any age may be engaged in a variety of preparation behaviors (Kim-Knauss and Lang 2020; Kornadt and Rothermund 2014). Age-related differences in planning may be associated with the acquisition of knowledge that may be motivated by stage of life and degree of focus on the future. Although a few studies have examined preparations beyond finances (Eismann et al. 2019; Petkoska and Earl 2009), they have only focused on a limited number of domains (e.g., bridge employment, leisure) within a specific group of individuals (i.e., older workers). Kornadt and Rothermund (2014) dealt with these limitations by investigating preparations for old age in different life domains across adulthood. They found that preparations varied in intensity depending on both the domain of functioning and participants’ age. These results highlight the importance of considering preparations for old age as a multidimensional construct dependent upon context.

Similar to preparations for old age, perceived control has been identified as a multidimensional construct (Lachman et al. 2011; Lachman and Weaver 1998). Indeed, Lachman and Weaver (1998) found that domain-specific control beliefs differed by age in their sample of adults aged 25 to 75. For instance, greater control over work, finances, and marriage was found with advancing age, whereas decreasing control was found for the domains of relationship with children and sex life. That is, an individual may perceive high control in some domains, but not in other domains simultaneously, with this domain-specificity varying across ages. Likewise, well-being has also been shown to be multidimensional, exhibiting variability across individuals and domains of functioning (e.g., Kornadt and Rothermund 2011). Taking into account the context-specificity of preparations, perceived control, and well-being, the present study examines the relationship of these three factors within four different domains of functioning. This approach allows us to match the three constructs in the respective life domains, which should increase predictive validity.

Culture as context of preparations

Bronfenbrenner’s Ecological System Theory (1994) proposes that an individual exists within a set of nested contextual systems, and that the interaction between the individual and environment occurs over time. That is, preparations do not occur in a vacuum, but are influenced not only by psychological factors at the individual level and by life context, but also by factors at cultural and societal levels. For example, preparations may be less of an issue for individuals in a country where a broad structure of support systems exists when compared to those in countries with weaker support. Supporting this assumption, Kornadt et al. (2019) examined preparation in three countries (i.e., Germany, Hong Kong, and the USA) and found cultural differences in preparations for old age for each domain. For example, Americans reported more preparations in the domain of finances and work than did Germans, which was expected given strong government-supported programs in Germany. However, against expectations, the lowest level of preparation was found in people from Hong Kong, which might be based in other factors, such as high expectations from family support systems. Above all, the interaction between different institutional structures and domains of functioning (i.e., exosystems; Bronfenbrenner 1994) may lead to different levels of individuals’ preparations in each country.

Given these contextual effects, the relationship between well-being, perceived control, and preparations for old age at the individual level should also be buffered by the extent to which social or structural supports exist within specific life domains within each culture. Presumably, a culture where weak support systems exist within a specific domain may increase the importance of person-based attributes (e.g., perceived control), as evidenced by greater associations between these factors and preparations for old age. For instance, strong government pension programs and a mandatory retirement age in Germany may be expected to lessen the level of, or need for personal control over financial preparations, whereas less sufficient social security in the USA than in Germany may increase the necessity of preparations along with the positive impact of perceived control on those preparations. Also, less certainty regarding government-supported health care as well as higher medical expenses in the USA than in Germany or Hong Kong may increase the need for personal preparations for health in later life. Perceptions of old age in different cultures also vary in terms of attitudes towards older adults needing care-giving. For example, rapidly aging and industrializing societies in East Asia may show negative attitudes toward older adults, such as seeing them as a burden (North and Fiske 2015). Within the domain of social relations, cohesive family relations along with a tradition of filial piety in Hong Kong may lessen the importance of other social relations (Adams and Kurtis 2015; Cheng and Chan 2006; Lee and Hong-kin 2005). On the other hand, western cultures (i.e., the USA and Germany) may have more emphasis on having new social relationships due to geographic dispersion of family members (Yuan and Wang 2011). That is, preparations are influenced by cultural differences in social values, traditions, and support system.

The current study

In the present study, we investigated the relationship between perceived control, preparations for old age, and well-being in each of four domains of functioning (social relations, finances, work, and health) across three different cultures (Germany, Hong Kong, and the USA). We used longitudinal data collected across a five-year period from adults initially aged from 30 to 85 years. Based on the previously reviewed literature, we were interested in the impact of control beliefs on both preparations and well-being, as well as the impact of well-being on control. More than three waves of data collection would be ideal to examine the entire chain in a series of longitudinal effects (e.g., perceived control → preparations → well-being → perceived control). However, as our data were only collected at two-time points, we examined directional relationships between one construct on the other two constructs over time, while controlling for stabilities and cross-sectional associations of these constructs (Little et al. 2007; Vötter and Schnell 2019).

We predicted that greater perceived control at T1 would predict greater levels of preparations and well-being at T2. We also predicted that higher levels of well-being and preparations at T1 would serve as a resource leading to greater levels of control at T2. We further hypothesized that the strength of all these effects would be moderated by age, domains of functioning, and culture. First of all, the strengths of relationships involving preparations (i.e., perceived control and preparations) were expected to increase with advancing age as preparation for old age becomes more self-relevant. We also hypothesized that there would be variation in the strength of these relationships depending on cultures and domains. In contrast, the relationships between well-being and perceived control were expected to be positive, with fewer variations, assuming that well-being may lead to an enhanced sense of control regardless of age or culture.

Within the domain of social relations, we expected variations in effects between three cultures (Germany, Hong Kong, and the USA), with these variations primarily reflecting differences in social and institutional structures as opposed to more global cultural differences. Older adults in the USA may feel a stronger need for social relations in later life due to geographic dispersion of family members. Likewise, for older Germans, lower home-ownership compared to the USA (e.g., Kaas et al. 2017) and good financial support for retiree communities or apartments by the German government may facilitate moving into those places in later life, with increasing need to build new social relations. In contrast, the tradition of familism in Hong Kong may lead older adults to engage in narrowed relationships involving family members or relatives, with less need to create new social relations in later life (Adams and Kurtiş, 2015; Fung et al. 2008). For the domains of finances and work, we hypothesized variations between cultures depending on whether a well-provided pension system and mandatory retirement age exist. Thus, due to mandatory retirement ages and stronger state-supported pension systems, the need for personal preparations for finances and work with advancing age in Germany was expected to be less than that in the USA and, to a lesser extent, in Hong Kong, leading to the prediction that the longitudinal pathways involving control would be the strongest in the USA. In the health domain, the effects were expected to be the strongest in the USA, where people may feel a strong need to engage in preparations due to higher medical expenses and lower government-provided health care systems compared to the other two cultures.

Method

This project has been reviewed and approved by the Institutional Review Boards at the Chinese University of Hong Kong, Friedrich-Schiller-University, and North Carolina State University.

Participants

As part of the Ageing as Future study, participants were recruited from: (a) Jena and Erlangen, Germany; (b) Hong Kong, China; and (c) Wake County, North Carolina, USA. Randomly selected people for targeted age and gender distributions were identified through private marketing companies (China and USA) or local registry offices (Germany) at Time 1 (T1, 2013). Upon consenting, participants were mailed a questionnaire packet and received a gift card worth approximately $25 for reimbursement. Participants at T1 were contacted again and asked to participate in our follow-up study at Time 2 (T2, 2018). We followed the same procedure as T1, and 77% (Germany), 65% (Hong Kong), and 62% (USA) of participants at T1 completed our study at T2. Drop-out with regard to central variables (e.g., age, gender, income; Voss et al. 2017) indicated that the US participants dropping out at T2 were significantly younger (M = 54.71, SD = 16.90) than those who participated at both T1 and T2 (M = 57.63, SD = 14.26), t(570) = -2.24, p = 0.025. Participants in Hong Kong and Germany did not show any systematic dropout.

Information about the final sample (N = 1255) is listed in Table 1. For further group comparison analyses by age, we divided the sample into three age groups of relatively comparable sizes based on ages at T1: younger (aged 30–49), middle-aged (aged 50–65), and older (aged 66–85). These age groups correspond reasonably well to periods of adulthood associated with (a) the early stages of work and family life, (b) established adulthood, and (c) the age at which old age is often associated. Subjective health was assessed with a single item (“How would you describe your current state of health?”) on a five-point scale (1 = not good at all, 5 = very good). Mean ratings of self-rated health at T1 were higher in the USA than in the other two cultures, F(2,1244) = 60.75, p < 0.001, ƞ2partial = 0.09. Moreover, significant age group differences in subjective health were found in Germany, F(2,619) = 20.37, p < 0.001, ƞ2partial = 0.06, such that mean health rating was significantly higher in younger adults than in middle-aged adults, and both were significantly higher than in older adults (p < 0.001). Household income was assessed on an eight-point scale with adjusted currency for each country. Higher income levels were observed in the USA than in Germany, and the income levels of Germany and the USA were significantly higher than that in Hong Kong, F(2, 1208) = 47.24, p < 0.001, ƞ2partial = 0.17. Significant age group differences in household income were found in Germany, F(2, 617) = 25.19, p < 0.001, ƞ2partial = 0.08, such that household income was significantly higher in young adults than in middle-aged adults, both of which were significantly higher than in older adults (ps < 0.001). Age differences in household income were also found in the Hong Kong subsample, F(2, 284) = 28.66, p < 0.001, ƞ2partial = 0.20, with significantly higher income in younger adults than in middle-aged adults, and both were higher than in older adults (ps < 0.001). Means for all comparisons are presented in Table 1.

Table 1.

Demographic characteristics at Time 1, in 2013

Culture Age group
(Total N, % female)		 Age
M (SD)		 Subjective Health
M (SD)		 Household Income
M (SD)
US Younger (95, 60%) 39.94 (5.10) 3.25 (0.72) 6.09 (1.31)
Middle-aged (110, 56.4%) 57.82 (4.66) 3.23 (0.79) 5.92 (1.75)
Older (108, 47.2%) 72.99 (5.64) 3.18 (0.86) 5.66 (1.61)
All ages (315, 54.3%) 57.63 (14.26) 3.22 (0.79) 5.88 (1.58)
Germany Younger (205, 49.0%) 42.14 (4.63) 2.91 (0.93) 5.44 (1.32)
Middle-aged (209, 57.3%) 57.90 (4.37) 2.64 (0.90) 5.11 (1.40)
Older (209, 52.0%) 73.71 (5.06) 2.33 (0.95) 4.52 (1.28)
All ages (623, 50.7%) 58.02 (13.71) 2.62 (0.95) 5.02 (1.39)
Hong Kong Younger (100, 48.8%) 43.07 (4.21) 2.58 (0.71) 5.58 (1.16)
Middle-aged (117, 53.1%) 57.62 (4.92) 2.58 (0.62) 4.87 (1.89)
Older (100, 50.2%) 72.73 (5.21) 2.53 (0.73) 3.59 (2.12)
All ages (317, 53.0%) 57.80 (12.73) 2.56 (0.68) 4.67 (1.96)
Open in a new tab

Measures

Preparations for old age

Preparations for age-related changes at both time points were assessed with a questionnaire developed by Kornadt and Rothermund (2014). Although this measure includes nine different domains of preparations, we focused on four domains that matched domains for the other measures of interest in the present study: social relations, finances, work, and health. For each domain, respondents rated three items on a four-point scale (1 = not at all; 4 = a lot). The first item assessed active preparations: Social relations, “I am actively working to maintain my personal relations in old age (e.g., by fostering contacts, being included in social groups, etc.); Finances, “I am actively providing for my financial situation in old age (e.g., retirement accounts, savings, etc.); Work, “I am actively preparing for successful professional activity in old age (e.g., through continuing education, avoidance of job-related illness and disability, etc.); Health, “ I am actively working to maintain my health in old age (e.g., by regular checkups, avoidance of behavior that is harmful to my health, etc.). The other two items assessed thinking about the topic (e.g., “I think about the topic”) and gathering information (e.g., “I try to gather information about the subject and discuss it with others”). These two items were worded the same for all domains.

Well-being

Well-being was assessed within each of the four target domains by three domain-specific items that measured participants’ perceptions of their current level of functioning. Each item focused on a specific aspect of functioning related to each domain that contrasted two opposing statements (e.g., “I have a negative attitude towards retirement” versus “I have a positive attitude towards retirement” in the work domain). Participants responded to each item on an eight-point scale, with higher scores indicating more positive perceptions of their own current level of functioning. The complete questionnaire can be found in Hess et al. (2017).

Perceived control

Participants rated a single item asking about perceived control in each of the four life domains (e.g., “Regarding my personal situation in the domain of social relations”) on a five-point scale (1 = “I have no control at all”; 5 = “I have a lot of control”).

Statistical analysis

Statistical analyses were implemented using SPSS 26.0 (IBM Corp 2018) and AMOS 26.0 (Arbuckle 2019). To establish measurement invariance, configural invariance and metric invariance across cultures and age groups over time were tested. Corresponding factor loadings were constrained to be equal across time, age groups, and cultures. Our main model testing our primary predictions involving control, preparations, and well-being (Fig. 1) included stability paths, initial correlations, and change correlations when estimating prospective effects, controlling for all bivariate effects. We initially tested this model using scores collapsed across domains for the entire sample. We next explored contextual effects by examining this model within each domain, and then testing for culture and age effects by introducing model constraints across cultures and age groups, using χ2 difference tests to compare competing models.

Fig. 1.

Fig. 1

Open in a new tab

Statistical model depicting the proposed variable relations over time

Based on previous findings with these data (e.g., Kornadt et al. 2019), we also allowed correlated errors between two items (i.e., “I think about the topic”, “I try to gather information about the subject and discuss it with others”) on the preparations scale due to common item stems. With respect to control variables, gender, income, retirement status as a dichotomous variable, and self-rated health as a continuous variable were included as common covariates across cultures when examined in the domains of social relations, finances, work, and health, respectively, due to relevance to specific domains.

Results

Initial model tests

In evaluating each model, we used the following guidelines as measures of good fit (Awang 2012): CFI > 0.90, and RMSEA < 0.05 (< 0.08 is considered an acceptable fit). In general, all estimated models yielded good to acceptable model fits (Table 2).

Table 2.

Results of Relations between Well-being, Perceived Control, and Preparations for Old Age for Domain-General and Domain-Specific Measures

Model Initial correlation β Change correlation β Autoregressive path β Cross-lagged path β χ2 df p RMSEA [90%CI] CFI
General CON1 ↔ WB1 .752 CON2 ↔ WB2 .631 CON1 → CON2 .431 CON1 → WB2 .275 5444.557 1411 .000 .048 .909
CON1 ↔ PREP1 .227 CON2 ↔ PREP2 .292 WB1 → WB2 .519 CON1 → PREP2 .118 [.046, .049]
WB1 ↔ PREP1 .222 WB2 ↔ PREP2 .232 PREP1 → PREP2 .648 WB1 → CON2 .324
PREP1 → CON2 .041
Social Relations CON1 ↔ WB1 .381 CON2 ↔ WB2 .215 CON1 → CON2 .426 CON1 → WB2 .064 292.886 66 .000 .052 .978
CON1 ↔ PREP1 .276 CON2 ↔ PREP2 .222 WB1 → WB2 .632 CON1 → PREP2 .136 [.046, .059]
WB1 ↔ PREP1 .447 WB2 ↔ PREP2 .186 PREP1 → PREP2 .498 WB1 → CON2 .059
PREP1 → CON2 .062
Finances CON1 ↔ WB1 .401 CON2 ↔ WB2 .304 CON1 → CON2 .411 CON1 → WB2 .068 343.316 66 .000 .058 .967
CON1 ↔ PREP1 .331 CON2 ↔ PREP2 .211 WB1 → WB2 .603 CON1 → PREP2 .116 [.052, .064]
WB1 ↔ PREP1 .389 WB2 ↔ PREP2 .183 PREP1 → PREP2 .593 WB1 → CON2 .008
PREP1 → CON2 .087
Work CON1 ↔ WB1 .294 CON2 ↔ WB2 .212 CON1 → CON2 .342 CON1 → WB2 .121 248.507 66 .000 .047 .982
CON1 ↔ PREP1 .258 CON2 ↔ PREP2 .174 WB1 → WB2 .435 CON1 → PREP2 .041 [.041, .053]
WB1 ↔ PREP1 .208 WB2 ↔ PREP2 .131 PREP1 → PREP2 .422 WB1 → CON2 .103
PREP1 → CON2 .039
Health CON1 ↔ WB1 .465 CON2 ↔ WB2 .247 CON1 → CON2 .300 CON1 → WB2 .036 221.056 66 .000 .043 .985
CON1 ↔ PREP1 .185 CON2 ↔ PREP2 .133 WB1 → WB2 .357 CON1 → PREP2 .019 [.037, .050]
WB1 ↔ PREP1 .064 WB2 ↔ PREP2 .092 PREP1 → PREP2 .642 WB1 → CON2 .074
PREP1 → CON2 .061
Open in a new tab

Values in bold are significant at p < .05

General model

Using measures collapsed across domains, significant initial correlations and change correlations between well-being, perceived control, and preparations for old age were observed. Considerable stabilities in well-being, perceived control, and preparations were also observed. To test our main hypotheses, we were especially interested in the cross-lagged paths and, consistent with expectations, most cross-lagged paths were positive. Specifically, control at T1 was a significant predictor of both preparations and well-being at T2, and well-being at T1 predicted perceived control at T2. However, inconsistent with predictions, preparation at T1 did not predict T2 control; in other words, the prediction that preparations would serve as a resource for enhancing control was not supported.

Domain-specific models

We next used domain-specific indices of control, preparations, and well-being to investigate the degree to which the aforementioned relationships were modified by domain of functioning. The models within each domain of functioning yielded good to acceptable model fits (See Table 2). Moreover, significant initial correlations and change correlations between well-being, perceived control, and preparations for old age, along with considerable stabilities were found in all domains. Similar to the general model, cross-lagged effects were also significant with a few exceptions. In general, evidence for the positive impact of resources on the subsequent sense of control was observed in all domains, with the strength of the specific resource influence (i.e., well-being or preparations) varying somewhat over domains. In contrast, significant positive effects of control on well-being or preparations were only observed in the domains of social relations, finances, and work.

Culture and age effects with each domain

We next examined the degree to which the relationships observed within each domain were modified by culture or age.

Cultural differences

As our primary interest was to investigate cultural differences in cross-lagged effects within each domain, we examined variation by constraining cross-lagged pathways to be equal across cultures. To do so, we used Germany as the basis for comparison (i.e., cross-lagged effects in the Hong Kong and USA subsamples were constrained to those obtained in the German subsample) given the larger sample size of this subsample and assumed associated greater stability of estimates. A significant χ2 change between the constrained and unconstrained model was only found in the domain of work: ∆χ2 (8) = 22.66, p < 0.01. Adding constraints did not significantly alter the fit in the other domains (ps > 0.25), suggesting little variation across cultures: Social relations, ∆χ2 (8) = 6.29; Finances, ∆χ2 (8) = 7.83; and Health, ∆χ2 (8) = 10.07. We next investigated multi-group comparisons of cross-lagged paths between cultures within the domain of work through pairwise comparisons. Each path was constrained one-by-one in the reference group. Results did not indicate significant differences (∆χ2 = −0.84 ~ 3.30 with df = 2 < 5.99), suggesting relatively minor variations across cultures. Thus, contrary to our expectations, culture did not significantly affect the strength of the observed cross-lagged effects.

Age group differences

Given that cultural differences were minimal, we investigated age effects within each domain collapsing across cultures. Using the younger group (aged 30–49) as the reference group, a significant χ2 change between the unconstrained and constrained model was only found in the domain of social relations, ∆χ2 (8) = 27.77, p < 0.01. Constraining the cross-lagged pathways to be equal across age groups did not significantly alter model fit in the domain of Finances, ∆χ2 (8) = 11.29, p > 0.15, Work, ∆χ2 (8) = 13.83, p > 0.09, and Health, ∆χ2 (8) = 3.15, p > 0.09. To investigate age effects within the domain of social relations, pairwise comparisons were performed between age groups. No significant differences in cross-lagged paths between age groups were found (∆χ2 = −0.397 ~ 4.546 with df = 2 < 5.991), providing no support for our hypothesis that the strength of cross-lagged effects would increase with age.

Discussion

This study investigated the relationships between well-being, perceived control, and preparations for old age over a time span of five years, examining cross-lagged effects between these three constructs at two-time points. Our initial examination of models with measures collapsed across age, culture, and domains of functioning revealed results that were generally consistent with our predictions. Specifically, we found that high levels of control at T1 predicted greater levels of both preparations and well-being at T2. In line with the resource perspective, we also found that greater well-being at T1 predicted greater levels of control at T2; however, predictive relations between preparation and control were not significant. The relations between control, preparations, and well-being are similar to effects obtained in previous research using cross-sectional data (e.g., Park et al. 2020; Prenda and Lachman 2001). However, the present longitudinal data extend this work by suggesting directional pathways, while also highlighting the reciprocal nature of the relationship between control and well-being. This finding is consistent with the expectation that well-being might serve as a resource for bolstering future control beliefs and preparations, a relationship that is suggested by previous work (e.g., Growney and Hess 2019b). The absence of a stronger association between preparations at T1 and control at T2 might be suggestive of an indirect relationship between these two factors, with well-being perhaps serving as a mediator. Unfortunately, the absence of three time points in the present data set did not allow us to test this specific hypothesis.

Moderating effects of domains, culture, and age

We also examined the degree to which these relationships varied by domains of functioning (i.e., social relations, finances, work, and health) and culture (i.e., the USA, Hong Kong, and Germany). Across all domains, all initial and change correlations between our three measures were significant. Of greater interest, however, were the cross-lagged associations between these variables.

Domain effects

Within-domain analyses revealed some differences in the strength of the cross-lagged effects as well as in the fit of the most parsimonious model. Within the domains of social relations, finances, and work, evidence for reciprocal cross-lagged relationships was observed. High initial control beliefs were associated with subsequent greater ratings of well-being and preparations, although the association involving this latter variable was not significant in the work domain. The positive impact of control on well-being and preparations was in line with previous cross-sectional analyses (e.g., Park et al. 2020; Prenda and Lachman 2001), but the observation that this cross-lagged relationship involving preparations was not evident in the domains of work and health is interesting. The reasons for this domain-specificity are unclear at present, but may relate to specific aspects of those domains. For example, the often-scripted nature of work (e.g., mandatory or cultural expectations regarding retirement) and uncertainty of many health outcomes (e.g., cancer, injury) may limit the importance of preparations relative to the domains of finances and social relations.

Control was also positively influenced by preparations or well-being in all four domains, a finding in line with our expectation that these factors would serve as a resource that enhanced future feelings of control. Notably, the associations between these resources and control were reciprocal in nature in all domains except health—although it should be noted that, taken together, the cross-lagged associations in this domain were somewhat weaker than those in the other three. These reciprocal effects are important in highlighting possible synergistic effects, with control promoting preparations and well-being, which in turn serve to build up a positive sense of control. The variation across domains in the strength of these effects, however, emphasizes the importance of context and cautions against making overgeneralizations about these relationships.

Culture and age effects within each domain

We investigated the extent to which the strength of the associations in specific domains varied based on variations in cultural views and structural supports. However, contrary to previous research supporting contextual specificity and multidimensionality of aging-related processes (Kornadt and Rothermind 2011; Kornadt et al. 2019), culture and age were unrelated to systematic variations of differences within each domain in our study. The relatively weak effect of culture and age is suggestive of stable relationships across cultures and adulthood. The absence of culture effects is particularly surprising, given variations across the three cultures relating to factors such as family relationships and health care. One possibility is that the absence of effects relates to the somewhat smaller sample sizes in the Hong Kong and US subsamples when compared to that of Germany, leading to the possibility of greater measurement error associated with the former subsamples.

Limitations

There are certain caveats regarding this research. First, although we investigated the relationships between well-being, perceived control, and preparations within a longitudinal design, other potential variables not included in the current study might also be influential in determining preparations for old age. For example, individuals who have a more concrete future time perspective may be engaged in preparations because they regard their future as providing some opportunities to pursue goals. In fact, Kornadt et al. (2019) found positive relationships between concreteness of time perspective and preparations. We chose, however, to focus on control beliefs given previous findings illustrating the important connection between control and both well-being and activities promoting well-being. In addition, although a measure to assess well-being would be ideal to include two dimensions evenly-eudaimonic feeling and optimal functioning suggested by Vittersø’s (2013), our measure tapped more into optimal functioning rather than eudaimonic feeling. Also, our study only compared three cultures, speculating about sources of effects based on knowledge of general differences in support structures between these cultures. Further attempts to replicate our study in more diverse cultures by measuring availability of resources—perceived or factual—may identify more systematic differences, potentially further highlighting the factors underlying cultural variation. In addition, with respect to age effects, our study divided participants into three age groups (i.e., younger adults 30–49, middle-aged adults 50–65, and older adults 66–85) corresponding to earlier work using the Ageing as Future study (e.g., Hess et al 2017) as well as each period adulthood. Thus, results might have confounding effects of cohort differences. Finally, only two waves of data were available, whereas three or four waves would have been more desirable in investigating a series of longitudinal effects of three constructs. For example, given some consistency in the pattern of cross-lagged effects, our models may predict that perceived control is a positive factor for future well-being, and vice versa. Also, well-being and preparations may also be resources perpetuating a heightened sense of control that may lead to continued engagement in preparations. Although methods for statistically exploring relationships between three variables over time, including mediation, have been proposed in the literature (e.g., Newsom 2015), they require strong assumptions regarding temporal sequences of action (e.g., control at T1 affects preparations at T2, which then have an immediate effect on well-being at T2). We were unwilling to make these assumptions and thus chose a more conservative approach.

Conclusion

The goal of the present research was to further understand the bidirectional relationships between well-being, perceived control, and preparations for old age across cultures and domains of functioning in adulthood. Our main finding is that positive associations between three constructs suggest that well-being and preparations also serve as resources for promoting control in later life, with the benefits leading to future preparations and well-being. Of course, this also implies the possibility of lower well-being or preparations having a maladaptive impact on perceived control. Notably, the bidirectional influences between constructs suggest many avenues for application. For example, interventions focused on improving well-being may be beneficial in promoting future preparations for old age and well-being by building psychological resources that may enhance feelings of control. Similarly, providing support for domain-specific preparations should result in enhanced well-being that in turn promotes positive control beliefs. Notably, the relatively weak effects of age and culture in our study suggest that any interventions designed to enhance preparations should be beneficial regardless of age and culture—at least with respect to the three cultures investigated. Last but not least, the finding that longitudinal effects seem to be effective across a time span of 5 years also implies possible prolonged effects of any interventional approach, increasing synergetic effects between control, preparations, and well-being. Along with stabilities across cultures and ages, the varied effects across domains of functioning highlight the variability in the experience of aging and the factors that influence adaptive functioning.

Funding

This work was supported by a grant of the VolkswagenStiftung (Az. 93 272) to Klaus Rothermund and Hong Kong Research Grants Council General Research Fund #14604220. A preliminary report based on this manuscript was presented at the annual meeting of the Gerontological Society of America, 2021.

Declarations

Conflict of interest

We have no known conflicts of interest.

Footnotes

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  1. Adams G, Kurtiş T. Friendship and gender in cultural-psychological perspective: Implications for research, practice, and consultation. Int Perspect Psychol Res Pract Consult. 2015;4(3):182–194. doi: 10.1037/ipp0000036. [DOI] [Google Scholar]
  2. Arbuckle JL (2019) Amos 26.0 User's Guide. Chicago: IBM SPSS.
  3. Awang ZH. A handbook on SEM: structural equation modeling. 4. Kualalumpur: Centre For Graduate Studies, University Teknologi MARA Kelantan; 2012. [Google Scholar]
  4. Baltes PB. On the incomplete architecture of human ontogeny: Selection, optimization, and compensation as foundation of developmental theory. Am Psychol. 1997;52(4):366–380. doi: 10.1037/0003-066X.52.4.366. [DOI] [PubMed] [Google Scholar]
  5. Bennett IM, Chen J, Soroui JS, White S. The contribution of health literacy to disparities in self-rated health status and preventive health behaviors in older adults. Annals Family Med. 2009;7(3):204–211. doi: 10.1370/afm.940. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bronfenbrenner U. Ecological models of human development. Int Encycl Educ. 1994;3:1643–1647. [Google Scholar]
  7. Chhatwani M. Personal control and financial well-being among the elderly: moderating role of the big five. Personal Individ Differ. 2022 doi: 10.1016/j.paid.2021.111171. [DOI] [Google Scholar]
  8. Cheng ST, Chan AC. Filial piety and psychological well-being in well older Chinese. J Gerontol Series B Psychol Sci Soc Sci. 2006;61:262–269. doi: 10.1093/geronb/61.5.p262. [DOI] [PubMed] [Google Scholar]
  9. Earl JK, Bednall TC, Muratore AM. A matter of time: Why some people plan for retirement and others do not. Work, Aging Retire. 2015;1(2):181–189. doi: 10.1093/workar/wau005. [DOI] [Google Scholar]
  10. Eismann M, Verbeij T, Henkens K. Older workers’ plans for activities in retirement: the role of opportunities, spousal support, and time perception. Psychol Aging. 2019;34(5):738–749. doi: 10.1037/pag0000377. [DOI] [PubMed] [Google Scholar]
  11. Fredrickson BL. The broaden-and-build theory of positive emotions. Philos Trans Royal Soc B: Biol Sci. 2004;359:1367–1377. doi: 10.1098/rstb.2004.1512. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Fung HH, Stoeber FS, Yeung DY, Lang FR. Cultural specificity of socioemotional selectivity: age differences in social network composition among Germans and Hong Kong Chinese. J Gerontol B Psychol Sci Soc Sci. 2008;63B(3):P156–P164. doi: 10.1093/GERONB/63.3.P156. [DOI] [PubMed] [Google Scholar]
  13. Gallant MP, Dorn GP. Gender and race differences in the predictors of daily health practices among older adults. Health Educ Res. 2001;16(1):21–31. doi: 10.1093/her/16.1.21. [DOI] [PubMed] [Google Scholar]
  14. Gerstorf D, Heckhausen J, Ram N, Infurna FJ, Schupp J, Wagner GG. Perceived personal control buffers terminal decline in well-being. Psychol Aging. 2014;29(3):612–625. doi: 10.1037/a0037227. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gerstorf D, Röcke C, Lachman ME. Antecedent-consequent relations of perceived control to health and social support: longitudinal evidence for between-domain associations across adulthood. J Gerontol B Psychol Sci Soc Sci. 2011;66(1):61–71. doi: 10.1093/geronb/gbq077. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Growney CM, Hess TM. The influence of mood versus relevant self-perceptions in older adults’ interest in negative health-related information. Psychol Aging. 2019;34(3):348. doi: 10.1037/pag0000333. [DOI] [PubMed] [Google Scholar]
  17. Growney CM, Hess TM. Affective influences on older adults’ attention to self-relevant negative information. J Gerontol Series B. 2019;74(4):642–651. doi: 10.1093/geronb/gbx108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Henning G, Segel-Karpas D, Stenling A, Huxhold O. Subjective well-being across the retirement transition—historical differences and the role of perceived control. Psychol Aging. 2022;37(3):388–400. doi: 10.1037/pag0000662. [DOI] [PubMed] [Google Scholar]
  19. Hess TM, O'Brien EL, Voss P, Kornadt AE, Rothermund K, Fung HH, Popham LE. Context influences on the relationship between views of aging and subjective age: the moderating role of culture and domain of functioning. Psychol Aging. 2017;32(5):419–431. doi: 10.1037/pag0000181. [DOI] [PubMed] [Google Scholar]
  20. IBM Corp (2018) IBM SPSS Statistics for Windows, Version 26.0. Armonk, NY: IBM Corp.
  21. Kaas L, Kocharkov G, Preugschat E, Siassi N (2017) Low Homeownership in Germany-A Quantitative Exploration. (CESinfo Working Paper Series 6775). CESinfo Group Munich. Retrieved from http://ideas.repec.org/p/ces/ceswps/_6775.html
  22. Kim-Knauss Y, Lang FR. Late-life preparedness and its correlates: a behavioral perspective on preparatory activities. J Gerontol Series B. 2020 doi: 10.1093/geronb/gbaa088. [DOI] [PubMed] [Google Scholar]
  23. Kornadt AE, Rothermund K. Contexts of aging: assessing evaluative age stereotypes in different life domains. J Gerontol B Psychol Sci Soc Sci. 2011;66(5):547–556. doi: 10.1093/geronb/gbr036. [DOI] [PubMed] [Google Scholar]
  24. Kornadt AE, Rothermund K. Preparation for old age in different life domains: dimensions and age differences. Int J Behav Dev. 2014;38(3):228–238. doi: 10.1177/0165025413512065. [DOI] [Google Scholar]
  25. Kornadt AE, Voss P, Fung HH, Hess TM, Rothermund K. Preparation for old age: the role of cultural context and future perceptions. J Gerontol Series B. 2019;74(4):609–619. doi: 10.1093/geronb/gby075. [DOI] [PubMed] [Google Scholar]
  26. Lachman ME, Neupert SD, Agrigoroaei S. The relevance of control beliefs for health and aging. In: Schaie KW, Willis SL, editors. Handbook of the psychology of aging. New York: Elsevier Academic Press; 2011. pp. 175–190. [Google Scholar]
  27. Lachman ME, Weaver SL. The sense of control as a moderator of social class differences in health and well-being. J Pers Soc Psychol. 1998;74(3):763–773. doi: 10.1037/0022-3514.74.3.763. [DOI] [PubMed] [Google Scholar]
  28. Lee WKM, Hong-Kin K. Differences in expectations and patterns of informal support for older persons in Hong Kong: Modification to filial piety. Ageing Int. 2005;30(2):188–206. doi: 10.1007/s12126-005-1011-1. [DOI] [Google Scholar]
  29. Little TD, Preacher KJ, Selig JP, Card NA. New developments in latent variable panel analyses of longitudinal data. Int J Behav Dev. 2007;31:357–365. doi: 10.1177/0165025407077757. [DOI] [Google Scholar]
  30. Moffatt S, Heaven B. 'Planning for uncertainty’: narratives on retirement transition experiences. Ageing Soc. 2017;37(5):879–898. doi: 10.1017/S0144686X15001476. [DOI] [Google Scholar]
  31. Newsom JT. Longitudinal structural equation modeling: a comprehensive introduction. London: Routledge; 2015. [Google Scholar]
  32. Noone J, O’Loughlin K, Kendig H. Australian baby boomers retiring ‘early’: Understanding the benefits of retirement preparation for involuntary and voluntary retirees. J Aging Studies. 2013;27(3):207–217. doi: 10.1016/j.jaging.2013.02.003. [DOI] [PubMed] [Google Scholar]
  33. North MS, Fiske ST. Modern attitudes toward older adults in the aging world: A cross-cultural meta-analysis. Psychol Bull. 2015;141(5):993. doi: 10.1037/a0039469. [DOI] [PubMed] [Google Scholar]
  34. OECD (2017) Policies for an ageing society: recent measures and areas for further reform, Economics Department Working Paper
  35. Park J, Fung HH, Rothermund K, Hess TM. The impact of perceived control and future-self views on preparing for the old age: moderating influences of age, culture, and context. J Gerontol Series B. 2020;75(5):e18–e28. doi: 10.1093/geronb/gbz138. [DOI] [PubMed] [Google Scholar]
  36. Petkoska J, Earl JK. Understanding the influence of demographic and psychological variables on retirement planning. Psychol Aging. 2009;24(1):245–251. doi: 10.1037/a0014096. [DOI] [PubMed] [Google Scholar]
  37. Prenda KM, Lachman ME. Planning for the future: a life management strategy for increasing control and life satisfaction in adulthood. Psychol Aging. 2001;16(2):206–216. doi: 10.1037/0882-7974.16.2.206. [DOI] [PubMed] [Google Scholar]
  38. Reitzes DC, Mutran EJ. The transition to retirement: stages and factors that influence retirement adjustment. Int J Aging Human Develop. 2004;59(1):63–84. doi: 10.2190/NYPP-RFFP-5RFK-8EB8. [DOI] [PubMed] [Google Scholar]
  39. Robinson SA, Lachman ME. Perceived control and aging: a mini-review and directions for future research. Gerontology. 2017;63(5):435–442. doi: 10.1159/000468540. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Skinner EA. Perceived control, motivation, & coping. California: Sage Publications Inc; 1995. [Google Scholar]
  41. Sörensen S, Mak W, Chapman B, Duberstein PR, Lyness JM. The relationship of preparation for future care to depression and anxiety in older primary care patients at 2-year follow-up. Am J Geriatr Psychiatry. 2012;20(10):887–894. doi: 10.1097/JGP.0b013e31822ccd8c. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Topa G, Herrador-Alcaide T. Procrastination and financial planning for retirement: a moderated mediation analysis. J Neurosci Psychol Econ. 2016;9(3–4):169–181. doi: 10.1037/npe0000065. [DOI] [Google Scholar]
  43. Topa G, Moriano JA, Depolo M, Alcover C-M, Morales JF. Antecedents and consequences of retirement planning and decision-making: a meta-analysis and model. J Vocat Behav. 2009;75(1):38–55. doi: 10.1016/j.jvb.2009.03.002. [DOI] [Google Scholar]
  44. Trope Y, Igou ER, Burke CT. Mood as a resource in structuring goal pursuit. In: Forgas JP, editor. Affect in social thinking and behavior. Psychology Press; 2006. pp. 217–234. [Google Scholar]
  45. United Nations, Department of Economic and Social Affairs, Population Division (2019) World Population Prospects 2019: Ten Key Findings
  46. van Rooij MC, Lusardi A, Alessie RJ. Financial literacy, retirement planning and household wealth. Econ J. 2012;122(560):449–478. doi: 10.1111/j.1468-0297.2012.02501.x. [DOI] [Google Scholar]
  47. Vittersø J. Feelings, meanings, and optimal functioning: Some distinctions between hedonic and eudaimonic well-being. In: Waterman AS, editor. The best within us: positive psychology perspectives on eudaimonia. American Psychological Association; 2013. pp. 39–55. [Google Scholar]
  48. Voss P, Kornadt AE, Rothermund K. Getting what you expect? Future self-views predict the valence of life events. Dev Psychol. 2017;53(3):567–580. doi: 10.1037/dev0000285. [DOI] [PubMed] [Google Scholar]
  49. Vötter B, Schnell T. Cross-lagged analyses between life meaning, self-compassion, and subjective well-being among gifted adults. Mindfulness. 2019;10:1294–1303. doi: 10.1007/s12671-018-1078-x. [DOI] [Google Scholar]
  50. Wang M, Shultz KS. Employee retirement: a review and recommendations for future investigation. J Manag. 2010;36(1):172–206. [Google Scholar]
  51. Yeung DY, Zhou X. Planning for retirement: longitudinal effect on retirement resources and post-retirement well-being. Front Psychol. 2017;8:1300. doi: 10.3389/fpsyg.2017.01300. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Yuan X, Wang Q. A tentative study on differences and integration of Sino-Western filial piety culture. Asian Soc Sci. 2011;7:97. doi: 10.5539/ass.v7n8p97. [DOI] [Google Scholar]

Articles from European Journal of Ageing are provided here courtesy of Springer

RESOURCES