Interaction of widowhood, gender, and age in predicting loneliness among older adults in China

Abstract

Background

Loneliness is a significant issue for the elderly, and widowhood is considered a major risk factor. However, research on the intersectional effects of gender, age, and widowhood on loneliness is limited, especially within the Chinese cultural context.

Methods

Using six waves (2002–2018) of national longitudinal data from the Chinese Longitudinal Healthy Longevity Survey (N = 22,777), this study employed multilevel mixed-effects ordered logistic regression to analyze the impact of widowhood on loneliness. Moderating roles of gender and age were examined through interaction effects.

Results

Widowhood significantly increased loneliness across genders and age groups, but this effect diminished with age. Widowed men experienced greater loneliness than women, but this difference converged by age 90. The buffering effect of age on the widowhood-loneliness link was less pronounced among older women.

Conclusion

The study unravels the complexity of how gender, age, and widowhood interact to shape loneliness in later life. Targeted interventions considering these intersections are needed to alleviate loneliness among Chinese widowed elderly.

Keywords:

• Age
• gender
• interaction
• loneliness
• widowhood

Introduction

Loneliness, defined as an unpleasant subjective feeling of social isolation or lack of companionship (Peplau & Perlman, 1982), is prevalent among older adults (Iecovich et al., 2011). According to Weiss’ (1973) classification of loneliness, it encompasses emotional loneliness and social loneliness. Emotional loneliness originates from a lack of intimate attachment to others, while social loneliness stems from a lack of social relationship networks with shared interests. It is important to study loneliness because of its significant effects on older adults’ mental and physical health, including exacerbating conditions like depression, cognitive decline, and mortality (Hawkley & Cacioppo, 2010; Ong et al., 2016). There are several contributing factors to loneliness among older adults, such as poor health, limited mobility, retirement, communication difficulties, living alone, low income, and spousal bereavement (Dykstra et al., 2005;Collins et al., 2020; Lee et al., 2019). Widowhood, defined as the loss of a spouse through death, often leads to a change in social status as the surviving spouse transitions into a new phase of singlehood (Utz et al., 2002). The change in social relationships after spousal loss often leaves widowed older adults vulnerable to loneliness (Carr et al., 2018; Olawa et al., 2021; Yang, 2021), although loneliness is distinct from simply being alone. Many studies have consistently found an association between widowhood and heightened loneliness, especially among older adults (Christiansen et al., 2016; Dykstra et al., 2005; Yang & Gu, 2021). Research recognizes that losing a lifelong partner can leave an emotional void that is hard to fill (Malli et al., 2023).

Past studies have indicated that the experience of loneliness following widowhood is not uniform among individuals (Bennett, 2005). Some individuals may navigate through an intense phase of loneliness, marked by a profound absence and the accompanying loss of regular interaction. However, they may eventually forge new routines, create fresh connections, and cultivate a renewed sense of self as an individual rather than part of a couple (Davies et al., 2016). Earlier research also suggests that loneliness is multifaceted, encompassing a range of sentiments and emotional states. Widowed women may experience various degrees and types of loneliness, which may coalesce into situational feelings or even a chronic condition (Peplau & Perlman, 1982). Additionally, the severity of loneliness can be shaped by factors such as the quality of the marital relationship preceding the spouse’s death, the surviving spouse’s social network, their coping mechanisms, gender identity, socioeconomic status, health status, and living conditions (De Jong Gierveld et al., 2009; Abbas et al., 2020).

Previous studies have also found that experiences of loneliness can significantly vary across different genders groups. For example, one study revealed that widowed men experience significantly more loneliness than their married counterparts, while no significant difference was discerned among women who were married or widowed (Perlman et al., 1978; Dykstra & Jenny, 2004). Another study further substantiated this finding, indicating that the likelihood of experiencing loneliness was over four times higher for widowed men than for those who were married, suggesting that the loss of a spouse might be exacerbated by men’s relatively less diverse social networks (Menec et al., 2019). However, some studies have demonstrated that women typically report higher levels of loneliness than men. For instance, studies have found that women frequently report both loneliness and fear more than men (Jakobsson & Hallberg, 2005). Furthermore, additional research has determined that women are more likely to spend more time alone and report feelings of loneliness than men (Brittain et al., 2017; Dong & Chen, 2017). A study conducted in China also reported that women were 1.36 times more likely to experience loneliness than men (Yang & Victor, 2008).

Gender variances in the experience of loneliness and widowhood are influenced by numerous factors, including societal roles, coping strategies, emotional expressiveness, and social networks. Societal roles significantly shape these experiences, with traditional roles often designating men as primary earners and women as primary caregivers. Consequently, men may rely on their spouses for emotional support and daily care, rendering the loss of a spouse particularly impactful (Fischer & Beresford, 2015). Widowhood deprives them of an important source of emotional support, resulting in increased loneliness and a greater burden of household chores (Yoon et al., 2022). However, the loss of a spouse also significantly affects women, who generally have higher expectations for intimacy and emotional support from a marriage (Bennett, 1997). The loss of a spouse can critically undermine their need for intimacy, potentially engendering greater feelings of loneliness than experienced by men. Additionally, widowed women may also experience difficulties in adjusting, loss of financial independence, and a sense of emptiness due to the absence of life purpose (Abbas et al., 2020)

Differences in coping strategies between genders also play a role. Women tend to exhibit more internalizing emotional behaviors, such as loneliness and depression, while men often display more externalizing emotional behaviors (Yang, 2021). These gender-based differences in coping mechanisms can influence the outcomes of widowhood. Emotional expressiveness is another contributing factor, with women generally more expressive of their emotions, while men are typically less inclined to admit feeling lonely (Barreto et al., 2021).

Furthermore, social networks can contribute to these gender disparities. The loss of a spouse in later life may lead to increased sense of identity and solidarity among widowed women with female friends and family members (Collins, 2014). Women are often more socially embedded than men and have more emotionally rewarding relationships (Okun & Keith, 1998), potentially providing a buffer against loneliness. However, the gender gap in social support in old age may dissipate if the impact of widowhood is removed (Olawa et al., 2021). In summary, the gender differences in the experience of loneliness and widowhood are multifaceted and influenced by various factors. Understanding these differences can assist in the development of targeted interventions to mitigate loneliness among the widowed demographic.

Beyond gender differences, widowhood is a substantial predictor of loneliness across various age groups, with the effect being more pronounced among the “young-old” individuals aged 65–79 when compared to their older counterparts (Yang & Gu, 2021). This can be attributed to the fact that widowhood is less common among the young-old due to lower mortality rates, rendering the experience more isolating (Dahlberg et al., 2015). Nonetheless, the impact of widowhood on loneliness remains evident even among the “oldest-old” individuals aged 80 and above, albeit diminishing with age (Yang & Gu, 2021). Notably, the influence of widowhood on loneliness is not solely immediate but also long-term. Research has found that the effect of widowhood on loneliness is more potent in the initial years following the spouse’s bereavement but continues to be significant over time (Davies et al., 2016; Wenger et al., 1996). This indicates that widowhood is not a binary event but rather a complex process that evolves over time, with implications for loneliness that persist well beyond the initial bereavement period (Yang & Gu, 2021).

While a significant body of research has explored widowhood, gender, age, and loneliness, studies specifically probing the interaction of these factors are relatively limited. However, a few have investigated this complex intertwining to some degree. For instance, one study found that the odds ratio of loneliness declines with age in both older men and women (Yang & Gu, 2021). Long-term effects of widowhood on loneliness do not appear to differ significantly between older men and women, suggesting that feelings of loneliness due to widowhood impact both genders equally over time (Nicolaisen & Thorsen, 2014). Yet, the association appears stronger in men, although such gender difference is not statistically significant in most instances. This could be attributed to the fact that older men who lose their spouse often have fewer social contacts and friends, increasing their vulnerability to loneliness (Nicolaisen & Thorsen, 2014). Nevertheless, compared to women, widowed men have better prospects in seeking new partners in later life as they view remarriage as a normal process and often choose to remarry (Carr, 2011; Collins, 2014), which could potentially alleviate their feelings of loneliness.

Upon reviewing the literature, it is evident that despite an expanding body of research on the individual aspects of widowhood, gender, age, and loneliness, there is a significant gap in the understanding of the intricate interplay among these factors. Most studies have concentrated on one or two of these variables, yielding valuable insights into their individual impacts on loneliness. However, the interaction of gender, age, and widowhood and its relation to loneliness remain inadequately explored. For example, while numerous studies focus on gender differences in the experience of widowhood and others examine the role of age in loneliness, few have integrated all these factors to offer a comprehensive understanding of their collective effects on loneliness.

Moreover, existing studies often present conflicting findings, underscoring the need for more nuanced research. The literature also generally lacks focus on diversity within these groups. For instance, how does the interaction between gender, age, and widowhood in relation to loneliness change in different cultural contexts? Much of the current research on widowhood and loneliness has been conducted in Western societies, leaving a gap in understanding within non-Western contexts. There is a pressing need for more research within the Chinese context to inform local policies and support services.

Chinese culture, heavily influenced by Confucian ideologies, places considerable importance on family bonds and filial piety. The implications of losing a spouse may have unique consequences within this cultural context, potentially influencing loneliness differently as compared to Western societies. China is rapidly aging, with the elderly population projected to reach 400 million by 2050. This demographic transition implies that an increasing number of individuals will experience widowhood, rendering it a critical issue to investigate. Traditional gender roles are deeply entrenched in Chinese society. Understanding how these roles shape the experience of widowhood and loneliness can yield insights specific to the Chinese context. Rapid urbanization and the one-child policy have precipitated smaller family sizes and increased geographical separation between parents and children in China. These societal shifts may affect the support available to those who are widowed and influence their experiences of loneliness.

In conclusion, there is a pressing need for further research to address this gap in the literature and provide a more complete understanding of how the interplay of gender, age, and widowhood influences loneliness. By exploring this issue within the Chinese context, researchers can contribute to a broader global understanding of widowhood and loneliness while also yielding culturally relevant insights that can inform targeted interventions to alleviate loneliness among different subsets of the widowed population. Consequently, our primary research question is: How do age and gender interact to mediate the relationship between widowhood and loneliness within the Chinese context? In pursuing answers, we anticipate revealing nuanced patterns where the combined effects of age and gender yield experiences that may significantly differ from what would be predicted by examining each factor separately. Such findings could revolutionize the design of support systems and interventions, underscoring the importance of a tailored, intersectional approach in addressing the challenges of widowhood.

Methods

Data

The data used for this research were provided by the Chinese Longitudinal Healthy Longevity Survey (CLHLS). The CLHLS is a national, longitudinal survey conducted in China that has been ongoing since 1998 with follow-up surveys conducted every two to three years. The study is jointly conducted by Peking University and Duke University, covers 23 provinces in China representing 85% of the total population, and is supported by funds from the U.S. National Institutes on Aging (NIA), China Natural Science Foundation, China Social Science Foundation, and UNFPA. The study population mainly consisted of elderly people aged 65 and above. The details regarding sampling design and data quality assessment can be found in Zeng’s (2012) work. The CLHLS study obtained ethical approval from the Biomedical Ethics Committee of Peking University, Beijing, China (IRB00001052-13074), and all participants provided informed consent prior to completing the research questionnaire.

Since the CLHLS survey incorporated a sample group of individuals aged 60–79 starting from 2002, for the purpose of this study, we utilized data from the 2002, 2005, 2008, 2011, 2014, and 2018 waves of the CLHLS. We removed samples with missing or unreasonable values for key variables such as self-reported loneliness, marital status, age, and gender. Unreasonable samples included data that exceeded the range of the variable, as well as data with logical errors in key variables from the follow-up survey. Examples of logical errors included cases where the age reported in the initial survey was greater than that reported in the follow-up survey, changes in gender data, and contradiction in marital status before and after. After removing missing and unreasonable samples on the variables, our final sample size consisted of 22,777 individuals, with a total of 45,139 observations.

Variables

The dependent variable of this study was self-rated loneliness. Loneliness is measured in various ways. some studies use a scale specifically designed for the survey, while others use a single-item question (Morrish & Medina-Lara, 2021). In this study, we used a single item to measure loneliness among older adults: "How often do you feel lonely?" with four response values: never (coded as 1), rarely (coded as 2), sometimes (coded as 3), and often (coded as 4). This method is often used and accepted in research on loneliness in old age because it presents an everyday concept that is routinely used in daily interactions (Victor et al., 2006). Despite some arguments that this might be a weakness, there is substantial evidence supporting the appropriateness of single-item measures of loneliness. These single-item measures are highly correlated with multi-item questionnaires on loneliness such as the 20-item UCLA Loneliness Scale and its shorter forms (Hughes et al., 2004; Iecovich, 2013). The results of studies employing single-item measures of loneliness converge well with studies using multiple items.

Given the limited number of unmarried, divorced, and separated elderly individuals in China, this study categorized marital status into two groups: married (coded as 0), and widowed (coded as 1). The control variables in this study included gender (male coded as 0; female coded as 1), age (60 to 104 years), place of residence (urban coded as 1; town coded as 2; rural coded as 3), educational background, family economic status, health conditions, health behaviors, and living arrangements. Elderly education was divided into four categories: no formal education (coded as 1); 1–5 years of education (coded as 2); 6–9 years of education (coded as 3); more than 10 years of education (coded as 4). Family economic status was measured by two variables: self-assessed family economic level (low coded as 1; medium coded as 2; high coded as 3) and whether illness was treated in a timely manner (no coded 0; yes coded as 1).

In this study, cognitive function, basic activities of daily living (ADLs), instrumental ADLs (IADLs), and chronic diseases were used to measure elderly health status. Cognitive function was categorized into four grades based on the Mini-Mental State Examination (MMSE) scores (Folstein et al., 1975): normal (score 24–30, coded as 4); mild impairment (score 18–23, coded as 3); moderate impairment (score 10–17, coded as 2); severe impairment (score 0–9, coded as 1). ADLs included 6 basic daily functioning measurements, and disability in one or more item indicated overall ADL disability. IADLs included eight functional measurements categorized as: severe disability with disabilities in >7 items, moderate disability with 4–6 disabilities, mild disability with 1–3 disabilities, and no disability with 0 disabilities. Chronic diseases were categorized as: no chronic disease, 1 chronic disease, 2 chronic diseases, and ≥3 chronic diseases. An overall health index was constructed based on these four health indicators using Item Response Theory model (De Ayala & Little, 2022), ranging 0–100 with higher scores indicating better health. Health behaviors were measured by two variables: current smoking (yes coded as 1, no coded as 0) and current alcohol consumption (yes coded as 1, no coded as 0). Living arrangement was a dichotomous variable: lives with children (coded as 1); lives without children (alone or only with spouse, coded as 0). The statistical distribution of the variables at baseline is summarized in Table 1

Table 1. Descriptive Statistics for Variables in Baseline (N = 22777).

Variables		Frequency	Percent
Gender	Male	9984	43.83
	Female	12793	56.17
Age Group	60-74 years old	4915	21.58
	75-84 years old	4604	20.21
	85-94 years old	7369	32.35
	95-104 years old	5889	25.86
Residence	City	4205	18.46
	Town	4558	20.01
	Rural	14014	61.53
Years of Education	0 years	14196	62.33
	1-5 years	5209	22.87
	6-9 years	2341	10.28
	Over 10 years	1031	4.53
Family Economic Status	Low	3571	15.68
	Medium	15468	67.91
	High	3738	16.41
Timely Treatment	No	2129	9.35
	Yes	20648	90.65
Cognitive Function	Normal	14773	64.86
	Mild Impaired	4152	18.23
	Moderate Impaired	2628	11.54
	Severe Impaired	1224	5.37
ADL	No Disability	17935	78.74
	Disability	4842	21.26
IADL	No Disability	7678	33.71
	Mild Disability	5309	23.31
	Moderate Disability	4309	18.92
	Severe Disability	5481	24.06
Number of Chronic Diseases	No	9099	39.95
	1	7355	32.29
	2	3664	16.09
	3 and above	2659	11.67
Smoking	No	18425	80.89
	Yes	4352	19.11
Drinking	No	18144	79.66
	Yes	4633	20.34
Living Arrangements	Live without Children	8452	37.11
	Live with Children	14325	62.89
Marital Status	Married	7844	34.44
	Widowhood	14933	65.56
Loneliness	Never	8209	36.04
	Rarely	7186	31.55
	Sometimes	5602	24.60
	Often	1780	7.82

Analytical strategies

The dependent variable of interest is ordinal in nature; hence we employ the Mixed-effects ordered logistic regression strategy, which encompasses both fixed and random effects. Within the context of a longitudinal design, individuals are viewed as second-level units, while observations at each distinct time point are considered first-level units. A two-level random-intercept model is thus utilized. In constructing the model, we adopted a methodology that progressively increased in complexity. Initially, the model was estimated exclusively with control variables. Subsequently, the widowhood variable was incorporated into the model, followed by the inclusion of an interaction term between widowhood and gender. Then, a two-dimensional interaction term was added, accounting for the relationship between widowhood and age. Finally, a three-dimensional interaction term encompassing widowhood, gender, and age was added to investigate the potential interaction of these three variables and its potential significant impact on loneliness.

Results

The study’s results come from a multilevel mixed-effects ordinal logistic regression model. We used a likelihood ratio test to compare different models and see how including certain variables affects the models’ explanatory power. Likelihood ratio tests show the model fit significantly improves from Model 1 to Model 2 (ΔLL = 3.086, Δdf = 1, p < .05), Model 2 to Model 3 (ΔLL = 25.275, Δdf = 1, p < .001), and Model 3 to Model 4 (ΔLL = 4.406, Δdf = 2, p < .05). This suggests Model 4 is relatively strong.

Table 2 shows Widowhood has a significant positive effect on loneliness across all models (p < .001), indicating widowed individuals are more likely to report higher loneliness, all else equal. The Widowhood coefficient increases from Model 1 (b = 1.058, SE = 0.026) to Model 4 (b = 1.186, SE = 0.037), suggesting the widowhood-loneliness association strengthens when including interaction effects. The significant Widowhood × Female interaction term across Models 2–4 has a negative coefficient, meaning widowhood’s effect on loneliness is less pronounced for females. Specifically, the negative and significant Widowhood × Female coefficient in Model 4 (b = −0.208, SE = 0.052, p < .001) indicates widowhood increases loneliness, but this increase is less steep among females. The interaction term Widowhood × Age is significant and negative across Models 3 and 4, indicating that the effect of widowhood on loneliness decreases with age. In other words, widowhood has a great influence on the loneliness of the young elderly.

Table 2. Multilevel ordered logit models for the effects of widowhood on loneliness.

Variables	Model 1	Model 2	Model 3	Model 4
Female	−0.108***	−0.037	0.005	−0.005
	(0.025)	(0.038)	(0.039)	(0.041)
Age	−0.012***	−0.012***	−0.001	−0.001
	(0.001)	(0.001)	(0.002)	(0.002)
Town	0.479***	0.479***	0.480***	0.479***
	(0.031)	(0.031)	(0.031)	(0.031)
Rural	0.540***	0.539***	0.545***	0.542***
	(0.028)	(0.028)	(0.028)	(0.028)
1–5 years of education	−0.051	−0.054*	−0.057*	−0.057*
	(0.027)	(0.027)	(0.027)	(0.027)
6–9 years of education	−0.126***	−0.124***	−0.109**	−0.111**
	(0.036)	(0.036)	(0.036)	(0.036)
Over 10 years of education	−0.203***	−0.200***	−0.183***	−0.184***
	(0.053)	(0.053)	(0.053)	(0.053)
Medium Family Economic Status	−0.609***	−0.610***	−0.604***	−0.604***
	(0.029)	(0.029)	(0.029)	(0.029)
High Family Economic Status	−0.913***	−0.916***	−0.910***	−0.910***
	(0.036)	(0.036)	(0.036)	(0.036)
Timely Treatment	−0.562***	−0.562***	−0.565***	−0.564***
	(0.037)	(0.037)	(0.037)	(0.037)
Health Index	−0.012***	−0.012***	−0.012***	−0.012***
	(0.001)	(0.001)	(0.001)	(0.001)
Smoking	−0.086**	−0.083**	−0.077**	−0.082**
	(0.028)	(0.028)	(0.028)	(0.028)
Drinking	−0.175***	−0.173***	−0.165***	−0.166***
	(0.026)	(0.026)	(0.026)	(0.026)
Live with Children	−0.412***	−0.411***	−0.399***	−0.399***
	(0.022)	(0.022)	(0.022)	(0.022)
Widowhood	1.058***	1.105***	1.135***	1.186***
	(0.026)	(0.033)	(0.033)	(0.037)
Widowhood × Female		−0.109*	−0.144**	−0.208***
		(0.044)	(0.044)	(0.052)
Windowhood × Age			−0.016***	−0.023***
			(0.002)	(0.003)
Female × Age				−0.002
				(0.004)
Windowhood × Female × Age				0.011*
				(0.005)
Var(cons)	0.375***	0.373***	0.374***	0.373***
	(0.027)	(0.027)	(0.027)	(0.027)
Log likelihood	−53980.081	−53976.995	−53951.720	−53947.314
Degree of Freedom	15	16	17	19
Observations	45,139	45,139	45,139	45,139
Number of groups	22,777	22,777	22,777	22,777

Notes. Standard errors in parentheses; ***p < 0.001, ** p < 0.01, * p < 0.05.

The significant, positive Widowhood × Female × Age three-way interaction in Model 4 (b = 0.011, SE = 0.005, p < .05) indicates the mitigating effect of age on the widowhood-loneliness association is weaker for females. In other words, gender’s moderating influence on the widowhood-loneliness relationship diminishes with age. To better illustrate the interaction of gender, age, and widowhood, we calculated the probability of frequent loneliness among elderly individuals with different genders, ages, and marital statuses, as depicted in Figure 1. Figure 1 shows widowhood still significantly affects loneliness even at advanced ages, though its effect decreases with age without disappearing entirely. Before age 90, widowhood’s impact on loneliness substantially differs between genders, with a markedly stronger effect on men. Moreover, the gender gap is widest at young ages but gradually narrows with age. After age 90, there is no significant gender difference in widowhood’s effect on loneliness. These findings highlight the complex interplay between personal circumstances and demographic characteristics in shaping loneliness experiences.

Figure 1. The probability of frequent loneliness in the elderly with different genders, ages, and marital status.

In the preceding analysis, we did not account for potential differences based on time since widowhood. The CLHLS survey includes elderly individuals with highly variable widowhood durations, ranging from recent bereavement to widowhood lasting up to 85 years. The typical widowhood period among the elderly is around 19–22 years. Therefore, we further examined the impact of early widowhood on loneliness in the elderly. We restricted the sample to elderly individuals with a living spouse at baseline and used whether they were widowed during follow-up as the explanatory variable. For those widowed in a given survey year, subsequent observations were excluded.

Table 3 shows the model estimates for early widowhood’s impact on loneliness among the elderly with a living spouse at baseline. The results indicate loneliness still increases significantly in the initial period after spousal loss. This demonstrates widowhood affects elderly loneliness both immediately and over the long-term. Additionally, early widowhood’s effect on loneliness diminishes progressively with age, though no significant gender difference emerges, suggesting the gender differences in widowhood’s overall impact result from the cumulative effects of prolonged widowhood.

Table 3. Multilevel ordered logit models for the effects of widowhood on loneliness among the elderly widowed during the survey period.

Variables	Model 1	Model 2	Model 3	Model 4
Widowhood	1.092***	1.107***	1.209***	1.202***
	(0.052)	(0.072)	(0.076)	(0.081)
Widowhood × Female		−0.031	−0.128	−0.125
		(0.100)	(0.103)	(0.108)
Windowhood × Age			−0.025***	−0.022**
			(0.006)	(0.009)
Female × Age				0.002
				(0.004)
Windowhood × Female × Age				−0.006
				(0.012)
Var(_cons[id])	0.318***	0.318***	0.316***	0.316***
	(0.035)	(0.035)	(0.035)	(0.035)
Log likelihood	−21591.544	−21591.379	−21580.050	−21579.414
Degree of Freedom	15	16	17	19
Observations	19,553	19,553	19,553	19,553
Number of groups	7,788	7,788	7,788	7,788

Notes. Standard errors in parentheses; *** p < 0.001, ** p < 0.01, * p < 0.05; Due to space limitations, estimates of control variables are not given.

Figure 2 depicts the probability of frequent loneliness among elderly individuals with a living spouse at baseline during early widowhood. As in Figure 1, early widowhood’s effect on loneliness progressively declines with age for both genders, though remains significant even at age 95. Unlike Figure 1, the probability curves and confidence intervals for widowed men and women overlap at all ages, indicating no significant gender difference. This further demonstrates men and women experience similar impacts of widowhood in the initial bereavement period.

Figure 2. The probability of frequent loneliness among the elderly widowed during the survey period.

Discussion

This study aimed to elucidate the complex interplay between gender, age, and widowhood in shaping loneliness experiences among older adults in China. The findings reveal several key insights that advance our understanding of this issue. Consistent with previous research, widowhood emerged as a major predictor of loneliness (Aartsen & Jylhä, 2011; Carr et al., 2018; Dahlberg et al., 2015; Yang, 2021; Yang & Gu, 2021). Losing a spouse leaves an emotional void that tends to exacerbate isolation and disconnection. This loss of support and interaction can be particularly severe in the initial years after bereavement, when widowhood’s impact on loneliness peaks (Yang & Gu, 2021). This acute effect applies to both older men and women and endures even into very advanced ages. Our findings also indicate widowhood has a lasting influence on loneliness that persists regardless of age group or gender. This implies the loneliness experienced by widowed individuals is an enduring condition rather than a transient state.

Our findings reveal widowhood’s impact on loneliness differs by gender, with widowed men experiencing greater loneliness compared to widowed women. This gender difference aligns with some prior studies (Dahlberg et al., 2015; Nicolaisen & Thorsen, 2014) and may be attributed to several factors. Key among them are gender differences in social networks. Men, especially in older age, tend to have fewer social contacts and friends than women. This lack of a diverse social network can amplify the impact of spousal loss for men, rendering them more susceptible to loneliness (Menec et al., 2019). Women often maintain larger social networks with more friends, which can serve a “protective function” when they are widowed (Nicolaisen & Thorsen, 2014).

However, the relationship between gender and post-bereavement loneliness is complex. Our finding contrasts with some prior studies that found women tend to experience greater loneliness following spousal loss due to their higher widowhood rates (Carr et al., 2018; Lee et al., 2001; Olawa et al., 2021; Williams, 1988). Moreover, we found no significant gender difference in loneliness during early widowhood, indicating the disparity reflects widowhood’s long-term impacts. This conflicts with research suggesting men’s and women’s loneliness affects equalize five or more years after bereavement (Nicolaisen & Thorsen, 2014). Potential explanations include this study’s longer timeframe enabling cumulative effects in men and Chinese cultural norms discouraging emotional expression in women.

Van den Hoonaard (2013), through a comparative analysis of widowed men from retirement communities in rural Atlantic Canada and Florida, USA, with distinct cultural backgrounds, found that culture shapes two differing masculinities, with Florida widowed men, who value career and financial success, being more inclined to express loneliness following the loss of a spouse. There is research suggesting that the feeling of loneliness increases with the rise of individualism, and men are more susceptible to experiencing loneliness (Barreto et al., 2021). Other studies indicate that the sense of loneliness among Chinese elders is closely related to the cultural significance of intergenerational relationships (Chou & Chi, 2004). In Chinese culture, the emphasis on intergenerational relationships leads older adults to spend a significant amount of time helping with childcare and household chores, which hinders their ability to expand their social networks and makes them more prone to experiencing loneliness (Zhao et al., 2022). The traditional Confucian culture in China places great importance on filial piety. However, the long-standing one-child policy and the increasing influence of individualism make it difficult for modern Chinese elders to achieve expected levels of intergenerational reciprocity and familial care, further intensifying their feelings of loneliness (Ren et al., 2022; Zhao et al., 2022). This disparity in expectations may result in a heightened sense of loneliness, particularly for widowed men who have always relied on being cared for and holding a parental role. Further cross-cultural examinations and in-depth qualitative studies are needed to elucidate how gender nuances shape the widowhood-loneliness association across contexts.

Importantly, we found the gender difference in widowhood’s impact on loneliness diminishes with age and disappears after age 90. This aligns with research showing the odds of loneliness associated with widowhood decrease similarly for older women and men (Yang & Gu, 2021), indicating spousal loss’s effect becomes less dependent on gender in advanced years (Julsing et al., 2016). Another potential explanation is that greater loneliness in very old women stems from factors like declining health, shrinking social circles, and financial insecurity. As women outlive peers, they lose close confidants, eroding social support and worsening loneliness. In contrast, very old men may receive more family care and social support due to traditional gender norms, though spousal loss remains impactful given men’s reliance on wives for emotional intimacy. This gender convergence could reflect counterbalancing effects in very advanced ages. More qualitative research on the oldest old’s lived experiences could provide richer insight into this diminishing gender difference.

Consistent with some prior research, our findings importantly demonstrate widowhood’s impact on loneliness diminishes with advancing age (Menec et al., 2019; F. Yang & Gu, 2021). This pattern may stem from older adults’ enhanced capacity to cope with loneliness (TORNSTAM, 1992) and potential to adjust expectations regarding social support and relationships over time (Wei et al., 2021). Older seniors may be more mentally prepared and able to adapt to losing a spouse. Having experienced multiple losses during their lifetime could impart greater equanimity and improved coping skills and resilience when facing bereavement later in life (Revenson & Johnson, 1984). Moreover, older adults generally have had more time to cultivate social connections beyond marriage and establish independent living routines (Peplau & Perlman, 1982), which could buffer widowhood’s effects.

A key strength of this study is the use of national longitudinal data to examine the interaction of gender, age, and widowhood within the Chinese cultural context. The findings make a unique contribution by unraveling the complexity of how these factors combine to shape loneliness later in life. However, some limitations warrant mention. First, the single-item loneliness measure, though common, may lack nuance; incorporating multidimensional scales could enrich understanding. Second, the absence of data on marital relationship quality, coping styles, and other psychosocial factors constrains a deeper investigation of mechanisms driving the widowhood-loneliness association. Third, experiences of widowed older adults may differ across Chinese regions and rural/urban areas, which were not examined here.

The findings have several implications for policies and interventions to alleviate loneliness among older Chinese adults. Programs should target widowed older men, who tend to experience greater loneliness. However, very old widowed women also need attention. Potential interventions could involve peer support groups tailored to gender and age, counseling to improve coping skills, and increased healthcare and financial planning assistance for very old women. Raising children’s awareness about how spousal loss impacts parents’ wellbeing may help. Overall, a nuanced approach considering the intersections of age, gender, and widowhood is necessary to address the multifaceted needs of widowed seniors and mitigate loneliness.

Conclusion

This study aimed to unravel the complex interplay between gender, age, and widowhood in impacting loneliness among older Chinese adults. The findings uniquely reveal nuances in how these factors intersect to shape experiences of spousal loss and disconnection in later life. Key conclusions are:

1. Widowhood significantly influences loneliness across gender and age groups, though this effect lessens with age.

2. Men experience greater loneliness exacerbation after spousal loss compared to women, but this gender difference converges in very old age due to multidimensional factors.

3. The buffering effect of age on the widowhood-loneliness association is weaker for older women, likely due to constraints like health declines and caregiving limitations.

4. The cultural context plays a role in shaping widowhood and loneliness, necessitating further cross-cultural comparisons.

This study also has important policy and practical implications. Targeted interventions to support coping among widowed older men and very old widowed women could help alleviate isolation. Counseling, peer support, social engagement, and financial planning assistance may be beneficial.

Overall, this study makes key contributions on the intersectional nature of widowhood, gender, age, and loneliness within the Chinese cultural context. It advances knowledge to inform supportive policies and programs that can effectively address the multifaceted needs of widowed older adults.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The data used in this study come from the Chinese Longitudinal Healthy Longevity Survey (CLHLS), which is publicly available to researchers upon request. The CLHLS data access policy follows the Taylor & Francis share upon reasonable request principle. Researchers can obtain free access to the data by formal registration. Researchers who wish to access the CLHLS data files should submit a data usage agreement to the CLHLS team. More details about the data access application process can be found at [https://opendata.pku.edu.cn/dataverse/CHADS;jsessionid=943424e8b1b49c937c14468d19c8#]. The authors confirm they were granted access to the data used in this study by completing the CLHLS data usage agreement, which outlines ethical use of the data. The authors are unable to directly share the CLHLS data used, but interested researchers can apply to access the data via the contact details provided on the website The computer code used to generate results presented in this study is available from the corresponding author upon reasonable request.

Additional information

Funding

Funding is from the National Social Sciences Research Funds [grant number 19BRK013] for “the research on Health Transition and Health Needs of the Elderly”.