Within-Person Dynamics of Objective and Subjective Social Isolation in Midlife and Later Life

Abstract

Objectives

To investigate the within-person dynamics of objective and subjective social isolation among U.S. middle-aged and older adults and to explore gender differences in this relationship. Methods: Four waves of data from the Health and Retirement Study (HRS, 2006–2018, N = 5437) and the multiple group random intercept cross-lagged panel model were used. Results: Within-person deviation in expected subjective isolation predicted deviation in expected objective isolation years later. No corresponding cross-lagged effect of objective isolation on subjective isolation was found. Gender differences were detected: the within-person cross-lagged positive effect of subjective isolation on objective isolation was significant for men but not for women. Discussion: This study provides evidence for a unidirectional relationship between subjective and objective isolation at the within-person level: higher than expected increase in subjective isolation predicts higher than expected increase in subsequent objective isolation. This within-person process is more salient in men than in women.

Keywords

social disconnectedness social relationships social network loneliness

Introduction

Social isolation has objective and subjective dimensions. Objective social isolation is characterized by having few social interactions and contacts, and is related to situational factors such as small social networks and low participation in social activities (Anderson & Thayer, 2018). Subjective isolation refers to the subjective experience of having few social contacts such as feeling of loneliness. Both forms of social isolation have been identified as risk factors for poor physical and mental health (Courtin & Knapp, 2017; Wenger & Burholt, 2004), with effects comparable to well-established health risk factors such as smoking and obesity (Holt-Lunstad et al., 2010). Even though subjective and objective social isolation are different concepts, they have often been conflated in research and social programs (Wigfield et al., 2022). Further, the relationship between the two concepts is not well understood, hindering the development of effective intervention strategies.

A number of studies have shown that objective and subjective isolation are moderately correlated (Cornwell & Waite, 2009b; Steptoe et al., 2013; Victor et al., 2000). However, many of these prior studies are based on cross-sectional data which can only examine between-person differences, that is, whether people who are higher on one form of social isolation are also higher on the other form (Anderson & Thayer, 2018; Cigna, 2018; Cornwell & Waite, 2009b; Dahlberg et al., 2018; McHugh et al., 2017). A few longitudinal studies have reported a reciprocal relationship between objective and subjective isolation. Yet, they often conflate within-person and between-person variance in their investigations (Hu & Li, 2022; Petersen et al., 2016; Santini et al., 2020). Within-person predictive association reflects to what extent, for a given individual, changes in one dimension of social isolation predict future changes in the other dimension. Without disaggregation of within-person and between-person variance in the analysis, estimates obtained are potentially biased (Curran & Bauer, 2010).

In one study that has separated between-person and within-person variance, the analysis was set up to test predictive effects of objective social isolation on subjective social isolation (Petersen et al., 2016), making an assumption of a priori unidirectional relationship from objective to subjective isolation. Such assumption is seen in other studies as well, such as those that examine how social network characteristics and change (e.g., marital dissolution, kinship availability, number of friends) are related to loneliness (Dykstra & Fokkema, 2007; Nicolaisen & Thorsen, 2014; van Tilburg et al., 2015). Many previous studies have overlooked the possibility that subjective isolation can influence people’s objective isolation. Prior research have suggested that people who feel lonely hold negative expectations about social relationships, construe their world as threatening, and display a heightened hypervigilance for social threats in the environment (Cacioppo & Hawkley, 2009; DeWall et al., 2009). The negative attitude and thought of lonely people may inhibit them from engaging with others.

In sum, few studies to date have disaggregated between-person and within-person variation to investigate the association between objective and subjective social isolation, and fewer have tested the direction of effect at the within-person level. The current study fills this gap.

Gender Differences

Men and women might be differently affected by subjective and objective isolation in midlife and later life. Regarding objective isolation, women are more likely to have close social relationships than men at all ages (Victor et al., 2000). A recent study found that middle-age and older women showed an overall increase in objective social isolation over a 10-year period while their male counterparts showed no proportional change over time (Read et al., 2020). Regarding subjective isolation, while some studies found that women’s level of subjective isolation was higher than men’s (Kotwal et al., 2021; Petersen et al., 2016), other studies failed to reveal gender differences (Cigna, 2018). A study that examined gender-specific trends showed that there were substantial gender differences in age trends in subjective isolation (von Soest et al., 2020). Additionally, gender has also been proposed as a moderator in the subjective-objective isolation association (Takagi et al., 2020). For instance, infrequent social contacts are associated with women’s level of subjective isolation more than with men’s (Pinquart & Sorensen, 2001). A potential explanation is that relative to men, women have been socialized to value social relationships and thus experience a greater sense of loneliness when having few social contacts.

The Present Study

Numerous studies have shown that objective and subjective social isolation have independent negative effects on physical and mental health, and cognitive function and mortality in older adults (see reviews, Evans et al., 2019; Holt-Lunstad et al., 2015; Mehrabi & Béland, 2020). These studies have prompted calls to address social isolation as a public health issue (National Academies of Sciences & Medicine, 2020). A better understanding of the within-person process regarding the two forms of isolation and gender differences in this process would help inform the development of intervention strategies. Repeated measurements are required in order to understand the interrelationship between subjective and objective social isolation at the within-person level. Drawing on longitudinal data with four measurement points over a 12-year period from a nationally representative sample of adults aged 51 and older in the United States, this study aimed to examine the within-person dynamics of subjective and objective isolation (net of between-person differences) in middle and late adulthood. Additionally, this study examined whether gender moderates the subjective-objective isolation relationship.

Methods

Data

We used data from the Health and Retirement Study (HRS) which has collected data from a nationally representative sample of adults aged over 50 years old since 1992. From 2006 onward, a random half sample was selected to participate in the Enhanced Face-to-Face Interview (EFTF). The EFTF had a leave-behind questionnaire on psychosocial topics that included measures of subjective and objective social isolation. This random half sample was re-interviewed every four years and the latest data available are from 2018. The four waves (2006, 2010, 2014, and 2018) of the half sample comprised the longitudinal sample for the present study. The HRS was approved by the Institutional Review Board at the University of Michigan and all participants provided informed consent. Because the current study made use of the publicly available de-identified data of the HRS, this research was considered to be exempt from institutional review board approval or informed consent.

Because the analytic method used in this study requires a minimum of three waves of data to obtain unbiased estimates (Hamaker et al., 2015), this study restricted the sample to those who took part in at least three waves of the EFTF. Additionally, the analytic sample was restricted to those with no missing values in either objective or subjective social isolation measure (N = 5,437, n = 18,861). Estimates were weighted to adjust for differential probabilities of selection and nonresponse.

Objective Social Isolation

We followed the recent study by Read et al. (2020) to measure objective social isolation, which was derived from five binary items: whether the respondent (a) lived alone; had less than monthly contact, including face-to-face, telephone, or written/email contact with (b) child(ren), (c) other family members or (d) friends; and (e) did not attend meetings of non-religious organizations, such as political, community, or other interest groups in the past month. This measurement is comparable to indicators used in contemporaneous longitudinal cohort studies (Sommerlad et al., 2019; Stafford et al., 2018; Sutin et al., 2020). The objective isolation score ranged from 0 to 5, with higher scores indicating greater objective isolation.

Subjective Social Isolation

We used the 3-item UCLA Loneliness Scale to assess subjective social isolation (Hawkley et al., 2005). This scale has been widely used in research and practice (Perissinotto et al., 2019). It has robust construct validity across several countries (Russell, 1996). Respondents were asked how much of the time they feel lack companionship, left out, and isolated from others. Response could be: hardly ever or never (= 1), some of the time (= 2), or often (= 3). The three items were summed (range 1–9), with higher scores indicating more subjective isolation. Cronbach’s alpha of the loneliness scale ranges from 0.80 to 0.82 across the four waves.

Covariates

Gender, age, race/ethnicity, educational attainment, employment status and health status were included as covariates in the analysis as prior studies have consistently shown that they were associated with subjective and objective isolation (Anderson & Thayer, 2018; Dahlberg et al., 2018). Gender (male, female) was used as a control variable in the main analysis and was also examined as a moderator. Age was measured in years (51–96 at baseline, continuous); race (non-Hispanic white, racial/ethnic minorities), education (bachelors or higher, less than bachelors) and employment status (non-employed, currently working) were represented by dummy variables; and self-rated health was a single-item measure (excellent [= 1] to poor [= 5]). These covariates were assessed at the baseline year and were treated as time-invariant.

Analytic Approach

Several methods have been proposed to examine the rank-order changes and time-lagged associations between two variables. The typical modeling approach is the cross-lagged panel model (CLPM). It estimates the autoregressive effects and cross-lagged effects simultaneously. However, this approach assumes that there are no stable between-person differences in the variables of interest and thus may conflate within-person and between-person effects (Mund & Nestler, 2019). To address this limitation, the random intercept cross-lagged panel model (RI-CLPM) was proposed. By including a random intercept (i.e., a factor with all loadings constrained to 1), RI-CLPM accounts for trait-like, time-invariant stability and thus distinguishes the within-person process from the between-person difference (Hamaker et al., 2015). We report findings based on the RI-CLPM approach. We also fit the data using CLPM. Comparisons between CLPM and RI-CLPM are reported when appropriate.

We built the RI-CLPM for the estimation of the within-person dynamics of subjective and objective isolation in two steps. First, a RI-CLPM specifying a reciprocal relationship between subjective and objective isolation was estimated (see Supplementary Figure 1). Given that the intervals between the observations are equal over time, we tested whether the model should constrain autoregressive and cross-lagged parameters to be equal across measurement occasions. Model comparison test supports the constraints (chi-square (8) = 16.29, p < .05). Second, we compared a model that constrained the grand means of subjective and objective isolation to be equal over time with one that relaxed the constraints. The comparison did not support constraining the grand means (∆chi-square = 947, p < .001). Thus, the RI-CLPM used to estimate the within-person dynamics of subjective and objective isolation does not constrain the grand means but constrains autoregressive and cross-lagged parameters to be equal across measurement occasions.

To examine moderating effects of gender in the subjective-objective isolation relationship, we used the multiple group RI-CLPM approach. Equality constraints between men and women were placed on corresponding cross-lagged paths and chi-square difference tests were conducted.

All models were evaluated using chi-square, the comparative fit index (CFI), the Tucker–Lewis index (TLI), the root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR). CFI and TLI values of 0.95 or higher, RMSEA values of 0.06 or lower, and SRMR values of 0.08 or lower are considered a relatively good model fit (Hu & Bentler, 1999). In addition, the Akaike information criterion (AIC) and Bayesian information criterion (BIC) were used; the model with the lowest AIC and BIC was preferred (Schwarz, 1978). We used Stata 16 to prepare the data and used Mplus 8.1 to fit the cross-lagged panel models.

Results

Descriptive and Bivariate Analysis Findings

The HRS provides cross-sectional weights designed specifically for the EFTF sample who completed the leave-behind questionnaire. To account for the sampling structure and nonresponse, we provided weighted descriptive results for the time-invariant variables and longitudinal information of both dimensions of social isolation for the total sample and by gender (Table 1).

Table 1.

Weighted Baseline Characteristics.

	Total		Men		Women
	%	Mean	%	Mean	%	Mean
Age		61.27		60.66		61.79
Gender (female = 1)	54.22
Race (nonwhite = 1)	19.71		18.55		20.68
Education (college and above =1)	28.97		34.14		24.61
Employment (employed = 1)	48.95		54.59		44.19
Self-rated health (range: 1–5)		2.55		2.51		2.57
Longitudinal summaries of social isolation	Total		Men		Women
Wave	SI	OI	SI	OI	SI	OI
2006	4.46	1.72	4.40	1.87	4.51	1.60
2010	4.41	1.91	4.35	2.01	4.46	1.83
2014	4.40	2.11	4.30	2.18	4.48	2.05
2018	4.35	2.13	4.24	2.16	4.44	2.10

Note. SI = subjective isolation, OI = objective isolation.

Of the total sample at baseline, the average age was 61 years old, less than 20% were racial minorities, nearly a third had a college degree or above and a half were employed. The average self-rated health of the respondents was between very good and good. Men and women were about equally distributed. Compared to women, men were more educated and more likely to be employed. Across the four waves, men reported higher levels of objective isolation but lower levels of subjective isolation than women.

Supplementary Figure 2 depicts the general time trends of social isolation. Objective isolation tends to increase over the 12 years of observation period whereas subjective isolation seems to remain relatively stable across successive waves. These trends are consistent with prior research showing that with advancing age, there are losses in social relationships (Wrzus et al., 2013) but loneliness level tends to stay stable from middle adulthood into the beginning of very old age (Böger & Huxhold, 2018b).

Bivariate correlations between subjective and objective isolation across waves are presented in Table 2. Consistent with prior research (Böger & Huxhold, 2018a), the two forms of social isolation are positively correlated. Their within-wave correlations (ranging between 0.21 and 0.25) are slightly stronger than that between waves (ranging between 0.19 and 0.24). The within-wave correlation coefficients indicate a small to medium effect size (Cohen, 1992). Both subjective isolation and objective isolation are strongly correlated with themselves across successive waves (r_{subjective isolation} = 0.47–0.59, r_{objective isolation} = 0.50–0.63).

Table 2.

Correlations Among Subjective and Objective Isolation Across Waves.

	SI2006	SI2010	SI2014	SI2018	OI2006	OI2010	OI2014	OI2018
SI2006	1
SI2010	0.58*	1
SI2014	0.50*	0.56*	1
SI2018	0.47*	0.53*	0.59*	1
OI2006	0.25*	0.19*	0.15*	0.18*	1
OI2010	0.21*	0.24*	0.17*	0.20*	0.62*	1
OI2014	0.22*	0.24*	0.21*	0.21*	0.55*	0.63*	1
OI2018	0.20*	0.22*	0.19*	0.24*	0.50*	0.59*	0.61*	1

Note. Cells report correlation coefficients; Shaded areas show within-wave correlations; SI = subjective isolation, OI = objective isolation; *p < .05.

The RI-CLPM Results

The model fit of the RI-CLPM to estimate the within-person dynamics of subjective and objective isolation is rather good, with chi-square (37) = 149, AIC = 198,822, BIC = 199,363, CFI = 0.99, TLI = 0.99, SRMR = 0.01, and RMSEA = 0.02. To illustrate the difference between RI-CLPM and CLPM, we also fit the data using CLPM. Goodness-of-fit indices of the CLPM are not satisfactory (Supplementary Table 1). Chi-square test shows that the RI-CLPM is significantly better than the CLPM (∆chi-square = 1014, p < .001), suggesting that the RI-CLPM provides a better representation of the data.

Table 3 shows results from the RI-CLPM. Estimates for the cross-lagged paths from the RI-CLPM reflect within-person estimates. The statistically significant positive cross-lagged paths from subjective isolation to objective isolation indicate that a score above the person-specific mean in subjective isolation at one measurement point is associated with an objective isolation score above the person-specific mean at the next measurement point, controlling for previous deviations from the person-specific mean in objective isolation. The cross-lagged paths from objective isolation to subjective isolation are not statistically significant, which indicates that changes in a person’s level of objective isolation at one time point is not associated with changes in his/her subjective isolation at the subsequent time point. In comparison, estimates from using CLPM suggest that all cross-lagged paths are statistically significant, meaning that subjective and objective isolation have a reciprocal relationship (see Supplementary Table 2). The differences between CLPM and RI-CLPM results suggest that without accounting for between-person differences, CLPM conflates within-person and between-person variance in the estimation.

Table 3.

Standardized Parameter Estimates From RI-CLPM (Total sample: N = 5437).

Parameter	Estimate	SE	p
Autoregressive parameters
SI2006 → SI2010 (α2)	0.19	0.02	0.00
SI2010 → SI2014 (α3)	0.17	0.02	0.00
SI2014 → SI2018 (α4)	0.19	0.02	0.00
OI2006 → OI2010 (δ2)	0.17	0.02	0.00
OI2010 → OI2014 (δ3)	0.15	0.02	0.00
OI2014 → OI2018 (δ4)	0.16	0.02	0.00
Cross-lagged parameters
OI2006 → SI2010 (γ2)	−0.01	0.02	0.42
OI2010 → SI2014 (γ3)	−0.01	0.01	0.42
OI2014 → SI2018 (γ4)	−0.01	0.02	0.42
SI2006 → OI2010 (β2)	0.03	0.02	0.03
SI2010 → OI2014 (β3)	0.03	0.01	0.03
SI2014 → OI2018 (β4)	0.03	0.02	0.03

Note. SI = subjective isolation, OI = objective isolation. The model controls for gender, age, race/ethnicity, educational attainment, employment status, and self-rated health.

Multiple-Group RI-CLPM Results

Multiple group analysis in RI-CLPM was conducted to test moderating effects of gender on the interrelationship between subjective and objective isolation. Supplementary Table 3 and Figure 1 present the standardized path coefficients for men and women, respectively. At the between-person level, there is a significant correlation between subjective and objective isolation for both men and women (r_s = 0.34 and 0.35 for men and women, respectively, p < .01), indicating that men and women who report higher levels of objective isolation across the four measurement points are likely to report higher levels of subjective isolation across the 12 years. After controlling for between-person differences, consistent evidence is found for the within-person carryover effects for both genders: subjective isolation in 2006, 2010, and 2014 predicts within-person changes in subjective isolation four years later and objective isolation predicts within-person changes in objective isolation as well.

Figure 1.

Standardized path coefficients derived from the multigroup RI-CLPM for men (left) and women (right).

However, there are gender differences in the within-person cross-lagged effects. For men, consistent with estimates for the overall sample, there are significant predictive long-term effects of subjective isolation on objective isolation: when a man experiences more subjective isolation than usual at one measurement point, he has more objective social isolation than usual at the next measurement point. This finding holds for 2006, 2010, and 2014. For women, the non-significant cross-lagged paths suggest that neither her subjective isolation predicts within-person changes in objective isolation nor her objective isolation predicts within-person changes in subjective isolation years later. We compared the model fit of one allowing the within-person autoregressive and cross-lagged parameters to differ by gender to one constraining the corresponding paths to be the same across gender. The chi-square is 15.69 (p = .003), suggesting that the gender-specific model is necessary when exploring the within-person dynamics of social isolation. Further model comparison tests show that cross-lagged paths from objective isolation to subjective isolation and within-person carryover effects of objective isolation are different for men and women (chi-square values equal 4.27 and 9.78, respectively).

Discussion

Drawing on data from a nationally representative longitudinal study, the present study investigates the within-person dynamics of subjective and objective isolation in middle and late adulthood. Findings substantially increase our understanding of how the two forms of social isolation influence each other over time within individuals and how men and women may differ in the process.

We found a consistent, unidirectional cross-lagged effect of subjective isolation on objective isolation at the within-person level, net of between-person variation. Specifically, the finding shows that higher than expected subjective isolation levels at one time point are associated with higher-than-usual levels of objective social isolation four years later. Due to the consistency across years, this finding offers strong support for the idea that changes in subjective isolation precedes changes in objective social isolation in midlife and later life. This finding is consistent with prior research showing that subjective isolation itself can promote negativity. Subjectively isolated people tend to focus on negative social information, expect social interactions to be negative, and regard pleasant interpersonal interactions as less pleasant than individuals who feel socially connected (Cacioppo & Cacioppo, 2014; Cacioppo & Hawkley, 2009; DeWall et al., 2009). Those negative perceptions and experience in social interactions may eventually lead to social withdrawal and objective social isolation.

The within-person process shows that changes in objective isolation do not predict changes in subjective isolation. Although counterintuitive, this finding is in line with the model of selection, optimization, and compensation (SOC, Baltes & Baltes, 1990). According to the socioemotional selectivity theory, with advancing age, people may have fewer distressing ties and obtain greater satisfaction from close relationships (Carstensen et al., 1999). Thus, the increase in objective isolation does not necessarily lead to increase in subjective isolation. Also, in the face of shrinking social networks, people may choose to develop close relationships and shift expectations, leading to no greater subjective isolation (Cornwell & Waite, 2009a).

An intriguing finding of this study is gender differences in the within-person process. Specifically, we find that the above within-person, unidirectional cross-lagged effects of subjective isolation on objective isolation are more salient in men than in women, although in both genders intra-individual change in objective isolation is not significantly correlated with intra-individual change in subjective isolation. In addition, at the between-person level, subjective and objective isolation are significantly and positively correlated in both genders. In other words, although for both men and women, those who are high in one form of social isolation are likely to be high in the other form, for women, within-person changes in one form of isolation are not significantly associated with changes in the other form over years.

We offer two explanations for the gender differences found in this study. First, the gender socialization theory suggests that social scripts shape men’s and women’s identity and sense of self differently (Calasanti, 2004). Feeling of loneliness may make men feel their masculinity threatened, which may lead to avoidant behaviors and withdrawal from social interactions. Men also have learned to endure physical and emotional discomfort and may turn to solitary activities to cope with the distress caused by loneliness, further isolating themselves. Second, men’s subjective isolation may have a lot to do with the loss of their spouse. Recent research shows that after spousal loss, men decrease social engagement not only in the first year but also in subsequent years (Yoon et al., 2022). Losing their spouse can mean losing other social connections for men as wives are often the “connector” in men’s social networks. More research is needed to understand gender differences in the within-person process related to the two forms of social isolation.

Our findings suggest that gender-specific interventions to address social isolation may be appropriate. For middle-aged and older men, interventions should aim to reduce their subjective isolation which is likely to lead to reduction of objective social isolation as well. Interventions targeting objective social isolation may not have an effect on subjective isolation for men. For women, however, interventions should address subjective and objective isolation separately as improvement in one form of isolation does not necessarily result in changes in the other form. Additionally, autoregressive paths from the RI-CLPM are all statistically significant for both genders. This finding signifies that for a given individual, heightened subjective isolation foretells future increases in subjective isolation, and heightened objective isolation foretells future increases in objective isolation. Social service programs should pay attention to the long-term within-person effects when designing programs to promote adults’ social engagement or reduce loneliness.

We should note that estimates obtained from the standard CLPM that aggregates between-person and within-person variances reveal a bi-directional relationship between objective and subjective isolation. The discrepancy between the estimates of the CLPM and RI-CLPM underscores the importance of disaggregating the two types of variance in order to provide accurate estimates of predictive effects within individuals. The use of RI-CLPM to analyze four-wave of data spanning twelve years is a strength of this study. In addition, the analysis is based on a nationally representative sample of adults aged over 50 in the United States, increasing the generalizability of the findings. The measurement of subjective isolation has been widely used while the measure of objective isolation is comparable to prior studies (Perissinotto et al., 2019; Sommerlad et al., 2019; Stafford et al., 2018; Sutin et al., 2020).

Some limitations of this study should be noted. First, although this study provides the first evidence of a long-term, within-person effect of subjective isolation on objective isolation, this finding should not be interpreted as a causal effect. There might still be other unobserved time-varying variables that cause this cross-lagged effect. Second, estimating the within-person process is only a step toward a more complete understanding of the relationship between subjective and objective isolation. More research is needed to understand the developmental process of subjective and objective social isolation and to explore heterogeneities in various social contexts. Third, we tentatively conclude that there was no cross-lagged predictive effect of objective isolation on subjective isolation at the within-person level. However, this result may be due to the choice of time lags. The four-year time lag of the current study may be too long to detect short-term effects of objective isolation on subjective isolation.

To conclude, this study finds that within-person changes in subjective social isolation result in changes in objective social isolation, and this process is more salient in men than in women. Men who experience increase over their typical levels of subjective isolation at one time point are likely to experience increase over their typical levels of objective social isolation four, eight, and twelve years later. The corresponding within-person predictive effects of subjective isolation are not observed in women. In both genders, within-person changes in objective social isolation do not predict changes in subjective isolation.

Supplemental Material

Supplemental Material - Within-Person Dynamics of Objective and Subjective Social Isolation in Midlife and Later Life

Supplemental Material for Within-Person Dynamics of Objective and Subjective Social Isolation in Midlife and Later Life by Mengsha Luo and Lydia W. Li in Journal of Aging and Health

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Office for Philosophy and Social Sciences (Grant No. 20CRK007).

Ethical Approval

The HRS was approved by the Institutional Review Board at the University of Michigan and all participants provided informed consent. Because the current study made use of the publicly available de-identified data of the HRS, this research was considered to be exempt from institutional review board approval or informed consent.

ORCID iDs

Mengsha Luo

Lydia W. Li

Supplemental Material

Supplemental material for this article is available online.

References

Anderson

G. O.

Thayer

C. E.

(2018). Loneliness and social connections: A national survey of adults 45 and older. AARP Foundation.

Baltes

P. B.

Baltes

M. M.

(1990). Psychological perspectives on successful aging: The model of selective optimization with compensation. In Baltes

P. B.

Baltes

M. M.

(Eds.), Successful aging: perspectives from the behavioral sciences (Vol. 1, pp. 1–34). Cambridge University Press.

Böger

Huxhold

(2018a). Do the antecedents and consequences of loneliness change from middle adulthood into old age? Developmental Psychology, 54(1), 181–197, https://doi.org/10.1037/dev0000453

Böger

Huxhold

(2018b). Age-related changes in emotional qualities of the social network from middle adulthood into old age: How do they relate to the experience of loneliness? Psychology and Aging, 33(3), 482–496, https://doi.org/10.1037/pag0000222

Cacioppo

J. T.

Cacioppo

(2014). Social relationships and health: The toxic effects of perceived social isolation. Social and Personality Psychology Compass, 8(2), 58–72, https://doi.org/10.1111/spc3.12087

Cacioppo

J. T.

Hawkley

L. C.

(2009). Loneliness. In Handbook of individual differences in social behavior. (pp. 227–240). The Guilford Press.

Calasanti

(2004). Feminist gerontology and old men. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 59(6), S305–S314, https://doi.org/10.1093/geronb/59.6.s305

Carstensen

L. L.

Isaacowitz

D. M.

Charles

S. T.

(1999). Taking time seriously: A theory of socioemotional selectivity. American Psychologist, 54(3), 165–181, https://doi.org/10.1037/0003-066x.54.3.165

Cigna . (2018). Cigna US loneliness index: Survey of 20,000 Americans examining behaviors driving loneliness in the United States.

10.

Cohen

(1992). A power primer. Psychological Bulletin, 112(1), 155–199, https://doi.org/10.1037/0033-2909.112.1.155

11.

Cornwell

E. Y.

Waite

L. J.

(2009a). Measuring social isolation among older adults using multiple indicators from the NSHAP study. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 64B(suppl_1), i38–i46, https://doi.org/10.1093/geronb/gbp037

12.

Cornwell

E. Y.

Waite

L. J.

(2009b). Social disconnectedness, perceived isolation, and health among older adults. Journal of Health and Social Behavior, 50(1), 31–48, https://doi.org/10.1177/002214650905000103

13.

Courtin

Knapp

(2017). Social isolation, loneliness and health in old age: A scoping review. Health and Social Care in the Community, 25(3), 799–812, https://doi.org/10.1111/hsc.12311

14.

Curran

P. J.

Bauer

D. J.

(2010). The disaggregation of within-person and between-person effects in longitudinal models of change. Annual Review of Psychology, 62(1), 583–619, https://doi.org/10.1146/annurev.psych.093008.100356

15.

Dahlberg

Agahi

Lennartsson

(2018). Lonelier than ever? Loneliness of older people over two decades. Archives of Gerontology and Geriatrics, 75, 96–103, https://doi.org/10.1016/j.archger.2017.11.004

16.

DeWall

C. N.

Twenge

J. M.

Gitter

S. A.

Baumeister

R. F.

(2009). It’s the thought that counts: The role of hostile cognition in shaping aggressive responses to social exclusion. Journal of Personality and Social Psychology, 96(1), 45–59, https://doi.org/10.1037/a0013196

17.

Dykstra

P. A.

Fokkema

(2007). Social and emotional loneliness among divorced and married men and women: Comparing the deficit and cognitive perspectives. Basic and Applied Social Psychology, 29(1), 1–12, https://doi.org/10.1080/01973530701330843

18.

Evans

I. E. M.

Martyr

Collins

Brayne

Clare

(2019). Social isolation and cognitive function in later life: A systematic review and meta-analysis. Journal of Alzheimer’s Disease, 70(s1), S119–S144, https://doi.org/10.3233/jad-180501

19.

Hamaker

E. L.

Kuiper

R. M.

Grasman

R. P. P. P.

(2015). A critique of the cross-lagged panel model. Psychological Methods, 20(1), 102–116, https://doi.org/10.1037/a0038889

20.

Hawkley

L. C.

Browne

M. W.

Cacioppo

J. T.

(2005). How can I connect with thee?: Let me count the ways. Psychological Science, 16(10), 798–804, https://doi.org/10.1111/j.1467-9280.2005.01617.x

21.

Holt-Lunstad

Smith

T. B.

Baker

Harris

Stephenson

(2015). Loneliness and social isolation as risk factors for mortality: A meta-analytic review. Perspectives on Psychological Science, 10(2), 227–237, https://doi.org/10.1177/1745691614568352

22.

Holt-Lunstad

Smith

T. B.

Layton

J. B.

(2010). Social relationships and mortality risk: A meta-analytic review. Plos Medicine, 7(7), Article e1000316, https://doi.org/10.1371/journal.pmed.1000316

23.

L. t.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55, https://doi.org/10.1080/10705519909540118

24.

R. X.

L. W.

(2022). Social disconnectedness and loneliness: Do self-perceptions of aging play a role? The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, Advance Access published January 27, 2022.

25.

Kotwal

A. A.

Cenzer

I. S.

Waite

L. J.

Covinsky

K. E.

Perissinotto

C. M.

Boscardin

W. J.

Hawkley

L. C.

Dale

Smith

A. K.

(2021). The epidemiology of social isolation and loneliness among older adults during the last years of life. Journal of the American Geriatrics Society, 61(3), 643, https://doi.org/10.1016/j.jpainsymman.2021.01.018

26.

McHugh

J. E.

Kenny

R. A.

Lawlor

B. A.

Steptoe

Kee

(2017). The discrepancy between social isolation and loneliness as a clinically meaningful metric: Findings from the Irish and English longitudinal studies of ageing (TILDA and ELSA). International Journal of Geriatric Psychiatry, 32(6), 664–674, https://doi.org/10.1002/gps.4509

27.

Mehrabi

Béland

(2020). Effects of social isolation, loneliness and frailty on health outcomes and their possible mediators and moderators in community-dwelling older adults: A scoping review. Archives of Gerontology and Geriatrics, 90, 104119, https://doi.org/10.1016/j.archger.2020.104119

28.

Mund

Nestler

(2019). Beyond the cross-lagged panel model: Next-generation statistical tools for analyzing interdependencies across the life course. Advances in Life Course Research, 41, 100249, https://doi.org/10.1016/j.alcr.2018.10.002

29.

National Academies of Sciences, E., & Medicine . (2020). Social isolation and loneliness in older adults: Opportunities for the health care system. The National Academies Press.

30.

Nicolaisen

Thorsen

(2014). Who are lonely? Loneliness in different age groups (18–81 years old), using two measures of loneliness. The International Journal of Aging and Human Development, 78(3), 229–257, https://doi.org/10.2190/ag.78.3.b

31.

Perissinotto

Holt-Lunstad

Periyakoil

V. S.

Covinsky

(2019). A practical approach to assessing and mitigating loneliness and isolation in older adults. Journal of the American Geriatrics Society, 67(4), 657–662, https://doi.org/10.1111/jgs.15746

32.

Petersen

Kaye

Jacobs

P. G.

Quinones

Dodge

Arnold

Thielke

(2016). Longitudinalrelationship between loneliness and social isolation in older adults: Results from the Cardiovascular Health Study. Journal of Aging and Health, 28(5), 775–795, https://doi.org/10.1177/0898264315611664

33.

Pinquart

Sorensen

(2001). Influences on loneliness in older adults: A meta-analysis. Basic and Applied Social Psychology, 23(4), 245–266, https://doi.org/10.1207/s15324834basp2304_2

34.

Read

Comas-Herrera

Grundy

(2020). Social isolation and memory decline in later-life. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 75(2), 367–376, https://doi.org/10.1093/geronb/gbz152

35.

Russell

D. W.

(1996). UCLA loneliness scale (version 3): Reliability, validity, and factor structure. Journal of Personality Assessment, 66(1), 20–40, https://doi.org/10.1207/s15327752jpa6601_2

36.

Santini

Z. I.

Jose

P. E.

York Cornwell

Koyanagi

Nielsen

Hinrichsen

Meilstrup

Madsen

K. R.

Koushede

(2020). Social disconnectedness, perceived isolation, and symptoms of depression and anxiety among older Americans (NSHAP): A longitudinal mediation analysis. The Lancet Public Health, 5(1), Article e62–e70, https://doi.org/10.1016/s2468-2667(19)30230-0

37.

Schwarz

(1978). Estimating the dimension of a model. Annals of Statistics, 6(2), 461–464, https://doi.org/10.1214/aos/1176344136

38.

Sommerlad

Sabia

Singh-Manoux

Lewis

Livingston

(2019). Association of social contact with dementia and cognition: 28-year follow-up of the Whitehall II cohort study. Plos Medicine, 16(8), Article e1002862, https://doi.org/10.1371/journal.pmed.1002862

39.

Stafford

von Wagner

Perman

Taylor

Kuh

Sheringham

(2018). Social connectedness and engagement in preventive health services: An analysis of data from a prospective cohort study. The Lancet Public Health, 3(9), Article e438–e446, https://doi.org/10.1016/s2468-2667(18)30141-5

40.

Steptoe

Shankar

Demakakos

Wardle

(2013). Social isolation, loneliness, and all-cause mortality in older men and women. Proceedings of the National Academy of Sciences, 110(15), 5797–5801. https://doi.org/10.1073/pnas.1219686110

41.

Sutin

A. R.

Stephan

Luchetti

Terracciano

(2020). Loneliness and risk of dementia. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 75(7), 1414–1422, https://doi.org/10.1093/geronb/gby112

42.

Takagi

Saito

Chan

(2020). Gender differences in the association between social relationships and loneliness among older adults in Singapore. Journal of Population Research, 37(3), 243–263, https://doi.org/10.1007/s12546-020-09242-0

43.

van Tilburg

T. G.

Aartsen

M. J.

van der Pas

(2015). Loneliness after divorce: A cohort comparison among Dutch young-old adults. European Sociological Review, 31(3), 243–252, https://doi.org/10.1093/esr/jcu086

44.

Victor

Scambler

Bond

Bowling

(2000). Being alone in later life: Loneliness, social isolation and living alone. Reviews in Clinical Gerontology, 10(4), 407–417, https://doi.org/10.1017/s0959259800104101

45.

von Soest

Luhmann

Hansen

Gerstorf

(2020). Development of loneliness in midlife and old age: Its nature and correlates. Journal of Personality and Social Psychology, 118(2), 388–406, https://doi.org/10.1037/pspp0000219

46.

Wenger

G. C.

Burholt

(2004). Changes in levels of social isolation and loneliness among older people in a rural area: A twenty–year longitudinal study. Canadian Journal on Aging/La Revue canadienne du vieillissement, 23(2), 115–127, https://doi.org/10.1353/cja.2004.0028

47.

Wigfield

Turner

Alden

Green

Karania

V. K.

(2022). Developing a new conceptual framework of meaningful interaction for understanding social isolation and loneliness. Social Policy and Society, 21(2), 172–193, https://doi.org/10.1017/s147474642000055x

48.

Wrzus

Hänel

Wagner

Neyer

F. J.

(2013). Social network changes and life events across the life span: A meta-analysis. Psychological Bulletin, 139(1), 53–80, https://doi.org/10.1037/a0028601

49.

Yoon

Park

G.-R.

Kim

(2022). Psychosocial trajectories before and after spousal loss: Does gender matter? Social Science and Medicine, 294, 114701, https://doi.org/10.1016/j.socscimed.2022.114701

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.23 MB