Skip to main content
NCBI home page
PsyCh Journal logoLink to PsyCh Journal
. 2026 Jul 26;15(4):e70115. doi: 10.1002/pchj.70115

Roles of Self‐Stigma and Meaning in the Relationship Between Age and Loneliness: A Perspective From Gerotranscendence

Han Xiao 1,2,3,4, Melina Malli 2,✉, Pamela Qualter 5, Manuela Barreto 6, Christina Victor 7, Theodore D Cosco 2,8, Guangyu Zhou 1,3,4,✉
PMCID: PMC13402103  PMID: 42503438

ABSTRACT

Drawing on the theory of gerotranscendence, this research aimed to gain a more comprehensive and nuanced understanding of the age‐loneliness relationship and underlying factors. Data from the BBC Loneliness Experiment (n = 6875, 67.80% female, M age = 49.26, Rangeage 16–94) were analyzed using polynomial regression and Multi‐Group Structural Equation Modeling (MG‐SEM) for the quantitative data, and complemented by Latent Dirichlet Allocation (LDA) for free‐text data. The results showed: (1) there was a non‐linear, inverted U‐shaped relationship between age and loneliness, with an inflection point at 39.2 years of age; (2) in the older age group, loneliness self‐stigma and the meaning of loneliness played parallel mediating roles between age and loneliness, whereas no mediating effect was observed in the younger age group; (3) these mediating effects primarily attenuated the loneliness frequency and loneliness intensity rather than loneliness duration; (4) two overarching topics of loneliness (i.e., Limited Social Functions and Unsatisfactory Social Relationships) were identified, with differences between the two age groups. This study further enriches the understanding of gerotranscendence in the loneliness domain and provides suggestions for tailoring precision interventions (i.e., self‐stigma, meaning‐making) to distinct life stages and specific loneliness dimensions.

Keywords: age, gerotranscendence, loneliness, loneliness self‐stigma, meaning of loneliness

1. Introduction

Loneliness may occur when subjective feelings of pain or unhappiness are caused by the discrepancy between an individual's actual social relationships and ideal social relationships (de Jong‐Gierveld et al. 2018; Lim et al. 2023). Nowadays, loneliness is becoming an important public health issue (Badcock et al. 2022; Lim et al. 2020), which is seriously damaging people's social functions and leading to cognitive impairment (Huang et al. 2023; Tan et al. 2020). Although loneliness can emerge at any age (Mund et al. 2020), existing literature regarding the relationship between age and loneliness remains notably inconsistent. Some studies propose a linear decline in loneliness as individuals age (Suanet et al. 2024); others report evidence of a linear increase (Hülür et al. 2016), and several suggest complex non‐linear patterns (Hawkley et al. 2022; Victor and Yang 2012). Accordingly, it is warranted to examine the precise age‐loneliness relationship pattern and to identify the inflection age point (if any) in this relationship within a large sample. Crucially, while statistical trends have been widely debated, the underlying psychological mechanisms and nuanced differences in this relationship should be further elucidated.

The theory of gerotranscendence conceptualizes aging as a process of development and a transformation in meta‐perspective (Tornstam 1997), which includes three dimensions: the cosmic (a sense of connection with existence and the universe rather than a sense of isolation), the self (a decrease in self‐centeredness and enhanced self‐integration), and social and personal relationships (a reduced need for superficial interactions and increased need for deep bonds). In the context of loneliness, one potential consequence of these dimensions is their influence on how older adults cognitively cope with loneliness: reducing loneliness self‐stigma and enhancing meaning‐making. For instance, older adults may adopt a perspective that normalizes loneliness as a universal experience, thereby reducing self‐focus and loneliness self‐stigma (i.e., the cosmic and the self dimensions) (Tornstam 2011). Simultaneously, the shift in the need for social relationships and self‐integration (i.e., the self and the social and personal relationships dimensions) enables older adults to reappraise loneliness not as a deficit, but as a meaningful space for freedom and existential clarity (Baltes and Dickson 2001; Morlett Paredes et al. 2021).

Therefore, the current study aims to determine whether a non‐linear age‐loneliness relationship exists, identify its precise inflection point, and uncover the psychological mechanisms (i.e., self‐stigma and meaning) underlying the association between age and loneliness.

1.1. Loneliness Self‐Stigma

Self‐stigma is the incorporation of a negative stereotype into one's identity (Bos et al. 2013; Criswell 2025). Loneliness is often stigmatized, with lonely individuals being attributed more negative traits—such as being less adjusted, less competent, and less likable—compared to those who are not lonely (Department for Culture, Media and Sport 2023; Kerr and Stanley 2021). Research indicates that stigma can intensify feelings of loneliness, creating barriers to forming social connections (Pitman et al. 2018). However, gerotranscendence suggests that older adults demonstrate a decrease in self‐centeredness and ego‐preoccupation, coupled with greater self‐integration (Tornstam 2011). Specifically, older adults are less susceptible to external judgments and more attuned to their internal feelings, resulting in diminished feelings of shame or embarrassment when experiencing loneliness. Research demonstrates that disengaging from negative social evaluations (e.g., feeling shame and embarrassment about loneliness) and facilitating authentic emotional expression in later life are highly beneficial for health and well‐being (Milligan et al. 2016). We propose that this age‐related cognitive change elucidates why older adults internalize the stigma of loneliness to a lesser degree than younger adults, buffering them against loneliness.

Additionally, the widespread belief that loneliness is particularly prevalent among older adults (Gardiner et al. 2020; Rapolienė and Aartsen 2022) may normalize the experience for this age group, reducing its stigmatizing impact. Conversely, younger adults are usually framed as highly social, rendering loneliness less normative and potentially more stigmatizing for them (Malli et al. 2023; Pikhartova et al. 2016). Driven by elevated loneliness self‐stigma, people often actively conceal their condition to avoid negative social judgments (Mental Health Foundation 2022), which is a maladaptive coping strategy that impedes help‐seeking and ultimately aggravates their experience of loneliness (Kantar Public 2016; Ko et al. 2022).

Taken together, these ideas lead us to expect that loneliness self‐stigma plays a mediating role in the relationship between age and loneliness.

1.2. Meaning of Loneliness

Meaning‐making involves construing an experience as comprehensible and significant (Martela and Steger 2016; Park 2010). Existing research and the wider media overwhelmingly portray the meaning of loneliness as negative (Ferguson 2018; Prohaska et al. 2020). However, emerging empirical evidence challenges this uniformly negative stereotype by revealing that loneliness can also be defined variously (Switsers et al. 2023; Victor et al. 2019): when regarded as the painful experience of loss, exclusion and sadness, it is appraised as negative; when regarded as enabling personal creativity, growth, and freedom, it is appraised as positive (de Jong‐Gierveld et al. 2018; Wotherspoon 2024).

Establishing new connections and maintaining social relationships are important to younger adults (Maes et al. 2019). Conversely, the narrowing of social connections (e.g., due to retirement, bereavement, and illness) is seen as an inevitable part of growing older (Cohen‐Mansfield and Eisner 2020; Vedder et al. 2022). From a gerotranscendent perspective (Tornstam 1997, 2011), older adults exhibit a greater capacity to acknowledge and accept loneliness, cognitively appraising and conceptualizing it as a meaningful experience that is inherent to the purpose of life (Mansfield et al. 2021).

The selective optimization with compensation theory (Baltes and Baltes 1990) also suggests that people may employ selective optimization to cope with the losses associated with aging. In response to perceived threats, such as shrinking social networks, older adults might seek to alleviate lonely feelings through two strategies of meaning‐making: reaffirming meaning in areas where life is threatened (e.g., reflecting on the meaning of loneliness) and redirecting their attention to other aspects (e.g., finding positive meaning in loneliness). The latter strategy, known as compensation (Baltes and Dickson 2001; Morlett Paredes et al. 2021), helps older adults protect the comprehensibility and significance of their existence, while also reducing loneliness (Borawski et al. 2022).

Therefore, this study hypothesizes that the meaning of loneliness (i.e., how positively it is appraised) functions as a mediating factor between age and loneliness.

1.3. Loneliness Dimensions

In addition to understanding the effects of the factors mentioned above, it is necessary to comprehensively examine and carefully distinguish the different dimensions of loneliness (Weiss 1975). Loneliness should be viewed from the same perspective as other emotion‐focused experiences (such as depression) with frequency (e.g., how often loneliness happens: never, very often), intensity (e.g., what is the level of distress the loneliness causes: severe, moderate, mild), and duration (e.g., how long loneliness lasts: a day, a week, a month, a year) (Qualter et al. 2021).

Therefore, one auxiliary aim of the current study is to explore if and how these mechanisms differ across the loneliness frequency, intensity, and duration.

1.4. Textual Semantic Topics of Loneliness

Moreover, to deepen the interpretation of the quantitative findings, another auxiliary aim of the current study is to explore the textual semantic topics, which further capture the nuanced phenomenology of the loneliness experience within the framework of gerotranscendence. This study incorporated Latent Dirichlet Allocation (LDA) topic modeling. LDA is a statistical model that leverages an unsupervised machine learning approach to discover the underlying semantic structure in textual data (Blei et al. 2003; Hagg et al. 2022). The latent semantic structure consists of a set of related topics identified by co‐occurring words, rather than relying on predefined topical assumptions.

In the context of the current study, textual semantic analysis could facilitate the understanding of age‐related differences in loneliness, enriching our investigation into the loneliness self‐stigma, the meaning of loneliness, and the dimensions of loneliness.

1.5. Overview of the Current Study

Collectively, by integrating the theoretical frameworks of gerotranscendence, stigmatization, and meaning‐making, this study aims to investigate:

Hypothesis 1

Age exhibits a non‐linear relationship with loneliness, characterized by a precise inflection point.

Hypothesis 2

The loneliness self‐stigma plays a mediating role in the relationship between age and loneliness.

Hypothesis 3

The meaning of loneliness plays a mediating role in the relationship between age and loneliness.

Hypothesis 4 (exploratory)

The magnitude of the mediating effects (i.e., loneliness self‐stigma and meaning of loneliness) varies across the three loneliness dimensions (i.e., frequency, intensity, and duration) (H4a). When the sample is divided at the identified inflection point, the characteristics of textual semantic topics relating to loneliness are expected to differ between age groups.

The conceptual framework of the current study is shown in Figure 1.

FIGURE 1.

FIGURE 1

Open in a new tab

Conceptual framework of the current study. Dashed boxes denote the foundational theoretical constructs derived from theory of gerotranscendence (Tornstam 1997, 2011). Solid boxes denote the key variables operationalized within the context of the current study (i.e., age, loneliness self‐stigma, meaning of loneliness, loneliness frequency, loneliness intensity, and loneliness duration).

2. Methods

2.1. Participants

Participants were from a large‐scale global study—the British Broadcasting Corporation (BBC) Loneliness Experiment. The data were collected between February and May 2018. The survey received ethical approval from the University of Manchester (Reference No. 2017‐2710‐4594). The ethical guidelines of the British Psychological Society and the 2013 Declaration of Helsinki were followed for this survey. The online survey was launched through BBC Radio 4 and the BBC World Service and covered elsewhere on radio, TV, and social media, extending the coverage of the study.

An a priori power analysis was conducted using a structural equation model (SEM) sample size calculator (Soper 2025) to determine the sample size needed for testing the parallel mediation model. To detect a medium‐sized mediating effect with a desired power of 0.80 and a significance level of 0.05 (Cohen 1988; Westland 2010), a minimum of 2900 participants was required. In the current study, we analyzed data from 6875 participants (16–94 years old, M = 49.26, SD = 15.38; 32.20% male; 71.60% residing in the UK) who provided data for the variables of interest in this study from the BBC Loneliness Experiment. The detailed characteristics of participants are shown in Supporting Information S1. Participants took approximately 40 min to complete the survey. Information was collected about age, loneliness (including loneliness frequency, intensity, and duration), loneliness self‐stigma, and the meaning of loneliness.

2.2. Measures

2.2.1. Loneliness

Loneliness was measured using the 4‐item UCLA Loneliness Scale (Russell et al. 1980): “Do you feel a lack of companionship?” “Do you feel left out?” “Do you feel isolated from others?” and “Do you feel in tune with people around you?” (reverse‐coded). We adapted the response options by asking about frequency (“How often does that happen?”, from 1 = never to 5 = very often), intensity (“How intense is that feeling?”, from 1 = not at all to 5 = very intense), and duration (“How long does that feeling last when it occurs?”, 1 = hours, 2 = days, 3 = weeks, 4 = months, 5 = longer) of loneliness (Qualter et al. 2021). Each subscale was averaged across the four items to obtain total scores for these three loneliness dimensions. With higher scores indicated greater loneliness. Qualter et al. (2021) obtained reliable Cronbach's α for frequency, intensity, and duration = 0.85, 0.86, and 0.88 respectively. In this study, the measures were reliable: Cronbach's α for frequency, intensity, and duration = 0.83, 0.82, and 0.86, respectively (and for all items = 0.95). For more details about justifying the decomposition of loneliness into three dimensions (frequency, intensity, and duration) in the current sample, please refer to Supporting Information S2.

Additionally, participants answered a free‐text question to express their attitudes and experiences of loneliness on a voluntary basis, and no word limit was imposed on the length of their response. Among the 6875 participants in this study, 5682 of them (32.20% male) answered the question. The MANOVA showed that between the participants who answered the questions and those who did not, there were no significant differences in terms of loneliness (including loneliness frequency, intensity, and duration; Fs (1, 6874) ranged from 0.11 to 2.70, ps > 0.10, partial η 2 < 0.001); there were no significant differences in terms of SES‐related indicators (including subjective social status, years of education, and employment status; Fs (1, 6874) ranged from 0.52 to 0.55, ps > 0.46, partial η 2 < 0.001). A detailed description of SES‐related indicators is provided in the Covariates section.

2.2.2. Loneliness Self‐Stigma

Loneliness self‐stigma was measured with three items: “When I feel lonely, I feel ashamed about it,” “When I feel lonely, I am too embarrassed to admit it to others”, and “When I feel lonely, I don't talk to others about it.” The measurement obtained a reliable Cronbach's α = 0.80 (Barreto et al. 2022). Participants responded to each item on a 7‐point Likert‐type scale (from 1 = strongly disagree to 7 = strongly agree). Item‐level responses for these scales served as the observable variable and were synthesized to create a continuous composite index. Higher scores are indicative of greater loneliness self‐stigma. In this study, the measure was reliable (Cronbach's α = 0.80).

2.2.3. Meaning of Loneliness

We ascertained the meaning of loneliness using a multiple‐choice question, “What does loneliness mean to you?”, which was developed specifically for the BBC Loneliness Experiment. Participants were required to choose as many items as they thought applied. It contained 10 items, including 3 positive items (e.g., loneliness means “Spending time focusing on yourself”, “Being on your own”) and 7 negative items (e.g., loneliness means “Sadness”, “Feeling vulnerable”). The answer for each item was coded as 0 = not selected and 1 = selected. To capture the overall valence of meaning ascribed to loneliness, negative items were reverse‐coded, and a sum score was calculated across all items. Higher scores indicate a greater degree of positive meaning of loneliness. This aggregative scoring approach for mixed‐valence checklists aligns with the similar practices in psychometric research (Nunnally and Bernstein 1994; Yesavage et al. 1982; Zuckerman et al. 1983). The confirmatory factor analysis yielded the following fit indices: χ 2/df = 26.26, CFI = 0.90, TLI = 0.83, SRMR = 0.07, and RMSEA = 0.06.

2.2.4. Covariates

According to the previous evidence, gender (dummy: 0 = female, and 1 = male), socioeconomic status (SES), solitude (actively seeking and enjoying being alone), and living circumstances (whether living alone) are associated with loneliness (Switsers et al. 2023; Weinstein et al. 2023); thus, the current study included these factors as covariates.

Subjective social status, income, education, and employment are important indices of SES (Ganzeboom et al. 1992; Hoebel and Lampert 2020). Participants were asked to indicate their status within their country using the 10‐rung ladder (MacArthur Subjective Social Status measure, MSSS) (Adler et al. 2000). The subjective evaluation of how well their financial needs were met by their income was measured using a 3‐point Likert‐scale (1 = poorly, 2 = fairly well, 3 = very well) (Pfeiffer 1975). Education was measured in years (continuous), and employment status was dichotomized as inactive (coded as 0 = not active in the labor market, including retired, student, and unemployed) or active (coded as 1 = active in the labor market, including full‐time work and part‐time work) (Pan et al. 2023). The above indices were standardized separately and summed to generate an overall indicator of participants' SES (continuous).

Solitude was measured using 6 items on a 3‐point Likert‐type scale (e.g., “To think something over without distraction, I want to be alone”), and scores were summed after reversing the scoring of the reverse‐scored items. Higher scores are indicative of greater solitude (Barreto et al. 2021). To our knowledge, no prior studies have reported on the psychometric properties of the solitude scale. In the current study, the Cronbach's α of the solitude scale was 0.72. Living circumstances were dichotomized by the question “Do you live alone?” (0 = no, 1 = yes).

2.3. Data Analysis

The overall rate of missing data in the current sample was 0.70%. The full‐information maximum likelihood (FIML) approach was used to handle missing data robustly, which could mitigate estimation bias stemming from nonnormal distribution and is highly effective for datasets with missing data well below the 30% threshold (Graham 2009). Statistical analyses were conducted using IBM SPSS Statistics (version 29) and RStudio (version 2024.4.2.764) for quantitative data, and in Python (version 3.11.4) for textual data. The primary analytical codes of the current study are available on OSF (https://osf.io/ftxc2/overview).

2.3.1. Non‐Linear Relationship and Inflection Point Test

First, a preliminary descriptive analysis was conducted to calculate sums and/or means, standard deviations, and correlations among the variables. This enabled us to characterize the loneliness experience of our study population. Then, polynomial regression modeling (comparing linear age, quadratic age 2 , and cubic age 3 functional forms) was employed to explicitly test for non‐linear patterns in the relationship between age and loneliness and to identify the inflection age point in this relationship.

2.3.2. Structural Equation Modeling

To further test the gerotranscendence‐based hypotheses, the sample was subsequently stratified into different life‐stage groups based on the inflection age point. Multi‐Group Structural Equation Modeling (MG‐SEM) was utilized to evaluate the parallel mediating roles of loneliness self‐stigma and meaning of loneliness between age and loneliness frequency, intensity, and duration across the groups. To test whether the robustness of the parameter estimates was maintained across groups, measurement invariance was examined before MG‐SEM. Model specification and estimation path analyses were conducted using the lavaan package in RStudio.

In the model specification, age, the dimensions of loneliness, meaning of loneliness, and the covariates were entered as observed variables, and the loneliness self‐stigma was entered as a latent variable measured by 3 items (i.e., ashamed, embarrassed, and do not talk). Considering the scaling disparities among variables with different original metrics (e.g., chronological age in years, binary dummy codes, and aggregated Likert means), the variables were transformed into standardized Z‐scores before estimation (Kline 2023). To conservatively account for any potential deviations from multivariate normality, all models were estimated using the Maximum Likelihood Robust Estimator (MLR) (Rhemtulla et al. 2012). To generate bias‐corrected confidence intervals (MacKinnon 2008), the bootstrap samples were set as 5000. Where the 95% confidence intervals (CI) did not include zero, it indicated that the mediating effect was significant (Preacher and Hayes 2008). Additionally, cutoff values for fit indices were employed, including the χ 2/df ratio, the Tucker‐Lewis index (TLI), the comparative fit index (CFI), the root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). To prevent biased estimates due to outcome redundancy, the residual covariances among loneliness frequency, intensity, and duration were freely estimated across all structural models. Moreover, to examine the relative magitude of the mediation across the three loneliness dimensions, the absolute values of mediating effects for loneliness frequency, intensity, and duration were compared (Cheung 2007; Liu et al. 2024). To evaluate between‐group differences in the mediating effects, the custom contrasts were specified to calculate the exact effect size differences between groups with a 5000‐resample bootstrapping procedure (see Supporting Information S3). Owing to the large sample size, a more rigorous criterion of p < 0.01 (rather than p < 0.05) was utilized. Interpretations emphasized standardized effect sizes to ensure that the reported effects are not only statistically significant but also theoretically and practically meaningful.

Finally, to ensure the robustness of the findings against the demographic characteristics of the current sample, four additional analyses were conducted. First, an additional 2‐group MG‐SEM was tested, controlling for macro‐level cultural variance by assigning each participant a score on Hofstede's Individualism Index (IDV, 0–100 scale; Hofstede 1997) based on their country of residence. Second, an additional 4‐group MG‐SEM (stratified by both inflection point and gender) was tested for gender invariance. Third, to validate the results from a standard lifespan‐developmental perspective, an additional MG‐SEM was conducted to test the robustness of the primary findings in theoretically defined age cohorts (i.e., emerging adulthood, early‐middle adulthood, later‐middle adulthood, and later adulthood) (Arnett 2000; Morbey et al. 2025; Srivastava et al. 2003). Fourth, the competing sequential (chain) mediation models were evaluated and compared with the primary parallel mediation models. Full details of additional analyses are available in Supporting Information S4.

2.3.3. Latent Dirichlet Allocation

Gensim and NLTK libraries were mainly used for textual semantic analysis. Participants who failed to provide valid open‐ended responses (coded as −99) were excluded from the corpus. The raw text corpus (total word count = 155,343) underwent a series of preprocessing steps to facilitate subsequent computational analysis. Preprocessing is a natural‐language‐processing (NLP) workflow to clean and structure raw text, making it suitable for analysis techniques like Latent Dirichlet Allocation (LDA) topic modeling (Maier et al. 2021). The preprocessing pipeline was implemented using Natural Language Toolkit package (NLTK) (Bird et al. 2009). First, following the identification of the inflection age point, the textual dataset was partitioned. Second, the text was then tokenized into individual words, with corrections made for spelling and spacing. Third, the remaining words underwent lemmatization, a process that standardizes different versions of a word to its base form, thereby improving the coherence and accuracy of the resulting topic model. Fourth, to ensure the analysis focused on words with substantive meaning, hyperlinks, special characters, articles, conjunctions, colloquial language (e.g., “kinda,” “yeah”), and non‐lexical strings (e.g., characters composed only of consonants or vowels) were removed using a custom and standard English stop‐word dictionary (Bird et al. 2019). Subsequently, the model was initialized with a fixed random seed (random_state = 100) and trained over 10 passes, utilizing auto alpha to optimize the topic‐word distributions.

Determining the optimal number of topics relied on both quantitative coherence metrics and qualitative theoretical evaluation, with a topic count (k) ranging from one to six. This selection process was guided by quantitative model indices (by Gensim), including Perplexity, Topic Coherence, and Normalized Pointwise Mutual Information (NPMI) (Hagg et al. 2022; Kapadia 2019). Specifically, Perplexity denotes the difference between the expected value according to the model and the observed value, topic coherence measures the semantic similarity of words within a topic, and NPMI assesses word co‐occurrence. An optimal model demonstrates low perplexity, high topic coherence, and high NPMI values. While the utilization of quantitative coherence metrics narrowed the candidate models, existing research in topic modeling suggests that purely quantitative metrics alone do not always produce interpretable models (Chang et al. 2009). Therefore, to ensure the interpretability and plausibility of the final model and to avoid fragmented and redundant subtopics, the selection was confirmed through a qualitative review considering the metrics and semantic overlap. Finally, the visualization of the output was produced by pyLDAvis (Mabey 2020).

3. Results

3.1. Inverted U‐Shaped Relationship Between Age and Loneliness

Harman's single‐factor test was used to examine common method bias (Aguirre‐Urreta and Hu 2019; Lindell and Whitney 2001). The results of the unrotated factor analysis showed that there were 9 factors with eigenvalues greater than 1. The first principal factor accounted for 24.22% (< 40%) of the variance, indicating that the common method bias was not serious in this study (Zhou and Long 2004). In addition, there was no multicollinearity among those tested variables (Tolerance > 0.20, VIF < 5) (Shrestha 2020), and further analyses were conducted. Table 1 displays the correlations for key variables. Age was negatively correlated with loneliness (r frequency = −0.08, r intensity = −0.11, and r duration = −0.05, ps < 0.001) and loneliness self‐stigma (rs < −0.08, ps < 0.001), and positively correlated with positive meaning of loneliness (r = 0.28, p < 0.001).

TABLE 1.

Correlations for key variables.

M SD 1 2 3 4 5 6 7 8
1. Age 49.26 15.38 —
2. Loneliness frequency 2.83 1.09 −0.08*** —
3. Loneliness intensity 2.85 1.07 −0.11*** 0.89*** —
4. Loneliness duration 3.10 1.39 −0.05*** 0.87*** 0.83*** —
5. Stigma_ashamed 3.98 1.94 −0.21*** 0.30*** 0.33*** 0.28*** —
6. Stigma_embarrassed 4.71 1.88 −0.15*** 0.32*** 0.32*** 0.29*** 0.64*** —
7. Stigma_do not talk 5.10 1.81 −0.08*** 0.29*** 0.27*** 0.26*** 0.38*** 0.66*** —
8. Meaning of loneliness 3.04 2.00 0.28*** −0.22*** −0.26*** −0.21*** −0.28*** −0.22*** −0.12*** —
Open in a new tab

Note: N = 6875.

***

p < 0.001.

The results of the relationship test between age and loneliness are shown in Table 2. The linear relationship revealed that loneliness decreased with age. Moreover, a significant inverted U‐shaped relationship between age and loneliness was identified. Specifically, loneliness peaks at an inflection point at the age of 39.2, after which it begins a decline. The non‐linear inverted U‐shaped relationship is shown in Figure 2.

TABLE 2.

Variance explained by linear, quadratic, and cubic models.

β t p R 2 ∆R 2 ∆F p (∆)
Linear −0.007 −9.27 < 0.001 0.18
Non‐linear Quadratic −0.0004 −8.46 < 0.001 0.19 0.01 71.53 < 0.001
Cubic 0.000 −1.60 0.110 0.20 0.0003 2.55 0.110
Open in a new tab

Note: ∆ refers to quadratic index subtracts linear index or cubic index subtracts quadratic index. Adjusting for gender, SES, living alone, and solitude.

FIGURE 2.

FIGURE 2

Open in a new tab

Inverted U‐shaped relationship between age and loneliness.

3.2. Parallel Mediation Analysis

Based on the identified inflection point (age = 39.2), the whole sample was divided into two life‐stage groups: younger age group (<= 39.2 years, n = 1922) and older age group (> 39.2 years, n = 4953). The parallel mediation model revealed a satisfactory fit to the data in both groups (younger age group: χ 2/df = 8.81, p < 0.001, CFI = 0.98, TLI = 0.94, SRMR = 0.06, RMSEA = 0.03; older age group: χ 2/df = 8.81, p < 0.001, CFI = 0.99, TLI = 0.98, SRMR = 0.04, RMSEA = 0.01). The constrained metric invariance model showed no significant decrement in model fit compared to the configural invariance model (∆χ 2 = 1.34, p = 0.52) between both age groups, ensuring the constructs were operationalized consistently.

After adjusting for gender, SES, living alone, and solitude, the results of the parallel mediation effect analysis for the two groups are shown in Figure 3 and Table 3. The results indicated the age‐specificity of the mediating mechanisms: neither loneliness self‐stigma nor meaning of loneliness demonstrated parallel mediating effects in the younger age group. In contrast, the loneliness self‐stigma and meaning of loneliness exhibited significant and substantial mediating effects in the older age group. Specifically, in the younger age group, the R 2 for loneliness frequency, intensity, and duration is 0.25, 0.24, and 0.20, respectively. The parallel mediating pathways of loneliness self‐stigma and meaning of loneliness accounted for 16.67% and 17.65% of the total model effect, respectively. In the older age group, the R 2 for loneliness frequency, intensity, and duration is 0.33, 0.35, and 0.28, respectively. The parallel mediating pathways of loneliness self‐stigma and meaning of loneliness accounted for 54.48% and 21.01% of the total model effect, respectively.

FIGURE 3.

FIGURE 3

Open in a new tab

The parallel mediation models: (a) is the parallel mediation model of the younger age group, (b) is the parallel mediation model of the older age group. Adjusting for gender, SES, living alone, and solitude. The value outside the parentheses is standard regression coefficient, the value inside the parentheses is standard error. The straight line represents a significant path, and the dashed line represents an insignificant path. ***p < 0.001, **p < 0.01.

TABLE 3.

Parallel mediating effects of loneliness self‐stigma and meaning of loneliness.

Group Pathway β SE p Boot LLCI Boot ULCI
Younger age group Indirect effects
Age → Loneliness self‐stigma → Loneliness frequency 0.006 0.019 0.762 −0.032 0.043
→ Loneliness intensity 0.006 0.019 0.761 −0.031 0.043
→ Loneliness duration 0.005 0.017 0.762 −0.028 0.038
Total indirect effect of loneliness self‐stigma 0.017 0.055 0.762 −0.091 0.124
Age → Meaning of loneliness → Loneliness frequency −0.006 0.005 0.266 −0.015 0.004
→ Loneliness intensity −0.007 0.006 0.247 −0.019 0.005
→ Loneliness duration −0.005 0.005 0.268 −0.014 0.004
Total indirect effect of meaning of loneliness −0.018 0.016 0.256 −0.048 0.013
Total effects of the model
−0.102 0.145 0.483 −0.385 0.182
Older age group Indirect effects
Age → Loneliness self‐stigma → Loneliness frequency −0.134 0.008 < 0.001 −0.148 −0.119
→ Loneliness intensity −0.136 0.008 < 0.001 −0.151 −0.120
→ Loneliness duration −0.120 0.007 < 0.001 −0.134 −0.106
Total indirect effect of loneliness self‐stigma −0.389 0.022 < 0.001 −0.431 −0.347
Age → Meaning of loneliness → Loneliness frequency −0.048 0.005 < 0.001 −0.058 −0.037
→ Loneliness intensity −0.057 0.006 < 0.001 −0.068 −0.046
→ Loneliness duration −0.046 0.006 < 0.001 −0.057 −0.035
Total indirect effect of meaning of loneliness −0.150 0.016 < 0.001 −0.181 −0.120
Total effects of the model
−0.714 0.058 < 0.001 −0.828 −0.600
Open in a new tab

Note: Adjusting for gender, SES, living alone, and solitude.

Subsequently, the mediating effects comparison for three dimensions of loneliness (i.e., frequency, intensity, and duration) was further conducted within the older age group. The forest plot (see Figure 4) demonstrated the absolute differences in the standardized indirect effects (|∆β|). The mediating effect of loneliness self‐stigma in the relationship between age and loneliness duration was weaker than that in the relationship between age and loneliness frequency, and weaker than that in the relationship between age and loneliness intensity. Additionally, the mediating effect of the meaning of loneliness in the relationship between age and loneliness intensity was stronger than that in the relationship between age and loneliness frequency, and stronger than that in the relationship between age and loneliness duration.

FIGURE 4.

FIGURE 4

Open in a new tab

The absolute differences in the mediating effects across the dimensions of loneliness within the older age group. |∆ 𝛽| is the absolute difference in the mediating effect. Adjusting for gender, SES, living alone, and solitude.

Moreover, the contrast analyses indicated that the mediating effects were significantly stronger in the older age group compared to the younger age group. The results for all six contrast tests (i.e., the group differences for the mediating effects of loneliness self‐stigma between age and loneliness frequency, intensity, and duration, separately; the group differences for the mediating effects of meaning of loneliness between age and loneliness frequency, intensity, and duration, separately) are shown in Supporting Information S3.

3.3. Robustness and Sensitivity Analyses

Additional analyses indicated the robustness of the primary findings across demographic variations (i.e., cultural individualism, gender, and developmental stage) and competing sequential mediation models. First, after introducing IDV as a covariate into the model, structural paths and indirect effects remained consistent with the model without adjusting the IDV. Second, when the sample was stratified into a 4‐group MG‐SEM to account for gender (Younger Males, Younger Females, Older Males, and Older Females), the magnitude and direction of the mediational pathways remained consistent, indicating no significant gender differences in the underlying gerotranscendent mechanisms. Third, consistent with the primary findings, the older groups (i.e., later‐middle adulthood and later adulthood) demonstrated parallel mediating effects, whereas the younger groups (i.e., emerging adulthood and early‐middle adulthood) exhibited no significant mediating effects through the loneliness self‐stigma or the meaning of loneliness. These findings conceptually corroborated the data‐driven inflection point in the primary findings and aligned with the theory of gerotranscendence, indicating that the capacity to derive positive meaning from loneliness and experience lower self‐stigma to buffer loneliness is a distinct developmental transformation to later life stages. Fourth, the sequential mediating models failed to reach the meaningful magnitudes of the indirect effects and to meet the generally accepted cutoff criteria for adequate model fit. The results of the additional analyses are shown in Supporting Information S4.

3.4. Latent Dirichlet Allocation

LDA model fit indices for different number of topics (k, ranging from one to six) for both the younger age group and the older age group is shown in Supporting Information S5. The Topic Coherence value reached its maximum at k = 2, and there was no semantic overlap when k = 2. Therefore, a two‐topic solution was selected as the optimal model for both the younger age group and the older age group. The visualization of the inter‐topic distances for the two groups was shown in Figure 5, where the area of each circle is proportional to the topic's prevalence in the corpus. The most representative words for each topic and the distribution probabilities of the words for each topic are listed in Supporting Information S5.

FIGURE 5.

FIGURE 5

Open in a new tab

Inter‐topic distance map in the LDA model: (a) is the inter‐topic distance map of the younger age group, (b) is the inter‐topic distance map of the older age group. The default for computing inter‐topic distances is Jensen‐Shannon divergence, and for scaling the set of inter‐topic distances defaults to principal components (see Sievert and Shirley 2014). The areas of the circles are proportional to the relative prevalence of the topic in the corpus.

Although both groups manifested two shared themes, subtle yet crucial differences emerged in their specific semantic content. Topic 1 is Limited Social Functions (e.g., “Not being able to talk to people around me,” “Having nobody to turn to for a chat, meet up with, to confide in, share experiences with”). The younger age group predominantly focused on this topic, accounting for 79.90% of the younger age group's corpus, such as the lack of talk, sharing, and companionship. In contrast, the older age group showed a relatively lower focus on general social deficits (accounting for 53.30% of the older age group's corpus). While the older age group also perceived the lack of talking and sharing, they uniquely emphasized a lack of opportunities of a deep exchange of ideas. Topic 2 is Unsatisfactory Social Relationships (e.g., “I have friends but no best friend,” “Loneliness is the feeling of your heart jumping out of your chest when you worry about being alone without a partner or family”). In this topic, the older age group exhibited relatively less self‐centeredness compared to the younger age group. Instead, the older age group's narratives centered on their closest, most meaningful ties (e.g., spouses, children, long‐term friends). Moreover, the affection associated with loneliness was different between the two groups. The younger age group predominantly expressed anxiety (a high‐arousal negative emotion), and the older age group felt a sense of sadness (a low‐arousal negative emotion).

4. Discussion

Drawing upon the theory of gerotranscendence and the frameworks of stigma and meaning‐making (Bos et al. 2013; Martela and Steger 2016; Tornstam 2011), the current study employed polynomial regression and MG‐SEM to analyze the quantitative scale data, complemented by LDA to explore the textual data. This dual analytical approach was adopted to gain a more comprehensive and nuanced understanding of the age‐loneliness relationship and underlying factors.

4.1. Age and Loneliness

The current study advances the understanding of loneliness across the lifespan by empirically confirming an inverted U‐shaped relationship between age and loneliness. Specifically, a precise inflection point at 39.2 years of age was identified, indicating that loneliness steadily increases in the younger age group, peaks in midlife, and subsequently demonstrates a gradual decline into later life.

This non‐linear relationship provides nuanced support for age‐related emotional resilience (Charles 2010; Scheibe and Carstensen 2010). As the theory of gerotranscendence suggests, transcendence is a natural process that comes with aging (Tornstam 1997). For instance, social networks tend to narrow with age (Hajek and König 2020; Kim et al. 2023), and the quality of emotionally close social partners is more important for older individuals (Xing et al. 2017). As a result, older individuals may redefine social relationships and decrease the need for superficial social contact (Tornstam 1997), thus reducing the experience of loneliness. However, before the inflection point (in the younger age group), individuals often face intense social pressures regarding social network expansion, exacerbating the experience of loneliness (Qualter et al. 2015).

In summary, this finding calls for reconsidering the conventional perspective of the association between age and loneliness. It highlights that loneliness is not unique to older individuals and suggests that the finding of lower loneliness in older agewas observed in the current sample.

4.2. The Mediating Role of the Loneliness Self‐Stigma

Consistent with the hypothesis, loneliness self‐stigma was found to mediate the relationship between age and loneliness. Specifically, the older age group demonstrated lower loneliness self‐stigma and further experienced lower loneliness frequency, intensity, and duration. Corroborating the results of the mediation, younger participants expressed feelings of shame and embarrassment of loneliness in the free‐text question. For instance, they reported “Feeling ashamed of being lonely and uncomfortable admitting to my friends that I am lonely” (20 years old) and described “Being embarrassed of going alone” (36 years old).

In light of the theory of gerotranscendence (Tornstam 1997, 2011), the older age group demonstrated a significant decrease in self‐centeredness alongside an increase in self‐integration. This psychological maturation renders them less vulnerable to the perceived judgments of others, effectively buffering them against the self‐stigma associated with loneliness. Another possible explanation is that public awareness of loneliness is greatest for older individuals (Pitman et al. 2018), with much of the scientific research and social campaigns on loneliness being focused on older individuals (Chawla et al. 2021; Gerst‐Emerson and Jayawardhana 2015). Thus, loneliness is widely perceived as a more normative and socially acceptable experience in later life, thereby diminishing the stigmatization (Barreto et al. 2022). However, although the current study did not find a significant pathway from age to loneliness self‐stigma in the younger age group, previous studies suggested that younger individuals attach great importance to building and maintaining social relationships; they are more vigilant about personal evaluations from society and are particularly sensitive to accepting and associating themselves with stereotypes pertinent to loneliness (Tsai and Reis 2009; Zheng et al. 2023). Moreover, higher self‐stigma further increases loneliness (Qin et al. 2024). Because they fear embarrassment and of negative social evaluation, individuals experiencing higher loneliness self‐stigma suppress their emotions and avoid disclosing their feelings (Mental Health Foundation 2022). This reluctance to share can lead to greater emotional isolation and exacerbate loneliness (Ko et al. 2022; Prizeman et al. 2023).

4.3. The Mediating Role of the Meaning of Loneliness

Consistent with the hypothesis, the meaning of loneliness played a mediating role between age and loneliness. Specifically, the older age group assigned more positive than negative meanings to loneliness and experienced lower loneliness. The free‐text responses also revealed that some older participants framed loneliness with a more positive perspective. For example, they noted that it is “nice to spend ‘alone’ time” (74 years old) and stated “I am used to my own company and have a hobby” (70 years old).

According to the theory of gerotranscendence (Tornstam 1997, 2011), older individuals tend to reduce superficial connections, reconstructing cognitive strategies to reframe negative interpretations of loneliness (Carstensen and Reynolds 2023). This includes reaffirming meaning in areas where one feels vulnerable, such as reflecting on the negative meaning of loneliness (e.g., loss, being left out, and sadness), and redirecting focus toward discovering other potential aspects. For instance, older individuals consider loneliness as commonplace and focus on its benefits (Kharicha et al. 2021), such as freedom and self‐growth (Kvaal et al. 2014). This may enable older individuals to “open the door” of loneliness (Kitzmüller et al. 2018) and, thus, experience less loneliness. On the contrary, previous evidence indicates that younger individuals attach great importance to social networks (Moore and Leung 2002). When there is a gap between actual and ideal social relationships, younger individuals may ascribe the circumstance to undesirable meanings that ultimately link to loneliness.

4.4. The Three Dimensions of Loneliness in the Mediation Model

The finding suggests that loneliness is a multidimensional construct, including frequency, intensity, and duration. In addition to aligning with the two overarching topics (discussed in the next section), participants' free‐text answers also revealed nuanced dimensions of the loneliness experience. These included: loneliness frequency, with descriptions such as “Sometimes the expectation of company can be disappointing” (37 years old) and “I feel cut off from society sometimes” (42 years old); loneliness intensity, with statements ranging from “I still get very lonely” (25 years old) to “I am quite lonely, but not very lonely” (57 years old); loneliness duration, with participants mentioning specific timeframes like “Not speaking to anyone all day” (27 years old) or “Spending a lot of time alone” (27 years old).

Comparing the mediating effects among different dimensions of loneliness provided critical and unique information that may be obscured or confounded in previous literature. We found that the mediating effect of loneliness self‐stigma in the relationship between age and loneliness frequency/intensity was stronger than in the relationship between age and loneliness duration; additionally, the mediating effect of meaning of loneliness in the relationship between age and loneliness intensity was stronger than in the relationship between age and loneliness duration. These findings further imply that the gerotranscendent processes (e.g., diminished self‐stigma and enhanced positive meaning‐making in the context of the current study) protect older individuals against loneliness frequency and intensity, rather than attenuating its prolonged or chronic duration. One potential explanation is related to negative emotional sensitivity (Sauer‐Zavala et al. 2012). Cognitive factors (e.g., self‐stigma and negative meaning‐making) may heighten an individual's sensitivity to feelings of loneliness, with even brief moments of isolation potentially magnified into more frequent and intensive experiences of loneliness. These findings highlight the necessity of delineating loneliness into distinct dimensions, thereby guiding future research and enabling the design of tailored interventions for older individuals. For instance, strategies designed with a gerotranscendent perspective may prove efficacious in ameliorating frequent and severe loneliness.

4.5. The Textual Semantic Topics Towards Loneliness

The textual semantic topics towards loneliness are consistent across the younger and older age groups, while there are still discrepancies between the two age groups. First, although the two groups both experience the Limited Social Functions (e.g., the lack of talk and sharing with other people), the younger age group expressed more experiences in this topic compared with the older age group, and the older age group uniquely emphasized a lack of opportunities to exchange inner thoughts with others. Second, in terms of Unsatisfactory Social Relationships, older individuals tended to focus more on closer relationships. This finding was consistent with gerotranscendence (Tornstam 1997) and socioemotional selectivity theory (Carstensen et al. 1999): aging is accompanied by a decreased centrality on the self and a deliberate withdrawal from superficial relationships, favoring instead highly meaningful connections where deep ideas can be explored and understood. Third, the negative affective vocabulary accompanying loneliness shifted across the younger age group and the older age group. Specifically, high‐arousal negative emotion (e.g., anxiety) was predominantly expressed by the younger age group, while low‐arousal negative emotion (e.g., sadness) was expressed more by the older age group. This finding aligns with people's developmental stage. Social deficits threaten younger individuals' urgent need to build social capital, thereby triggering an anxious response. Conversely, the older individuals are not anxiously striving to build new social relationships; rather, the sadness reflects a mature, reflective acceptance of the irreversible loss of irreplaceable and deep relational figures (Charles 2010; Scheibe and Carstensen 2010).

Furthermore, the textual semantic findings in the LDA provided the phenomenological context to enrich the interpretation of the psychological mediators in MG‐SEM. On the one hand, younger individuals' heightened self‐focus and hypervigilance regarding social relationships increase vulnerability to social threat and loneliness stigma (Hawkley and Cacioppo 2010; Osborn et al. 2021). In contrast, the older individuals demonstrated a shift away from self‐focus and superficial validation (Tornstam 2005). By reducing the need to maintain a broad social network, older individuals escape from the social pressures and normative expectations. On the other hand, the younger age group expressed loneliness more as a deficit in social functioning, whereas the older age group particularly described it as the absence of profound intellectual and relational depth. By framing loneliness in a way that honors the value of irreplaceable, close social ties (Carstensen et al. 1999; Charles 2010) rather than an anxious experience, the older individuals transform it into a more meaningful existential state (Tornstam 2005).

4.6. Limitations and Future Directions

Several limitations should be addressed in future research. First, self‐reported data were inherently susceptible to the influence of social desirability bias. Although a preliminary Harman's single‐factor test indicated that common method bias was not a serious problem in the current study, this diagnostic tool cannot definitively rule out or isolate shared method variance (Podsakoff et al. 2003). Consequently, the magnitude of the observed structural associations should be interpreted with the understanding that relying exclusively on self‐reports may still partially inflate some relationships. Integrating neurophysiological indicators (Franco‐O'Byrne et al. 2023) and more advanced diagnostic tools for detecting methodological biases (Podsakoff et al. 2024) could further enhance the interpretability and generalizability of the findings.

Second, the reliance on an online survey introduces self‐selection biases. On the one hand, people with better physical and mental health, higher educational attainment, greater digital literacy, and a pre‐existing interest in loneliness‐related topics were more likely to participate in the study. Accordingly, these selection processes may systematically exclude the most vulnerable or technologically disadvantaged older individuals. The demographic composition was also predominantly female and concentrated in the United Kingdom. Although additional analyses (after considering the gender and cultural background) yielded results consistent with the primary findings, these imbalances limited the generalizability of the conclusions. On the other hand, while the age‐stratified MG‐SEM based on a non‐linear relationship yielded theoretical insights into developmental transformations, the group sample sizes were unequal in the current study. Although the absolute sample size within each derived subgroup remained relatively large for adequate statistical power in SEM (Kline 2023), the demographic imbalance is noted. Future research utilizing nationally representative and balanced sampling for age, gender, and cultural background is recommended to rigorously cross‐validate and extend the mechanisms identified in this study.

Third, conceptual complexities are inherent in measuring multidimensional psychological constructs (e.g., different subdimensions of meaning and loneliness). For instance, as with many checklist‐style inventories, summing scores of the meaning of loneliness (i.e., summing positive and reverse‐coded negative items) may simplify the construct. Future research should aim to unpack this complexity by utilizing person‐centered methodologies, such as Latent Profile Analysis (LPA) and Latent Transition Analysis (LTA), to identify distinct profiles and how these profiles shift across the lifespan.

Fourth, the cross‐sectional nature of the data precludes further definitive conclusions regarding causality or intra‐individual developmental change. It is difficult to isolate pure developmental maturation (i.e., getting older) from the unique generational environments in which different cohorts were socialized. Longitudinal panel designs and experience sampling methodology (Costa and McCrae 1982; Myin‐Germeys et al. 2018) should be employed to track individuals' stigma and meaning‐making towards loneliness as they age.

Lastly, while LDA provides an unsupervised machine learning approach for extracting semantic topics from large‐scale text, future research should complement LDA with in‐depth qualitative interviews (Sholokhova et al. 2022). The phenomenological approaches are needed to compare and understand the nuanced experiences of loneliness‐related stigmatization and meaning‐making between younger and older age groups, informing the development of more targeted, age‐specific psychological interventions.

5. Conclusion and Implications

This study further enriched the theory of gerotranscendence in the loneliness domain with the perspective of stigma and meaning‐making. First, the findings empirically confirmed a non‐linear, inverted U‐shaped relationship between age and loneliness, identifying a precise inflection at 39.2 years of age. Second, we found that the older age group experienced lower overall loneliness due to two parallel gerotranscendent mechanisms: a reduction in loneliness self‐stigma and an enhanced capacity for positive meaning‐making. Third, the above mediations demonstrated dimensional specificity for loneliness, which primarily attenuated the frequency and acute intensity of loneliness rather than its chronic duration. Fourth, corroborated by the textual semantic topics extracted via LDA, the study further captured the nuanced, age‐specific variations in the lived experience of loneliness.

These findings suggest potential practical interventions for addressing loneliness. While increasing social interaction and providing social skills training have been shown to alleviate loneliness (Pearce et al. 2021), our results indicate it may be beneficial to target internal psychological factors. Mental health professionals should incorporate stigma‐reduction and meaning‐making approaches into therapeutic frameworks to help individuals re‐frame loneliness in a more constructive light (Rodriguez et al. 2025). Public campaigns and media narratives should aim to destigmatize loneliness by reframing it as a shared human experience with opportunities for personal growth and connection. In addition, we should note people's preconceived notions about a group (e.g., older individuals are bound to be lonely, while younger individuals are bound to be sociable) and alleviate stigmatization. Collectively, these findings emphasize the value of understanding loneliness through a gerotranscendent lens that considers psychological mechanisms, individual experiences, and social attitudes. Together the findings of self‐stigma, meaning‐making and age related differences in loneliness may inform the development of more effective interventions tailored to different age groups.

Funding

Wellcome Trust [Funder reference: 209625/Z/17/Z] funded the BBC Loneliness Experiment, awarded to Pamela Qualter, Manuela Barreto, and Christina Victor. This paper was also supported by the National Natural Science Foundation of China [Funder reference: 32271136] to Guangyu Zhou.

Ethics Statement

This study received ethical approval from the University of Manchester (Reference No.: 2017‐2710‐4594), and participants gave their informed consent before participating in this study.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Characteristics of participants.

Table S2: Between‐group differences of the mediating effects.

Table S3: The model fit of the 2‐group MG‐SEM controlling for IDV.

Table S4: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 1 (younger adults, <= 39.2).

Table S5: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 2 (older adults, > 39.2).

Table S6: The model fit of the 4‐group MG‐SEM.

Table S7: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 1 (younger female, <= 39.2).

Table S8: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 2 (younger male, <= 39.2).

Table S9: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 3 (older female, > 39.2).

Table S10: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 4 (older male, > 39.2).

Table S11: The model fit of the 4‐group MG‐SEM.

Table S12: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 1 (emerging adulthood, <= 25 years).

Table S13: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 2 (early‐middle adulthood, 25~44 years).

Table S14: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 3 (later‐middle adulthood, 45~64 years).

Table S15: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 4 (later adulthood, > = 65 years).

Table S16: The model fit and mediating effects of competing sequential (chain) mediation models.

Table S17: The LDA model index of topic number in the younger age group (<= 39.2 years old).

Table S18: The LDA model index of topic number in the older age group (> 39.2 years old).

Figure S1: Top‐20 most relevant words (terms) for each topic of the younger age group (<= 39.2 years old).

Figure S2: Top‐20 most relevant words (terms) for each topic of the older age group (> 39.2 years old).

Table S19: The words and their distribution probabilities for each topic.

PCHJ-15-e70115-s001.docx (313.2KB, docx)

Acknowledgements

This research was the result of a partnership between academics and All in the Mind on the British Broadcasting Corporation Radio 4. The authors thank Claudia Hammond and Geraldine Fitzgerald for support throughout the project, and Alice Eccles and Matt Richins for help with the preparation of materials. Han Xiao would like to express appreciation for the support provided by the Institute of Population Research and Office of International Relations at Peking University for supporting her study programme at the University of Oxford.

Contributor Information

Melina Malli, Email: melina.malli@ageing.ox.ac.uk.

Guangyu Zhou, Email: gyzhou@pku.edu.cn.

Data Availability Statement

Access to the dataset is available from the authors upon reasonable request.

References

  1. Adler, N. E. , Epel E. S., Castellazzo G., and Ickovics J. R.. 2000. “Relationship of Subjective and Objective Social Status With Psychological and Physiological Functioning: Preliminary Data in Healthy, White Women.” Health Psychology 19, no. 6: 586–592. 10.1037/0278-6133.19.6.586. [DOI] [PubMed] [Google Scholar]
  2. Aguirre‐Urreta, M. I. , and Hu J.. 2019. “Detecting Common Method Bias: Performance of the Harman's Single‐Factor Test.” ACM SIGMIS Database: The Database for Advances in Information Systems 50, no. 2: 45–70. 10.1145/3330472.3330477. [DOI] [Google Scholar]
  3. Arnett, J. J. 2000. “Emerging Adulthood: A Theory of Development From the Late Teens Through the Twenties.” American Psychologist 55, no. 5: 469–480. 10.1037/0003-066X.55.5.469. [DOI] [PubMed] [Google Scholar]
  4. Badcock, J. C. , Holt‐Lunstad J., Garcia E., Bombaci P., and Lim M. H.. 2022. Position Statements on Addressing Social Isolation, Loneliness, and the Power of Human Connection. Global Initiative on Loneliness and Connection. https://www.gilc.global/_files/ugd/410bdf_74fffc2d18984b0e8217288b1b12d199.pdf. [Google Scholar]
  5. Baltes, B. B. , and Dickson M. W.. 2001. “Using Life‐Span Models in Industrial Organizational Psychology: The Theory of Selective Optimization With Compensation.” Applied Developmental Science 5: 51–62. 10.1207/S1532480XADS0501_5. [DOI] [Google Scholar]
  6. Baltes, P. B. , and Baltes M. M.. 1990. “Psychological Perspectives on Successful Aging: The Model of Selective Optimization With Compensation.” In Successful Aging: Perspectives From the Behavioral Sciences, 1–34. Cambridge University Press. [Google Scholar]
  7. Barreto, M. , Victor C., Hammond C., Eccles A., Richins M. T., and Qualter P.. 2021. “Loneliness Around the World: Age, Gender, and Cultural Differences in Loneliness.” Personality and Individual Differences 169: 110066. 10.1016/j.paid.2020.110066. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Barreto, M. , van Breen J., Victor C., et al. 2022. “Exploring the Nature and Variation of the Stigma Associated With Loneliness.” Journal of Social and Personal Relationships 39, no. 9: 2658–2679. 10.1177/02654075221087190. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Bird, S. , Klein E., and Loper E.. 2019. Accessing Text Corpora and Lexical Resources. Natural Language Toolkit. Retrieved from. https://www.nltk.org/book/ch02.html. [Google Scholar]
  10. Bird, S. , Edward L., and Ewan K.. 2009. Natural Language Processing With Python. O'Reilly Media Inc. [Google Scholar]
  11. Blei, D. M. , Ng A. Y., and Jordan M. I.. 2003. “Latent Dirichlet Allocation.” Journal of Machine Learning Research 3, no. Jan: 993–1022. [Google Scholar]
  12. Borawski, D. , Nowak A., and Zakrzewska A.. 2022. “Lonely Meaning–Seekers: The Moderating Role of Search for Meaning in the Relationship Between Loneliness and Presence of Meaning.” Personality and Individual Differences 190: 111550. 10.1016/j.paid.2022.111550. [DOI] [Google Scholar]
  13. Bos, A. E. , Pryor J. B., Reeder G. D., and Stutterheim S. E.. 2013. “Stigma: Advances in Theory and Research.” Basic and Applied Social Psychology 35, no. 1: 1–9. 10.1080/01973533.2012.746147. [DOI] [Google Scholar]
  14. Carstensen, L. L. , and Reynolds M. E.. 2023. “Age Differences in Preferences Through the Lens of Socioemotional Selectivity Theory.” Journal of the Economics of Ageing 24: 100440. 10.1016/j.jeoa.2022.100440. [DOI] [Google Scholar]
  15. Carstensen, L. L. , Isaacowitz D. M., and Charles S. T.. 1999. “Taking Time Seriously: A Theory of Socioemotional Selectivity.” American Psychologist 54, no. 3: 165–181. 10.1037/0003-066X.54.3.165. [DOI] [PubMed] [Google Scholar]
  16. Chang, J. , Gerrish S., Wang C., Boyd‐Graber J. L., and Blei D. M.. 2009. “Reading Tea Leaves: How Humans Interpret Topic Models.” Advances in Neural Information Processing Systems (NIPS) 22: 288–296 Retrieved from. https://proceedings.neurips.cc/paper_files/paper/2009/file/f92586a25bb3145facd64ab20fd554ff‐Paper.pdf. [Google Scholar]
  17. Charles, S. T. 2010. “Strength and Vulnerability Integration: A Model of Emotional Well‐Being Across Adulthood.” Psychological Bulletin 136, no. 6: 1068–1091. 10.1037/a0021232. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Chawla, K. , Kunonga T. P., Stow D., Barker R., Craig D., and Hanratty B.. 2021. “Prevalence of Loneliness Amongst Older People in High‐Income Countries: A Systematic Review and Meta‐Analysis.” PLoS One 16, no. 7: e0255088. 10.1371/journal.pone.0255088. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Cheung, M. W. 2007. “Comparison of Approaches to Constructing Confidence Intervals for Mediating Effects Using Structural Equation Models.” Structural Equation Modeling: A Multidisciplinary Journal 14, no. 2: 227–246. 10.1080/10705510709336745. [DOI] [Google Scholar]
  20. Cohen‐Mansfield, J. , and Eisner R.. 2020. “The Meanings of Loneliness for Older Persons.” Aging & Mental Health 24, no. 4: 564–574. 10.1080/13607863.2019.1571019. [DOI] [PubMed] [Google Scholar]
  21. Cohen, J. 1988. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Lawrence Earlbaum Associates. [Google Scholar]
  22. Costa, P. T. , and McCrae R. R.. 1982. “An Approach to the Attribution of Aging, Period, and Cohort Effects.” Psychological Bulletin 92, no. 1: 238–250. 10.1037/0033-2909.92.1.238. [DOI] [Google Scholar]
  23. Criswell, K. R. 2025. “Effects of Internalized Stigma on Quality of Life and Stress in Undergraduate Students With Chronic Health Conditions.” Stigma and Health 10, no. 4: 586–598. 10.1037/sah0000539. [DOI] [Google Scholar]
  24. de Jong‐Gierveld, J. , Van Tilburg T. G., and Dykstra P. A.. 2018. “New Ways of Theorizing and Conducting Research in the Field of Loneliness and Social Isolation.” In The Cambridge Handbook of Personal Relationships, edited by Vangelisti A. L. and Perlman D., 391–404. Cambridge University Press. 10.1017/9781316417867.031. [DOI] [Google Scholar]
  25. Department for Culture, Media and Sport . 2023. “Loneliness Stigma Rapid Evidence Assessment (REA).” Available Online. https://www.gov.uk/government/publications/research‐exploring‐the‐stigma‐associated‐with‐loneliness/loneliness‐stigma‐rapid‐evidence‐assessment‐rea#impacts‐of‐loneliness‐stigma‐1.
  26. Ferguson, M. 2018. “How Does It Really Feel to Be Lonely?” The Economist. Retrieved from. https://www.1843magazine.com/features/how‐does‐it‐really‐feel‐to‐be‐lonely.
  27. Franco‐O'Byrne, D. , Sepúlveda J. P. M., Gonzalez‐Gomez R., et al. 2023. “The Neurocognitive Impact of Loneliness and Social Networks on Social Adaptation.” Scientific Reports 13, no. 1: 12048. 10.1038/s41598-023-38244-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Ganzeboom, H. B. , De Graaf P. M., and Treiman D. J.. 1992. “A Standard International Socio‐Economic Index of Occupational Status.” Social Science Research 21, no. 1: 1–56. 10.1016/0049-089X(92)90017-B. [DOI] [Google Scholar]
  29. Gardiner, C. , Laud P., Heaton T., and Gott M.. 2020. “What Is the Prevalence of Loneliness Amongst Older People Living in Residential and Nursing Care Homes? A Systematic Review and Meta‐Analysis.” Age and Ageing 49, no. 5: 748–757. 10.1093/ageing/afaa049. [DOI] [PubMed] [Google Scholar]
  30. Gerst‐Emerson, K. , and Jayawardhana J.. 2015. “Loneliness as a Public Health Issue: The Impact of Loneliness on Health Care Utilization Among Older Adults.” American Journal of Public Health 105, no. 5: 1013–1019. 10.2105/AJPH.2014.302427. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Graham, J. W. 2009. “Missing Data Analysis: Making It Work in the Real World.” Annual Review of Psychology 60: 549–576. 10.1146/annurev.psych.58.110405.085530. [DOI] [PubMed] [Google Scholar]
  32. Hagg, L. J. , Merkouris S. S., O'Dea G. A., et al. 2022. “Examining Analytic Practices in Latent Dirichlet Allocation Within Psychological Science: Scoping Review.” Journal of Medical Internet Research 24, no. 11: e33166. https://preprints.jmir.org/preprint/33166. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Hajek, A. , and König H. H.. 2020. “Which Factors Contribute to Loneliness Among Older Europeans? Findings From the Survey of Health, Ageing and Retirement in Europe: Determinants of Loneliness.” Archives of Gerontology and Geriatrics 89: 104080. 10.1093/geronb/gbaa044. [DOI] [PubMed] [Google Scholar]
  34. Hawkley, L. C. , and Cacioppo J. T.. 2010. “Loneliness Matters: A Theoretical and Empirical Review of Consequences and Mechanisms.” Annals of Behavioral Medicine 40, no. 2: 218–227. 10.1007/s12160-010-9210-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Hawkley, L. C. , Buecker S., Kaiser T., and Luhmann M.. 2022. “Loneliness From Young Adulthood to Old Age: Explaining Age Differences in Loneliness.” International Journal of Behavioral Development 46, no. 1: 39–49. 10.1177/0165025420971048. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Hoebel, J. , and Lampert T.. 2020. “Subjective Social Status and Health: Multidisciplinary Explanations and Methodological Challenges.” Journal of Health Psychology 25, no. 2: 173–185. 10.1177/1359105318800804. [DOI] [PubMed] [Google Scholar]
  37. Hofstede, G. 1997. Cultures and Organizations: Software of the Mind. McGraw‐Hill. [Google Scholar]
  38. Huang, Y. , Zhu X., Liu X., and Li J.. 2023. “The Effects of Loneliness, Social Isolation, and Associated Gender Differences on the Risk of Developing Cognitive Impairment for Chinese Oldest Old.” Aging & Mental Health 27, no. 7: 1360–1367. 10.1080/13607863.2022.2116396. [DOI] [PubMed] [Google Scholar]
  39. Hülür, G. , Drewelies J., Eibich P., et al. 2016. “Cohort Differences in Psychosocial Function Over 20 Years: Current Older Adults Feel Less Lonely and Less Dependent on External Circumstances.” Gerontology 62, no. 3: 354–361. 10.1159/000438991. [DOI] [PubMed] [Google Scholar]
  40. Kantar Public . 2016. Trapped in a Bubble: An Investigation Into Triggers for Loneliness in the UK. Co‐Operative/British Red Cross; Available Online. https://assets.ctfassets.net/5ywmq66472jr/5tKumBSlO0suKwiWO6KmaM/230366b0171541781a0cd98fa80fdc6e/Coop_Trapped_in_a_bubble_report.pdf. [Google Scholar]
  41. Kapadia, S. 2019. “Evaluate Topic Models: Latent Dirichlet Allocation (LDA).” Towards Data Science Available from. https://towardsdatascience.com/evaluate‐topic‐model‐in‐python‐latent‐dirichlet‐allocation‐lda‐7d57484bb5d0/. [Google Scholar]
  42. Kerr, N. A. , and Stanley T. B.. 2021. “Revisiting the Social Stigma of Loneliness.” Personality and Individual Differences 171: 110482. 10.1016/j.paid.2020.110482. [DOI] [Google Scholar]
  43. Kharicha, K. , Manthorpe J., Iliffe S., et al. 2021. “Managing Loneliness: A Qualitative Study of Older People's Views.” Aging & Mental Health 25, no. 7: 1206–1213. 10.1080/13607863.2020.1729337. [DOI] [PubMed] [Google Scholar]
  44. Kim, Y. M. , Jang S. N., and Cho S. I.. 2023. “Working Hours, Social Engagement, and Depressive Symptoms: An Extended Work‐Life Balance for Older Adults.” BMC Public Health 23, no. 1: 2442. 10.1186/s12889-023-17072-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Kitzmüller, G. , Clancy A., Vaismoradi M., Wegener C., and Bondas T.. 2018. ““Trapped in an Empty Waiting Room”—The Existential Human Core of Loneliness in Old Age: A Meta‐Synthesis.” Qualitative Health Research 28, no. 2: 213–230. 10.1177/1049732317735079. [DOI] [PubMed] [Google Scholar]
  46. Kline, R. B. 2023. Principles and Practice of Structural Equation Modeling. Guilford Publications. [Google Scholar]
  47. Ko, S. Y. , Wei M., Rivas J., and Tucker J. R.. 2022. “Reliability and Validity of Scores on a Measure of Stigma of Loneliness.” Counseling Psychologist 50, no. 1: 96–122. 10.1177/00110000211048000. [DOI] [Google Scholar]
  48. Kvaal, K. , Halding A. G., and Kvigne K.. 2014. “Social Provision and Loneliness Among Older People Suffering From Chronic Physical Illness. A Mixed‐Methods Approach.” Scandinavian Journal of Caring Sciences 28, no. 1: 104–111. 10.1111/scs.12041. [DOI] [PubMed] [Google Scholar]
  49. Lim, M. , Qualter P., Ding D., Holt‐Lunstad J., Mikton C., and Smith B.. 2023. “Advancing Loneliness and Social Isolation as Global Health Challenges: Taking Three Priority Actions.” Public Health Research and Practice 33, no. 3: e3332320. 10.17061/phrp3332320. [DOI] [PubMed] [Google Scholar]
  50. Lim, M. , Eres R., and Vasan S.. 2020. “Understanding Loneliness in the Twenty‐First Century: An Update on Correlates, Risk Factors, and Potential Solutions.” Social Psychiatry and Psychiatric Epidemiology 55: 793–810. 10.1007/s00127-020-01889-7. [DOI] [PubMed] [Google Scholar]
  51. Lindell, M. K. , and Whitney D. J.. 2001. “Accounting for Common Method Variance in Cross‐Sectional Research Designs.” Journal of Applied Psychology 86, no. 1: 114–121. 10.1037/0021-9010.86.1.114. [DOI] [PubMed] [Google Scholar]
  52. Liu, H. , Chen X., Yao Y., He W., Zhang Q., and Zeng H.. 2024. “Social Support and Depressive Symptoms Among Migrant Elderly Following Children: The Chain Mediating Role of Resilience and Loneliness.” BMC Psychiatry 24: 898. 10.1186/s12888-024-06349-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Mabey, B. 2020. “pyLDAvis.” Avalible from. https://github.com/bmabey/pyLDAvis.
  54. MacKinnon, D. P. 2008. Introduction to Statistical Mediation Analysis. Erlbaum. [Google Scholar]
  55. Maes, M. , Nelemans S. A., Danneel S., et al. 2019. “Loneliness and Social Anxiety Across Childhood and Adolescence: Multilevel Meta‐Analyses of Cross‐Sectional and Longitudinal Associations.” Developmental Psychology 55, no. 7: 1548–1565. 10.1037/dev0000719. [DOI] [PubMed] [Google Scholar]
  56. Maier, D. , Waldherr A., Miltner P., et al. 2021. “Applying LDA Topic Modeling in Communication Research: Toward a Valid and Reliable Methodology.” In Computational Methods for Communication Science, 13–38. Routledge. [Google Scholar]
  57. Malli, M. A. , Ryan S., Maddison J., and Kharicha K.. 2023. “Experiences and Meaning of Loneliness Beyond Age and Group Identity.” Sociology of Health & Illness 45, no. 1: 70–89. 10.1111/1467-9566.13539. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Mansfield, L. , Victor C., Meads C., et al. 2021. “A Conceptual Review of Loneliness in Adults: Qualitative Evidence Synthesis.” International Journal of Environmental Research and Public Health 18, no. 21: 11522. 10.3390/ijerph182111522. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Martela, F. , and Steger M. F.. 2016. “The Three Meanings of Meaning in Life: Distinguishing Coherence, Purpose, and Significance.” Journal of Positive Psychology 11: 531–545. 10.1080/17439760.2015.1137623. [DOI] [Google Scholar]
  60. Mental Health Foundation . 2022. “Loneliness and Mental Health Report, UK.” Available online. https://www.mentalhealth.org.uk/our‐work/research/loneliness‐and‐mental‐health‐report‐uk.
  61. Milligan, C. , Bingley A., and Gatrell A.. 2016. “‘Healing and Feeling’: The Place of Emotions in Later Life.” In Emotional Geographies, 49–62. Routledge. [Google Scholar]
  62. Moore, S. , and Leung C.. 2002. “Young People's Romantic Attachment Styles and Their Associations With Well‐Being.” Journal of Adolescence 25, no. 2: 243–255. 10.1006/jado.2002.0464. [DOI] [PubMed] [Google Scholar]
  63. Morbey, R. A. , Todkill D., Hughes H. E., Charlett A., and Elliot A. J.. 2025. “Evaluation of Age Groupings Used for Syndromic Surveillance.” medRxiv, 2025–01. 10.1101/2025.01.10.25320339. [DOI]
  64. Morlett Paredes, A. , Lee E. E., Chik L., et al. 2021. “Qualitative Study of Loneliness in a Senior Housing Community: The Importance of Wisdom and Other Coping Strategies.” Aging & Mental Health 25: 559–566. 10.1080/13607863.2019.1699022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Mund, M. , Freuding M. M., Möbius K., Horn N., and Neyer F. J.. 2020. “The Stability and Change of Loneliness Across the Life Span: A Meta‐Analysis of Longitudinal Studies.” Personality and Social Psychology Review 24, no. 1: 24–52. 10.1177/1088868319850738. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Myin‐Germeys, I. , Kasanova Z., Vaessen T., et al. 2018. “Experience Sampling Methodology in Mental Health Research: New Insights and Technical Developments.” World Psychiatry 17, no. 2: 123–132. 10.1002/wps.20513. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Nunnally, J. , and Bernstein I.. 1994. Psychometric Theory. 3rd ed. MacGraw‐Hill. [Google Scholar]
  68. Osborn, T. , Weatherburn P., and French R. S.. 2021. “Interventions to Address Loneliness and Social Isolation in Young People: A Systematic Review of the Evidence on Acceptability and Effectiveness.” Journal of Adolescence 93: 53–79. 10.1016/j.adolescence.2021.09.007. [DOI] [PubMed] [Google Scholar]
  69. Pan, H. , Qualter P., Barreto M., Stegen H., and Dury S.. 2023. “Loneliness in Older Migrants: Exploring the Role of Cultural Differences in Their Loneliness Experience.” International Journal of Environmental Research and Public Health 20, no. 4: 2785. 10.3390/ijerph20042785. [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Park, C. L. 2010. “Making Sense of the Meaning Literature: An Integrative Review of Meaning Making and Its Effects on Adjustment to Stressful Life Events.” Psychological Bulletin 136: 257–301. 10.1037/a0018301. [DOI] [PubMed] [Google Scholar]
  71. Pearce, E. , Myles‐Hooton P., Johnson S., et al. 2021. “Loneliness as an Active Ingredient in Preventing or Alleviating Youth Anxiety and Depression: A Critical Interpretative Synthesis Incorporating Principles From Rapid Realist Reviews.” Translational Psychiatry 11: 628. 10.1038/s41398-021-01740-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  72. Pfeiffer, E. 1975. Multidimensional Functional Assessment, the OARS Methodology: A Manual. Center for the Study of Aging and Human Development, Duke University. [Google Scholar]
  73. Pikhartova, J. , Bowling A., and Victor C. R.. 2016. “Is Loneliness in Later Life a Self‐Fulfilling Prophecy?” Aging & Mental Health 20, no. 5: 543–549. 10.1080/13607863.2015.1023767. [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. Pitman, A. , Mann F., and Johnson S.. 2018. “Advancing Our Understanding of Loneliness and Mental Health Problems in Young People.” Lancet Psychiatry 5, no. 12: 955–956. 10.1016/S2215-0366(18)30436-X. [DOI] [PubMed] [Google Scholar]
  75. Podsakoff, P. M. , Podsakoff N. P., Williams L. J., Huang C., and Yang J.. 2024. “Common Method Bias: It's Bad, It's Complex, It's Widespread, and It's Not Easy to Fix.” Annual Review of Organizational Psychology and Organizational Behavior 11: 17–61. 10.1146/annurev-orgpsych-110721-040030. [DOI] [Google Scholar]
  76. Podsakoff, P. M. , MacKenzie S. B., Lee J. Y., and Podsakoff N. P.. 2003. “Common Method Biases in Behavioral Research: A Critical Review of the Literature and Recommended Remedies.” Journal of Applied Psychology 88, no. 5: 879–903. 10.1037/0021-9010.88.5.879. [DOI] [PubMed] [Google Scholar]
  77. Preacher, K. J. , and Hayes A. F.. 2008. “Asymptotic and Resampling Strategies for Assessing and Comparing Indirect Effects in Multiple Mediator Models.” Behavior Research Methods 40, no. 3: 879–891. 10.3758/brm.40.3.879. [DOI] [PubMed] [Google Scholar]
  78. Prizeman, K. , Weinstein N., and McCabe C.. 2023. “Effects of Mental Health Stigma on Loneliness, Social Isolation, and Relationships in Young People With Depression Symptoms.” BMC Psychiatry 23: 527. 10.1186/s12888-023-04991-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  79. Prohaska, T. , Burholt V., Burns A., et al. 2020. “Consensus Statement: Loneliness in Older Adults, the 21st Century Social Determinant of Health?” BMJ Open 10, no. 8: e034967. 10.1136/bmjopen-2019-034967. [DOI] [PMC free article] [PubMed] [Google Scholar]
  80. Qin, Y. , Victor C., Qualter P., and Barreto M.. 2024. “Understanding the Psychological, Relational, Sociocultural, and Demographic Predictors of Loneliness Using Explainable Machine Learning.” Stigma and Health. 10.1037/sah0000594. [DOI] [Google Scholar]
  81. Qualter, P. , Vanhalst J., Harris R., et al. 2015. “Loneliness Across the Life Span.” Perspectives on Psychological Science 10, no. 2: 250–264. 10.1177/174569161556899. [DOI] [PubMed] [Google Scholar]
  82. Qualter, P. , Petersen K., Barreto M., Victor C., Hammond C., and Arshad S. A.. 2021. “Exploring the Frequency, Intensity, and Duration of Loneliness: A Latent Class Analysis of Data From the BBC Loneliness Experiment.” International Journal of Environmental Research and Public Health 18, no. 22: 12027. 10.3390/ijerph182212027. [DOI] [PMC free article] [PubMed] [Google Scholar]
  83. Rapolienė, G. , and Aartsen M.. 2022. “Lonely Societies: Low Trust Societies? Further Explanations for National Variations in Loneliness Among Older Europeans.” European Journal of Ageing 19, no. 3: 485–494. 10.1007/s10433-021-00649-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  84. Rhemtulla, M. , Brosseau‐Liard P. É., and Savalei V.. 2012. “When Can Categorical Variables Be Treated as Continuous? A Comparison of Robust Continuous and Categorical SEM Estimation Methods Under Suboptimal Conditions.” Psychological Methods 17, no. 3: 354–373. 10.1037/a0029315. [DOI] [PubMed] [Google Scholar]
  85. Rodriguez, M. , Pratt S., Bellet B. W., and McNally R. J.. 2025. “Solitude Can Be Good—If You See It as Such: Reappraisal Helps Lonely People Experience Solitude More Positively.” Journal of Personality 93, no. 1: 118–135. 10.1111/jopy.12887. [DOI] [PMC free article] [PubMed] [Google Scholar]
  86. Russell, D. , Peplau L. A., and Cutrona C. E.. 1980. “The Revised UCLA Loneliness Scale: Concurrent and Discriminant Validity Evidence.” Journal of Personality and Social Psychology 39, no. 3: 472–480. 10.1037/0022-3514.39.3.472. [DOI] [PubMed] [Google Scholar]
  87. Sauer‐Zavala, S. , Boswell J. F., Gallagher M. W., Bentley K. H., Ametaj A., and Barlow D. H.. 2012. “The Role of Negative Affectivity and Negative Reactivity to Emotions in Predicting Outcomes in the Unified Protocol for the Transdiagnostic Treatment of Emotional Disorders.” Behaviour Research and Therapy 50, no. 9: 551–557. 10.1016/j.brat.2012.05.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  88. Scheibe, S. , and Carstensen L. L.. 2010. “Emotional Aging: Recent Findings and Future Trends.” Journals of Gerontology. Series B, Psychological Sciences and Social Sciences 65, no. 2: 135–144. 10.1093/geronb/gbp132. [DOI] [PMC free article] [PubMed] [Google Scholar]
  89. Sholokhova, S. , Bizzari V., and Fuchs T.. 2022. “Exploring Phenomenological Interviews: Questions, Lessons Learned and Perspectives.” Phenomenology and the Cognitive Sciences 21, no. 1: 1–7. 10.1007/s11097-021-09799-y. [DOI] [Google Scholar]
  90. Shrestha, N. 2020. “Detecting Multicollinearity in Regression Analysis.” American Journal of Applied Mathematics and Statistics 8, no. 2: 39–42. 10.12691/ajams-8-2-1. [DOI] [Google Scholar]
  91. Sievert, C. , and Shirley K.. 2014. “LDAvis: A Method for Visualizing and Interpreting Topics.” In Proceedings of the International Working Conference on Advanced Visual Interfaces (pp. 63‐70).
  92. Soper, D. S. 2025. “A‐Priori Sample Size Calculator for Structural Equation Models [Software].” Available from. https://www.danielsoper.com/statcalc.
  93. Srivastava, S. , John O. P., Gosling S. D., and Potter J.. 2003. “Development of Personality in Early and Middle Adulthood: Set Like Plaster or Persistent Change?” Journal of Personality and Social Psychology 84, no. 5: 1041–1053. 10.1037/0022-3514.84.5.1041. [DOI] [PubMed] [Google Scholar]
  94. Suanet, B. , Drewelies J., Duezel S., et al. 2024. “Historical Change in Trajectories of Loneliness in Old Age: Older Adults Today Are Less Lonely, but Do Not Differ in Their Age Trajectories.” Psychology and Aging 39, no. 4: 350–363. 10.1037/pag0000803. [DOI] [PubMed] [Google Scholar]
  95. Switsers, L. , Qualter P., Pan H., et al. 2023. “Exploring the Demographic and Situational Characteristics of Older British People Experiencing Loneliness as Positive Within the BBC Loneliness Experiment.” Aging & Mental Health 27, no. 7: 1396–1402. 10.1080/13607863.2022.2088692. [DOI] [PubMed] [Google Scholar]
  96. Tan, M. , Shallis A., and Barkus E.. 2020. “Social Anhedonia and Social Functioning: Loneliness as a Mediator.” PsyCh Journal 9, no. 2: 280–289. 10.1002/pchj.344. [DOI] [PubMed] [Google Scholar]
  97. Tornstam, L. 1997. “Gerotranscendence: The Contemplative Dimension of Aging.” Journal of Aging Studies 11, no. 2: 143–154. 10.1016/S0890-4065(97)90018-9. [DOI] [Google Scholar]
  98. Tornstam, L. 2005. Gerotranscendence: A Developmental Theory of Positive Aging. Springer Publishing Company. [Google Scholar]
  99. Tornstam, L. 2011. “Maturing Into Gerotranscendence.” Journal of Transpersonal Psychology 43, no. 2: 166–180 Available from. https://www.atpweb.org/jtparchive/trps‐43‐11‐02‐000.pdf#page=56. [Google Scholar]
  100. Tsai, F. F. , and Reis H. T.. 2009. “Perceptions by and of Lonely People in Social Networks.” Personal Relationships 16, no. 2: 221–238. 10.1111/j.1475-6811.2009.01220.x. [DOI] [Google Scholar]
  101. Vedder, A. , Boerner K., Stokes J. E., Schut H. A., Boelen P. A., and Stroebe M. S.. 2022. “A Systematic Review of Loneliness in Bereavement: Current Research and Future Directions.” Current Opinion in Psychology 43: 48–64. 10.1016/j.copsyc.2021.06.003. [DOI] [PubMed] [Google Scholar]
  102. Victor, C. R. , and Yang K.. 2012. “The Prevalence of Loneliness Among Adults: A Case Study of the United Kingdom.” Journal of Psychology 146, no. 1–2: 85–104. 10.1080/00223980.2011.613875. [DOI] [PubMed] [Google Scholar]
  103. Victor, C. , Qualter P., and Barreto M.. 2019. “What Is Loneliness: Insights From the BBC Loneliness Experiment.” Innovation in Aging 3, no. Suppl 1: S373. 10.1093/geroni/igz038.1366. [DOI] [Google Scholar]
  104. Weinstein, N. , Vuorre M., Adams M., and Nguyen T. V.. 2023. “Balance Between Solitude and Socializing: Everyday Solitude Time Both Benefits and Harms Well‐Being.” Scientific Reports 13: 21160. 10.1038/s41598-023-44507-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  105. Weiss, R. 1975. Loneliness: The Experience of Emotional and Social Isolation. MIT Press. [Google Scholar]
  106. Westland, J. C. 2010. “Lower Bounds on Sample Size in Structural Equation Modeling.” Electronic Commerce Research and Applications 9, no. 6: 476–487. 10.1016/j.elerap.2010.07.003. [DOI] [Google Scholar]
  107. Wotherspoon, N. 2024. “Compelled Loneliness and Necessitated Social Isolation: “It's Like Being on the Other Side of a Mirror, Just Looking in”.” Sociology of Health & Illness 46: 744–761. 10.1111/1467-9566.13732. [DOI] [PubMed] [Google Scholar]
  108. Xing, C. , Zhang X., and Cheng S. T.. 2017. “Would More Social Partners Lead to Enhanced Health? The Paradox Between Quantity and Quality.” PsyCh Journal 6, no. 3: 219–227. 10.1002/pchj.181. [DOI] [PubMed] [Google Scholar]
  109. Yesavage, J. A. , Brink T. L., Rose T. L., et al. 1982. “Development and Validation of a Geriatric Depression Screening Scale: A Preliminary Report.” Journal of Psychiatric Research 17, no. 1: 37–49. 10.1016/0022-3956(82)90033-4. [DOI] [PubMed] [Google Scholar]
  110. Zheng, K. , Johnson S., Jarvis R., et al. 2023. “The Experience of Loneliness Among International Students Participating in the BBC Loneliness Experiment: Thematic Analysis of Qualitative Survey Data.” Current Research in Behavioral Sciences 4: 100113. 10.1016/j.crbeha.2023.100113. [DOI] [Google Scholar]
  111. Zhou, H. , and Long L.. 2004. “Statistical Remedies for Common Method Biases.” Advances in Psychological Science 12, no. 6: 942–950. 10.3969/j.issn.1671-3710.2004.06.018. [DOI] [Google Scholar]
  112. Zuckerman, M. , Lubin B., and Rinck C. M.. 1983. “Construction of New Scales for the Multiple Affect Adjective Check List.” Journal of Behavioral Assessment 5, no. 2: 119–129. 10.1007/BF01321444. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Table S1: Characteristics of participants.

Table S2: Between‐group differences of the mediating effects.

Table S3: The model fit of the 2‐group MG‐SEM controlling for IDV.

Table S4: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 1 (younger adults, <= 39.2).

Table S5: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 2 (older adults, > 39.2).

Table S6: The model fit of the 4‐group MG‐SEM.

Table S7: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 1 (younger female, <= 39.2).

Table S8: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 2 (younger male, <= 39.2).

Table S9: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 3 (older female, > 39.2).

Table S10: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 4 (older male, > 39.2).

Table S11: The model fit of the 4‐group MG‐SEM.

Table S12: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 1 (emerging adulthood, <= 25 years).

Table S13: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 2 (early‐middle adulthood, 25~44 years).

Table S14: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 3 (later‐middle adulthood, 45~64 years).

Table S15: Parallel mediating effects of loneliness self‐stigma and meaning of loneliness in Group 4 (later adulthood, > = 65 years).

Table S16: The model fit and mediating effects of competing sequential (chain) mediation models.

Table S17: The LDA model index of topic number in the younger age group (<= 39.2 years old).

Table S18: The LDA model index of topic number in the older age group (> 39.2 years old).

Figure S1: Top‐20 most relevant words (terms) for each topic of the younger age group (<= 39.2 years old).

Figure S2: Top‐20 most relevant words (terms) for each topic of the older age group (> 39.2 years old).

Table S19: The words and their distribution probabilities for each topic.

PCHJ-15-e70115-s001.docx (313.2KB, docx)

Data Availability Statement

Access to the dataset is available from the authors upon reasonable request.

Articles from PsyCh Journal are provided here courtesy of Wiley

RESOURCES