The relationship between family communication networks and elderly loneliness: A structural equation modeling approach across three generations

Mahdi Sepidarkish; Afsaneh Bakhtiari; Zeinab Baheshmat; Shabnam Omidvar

doi:10.1371/journal.pone.0358707

. 2026 Oct 5;21(10):e0358707. doi: 10.1371/journal.pone.0358707

The relationship between family communication networks and elderly loneliness: A structural equation modeling approach across three generations

Mahdi Sepidarkish ^1,², Afsaneh Bakhtiari ^2,^*, Zeinab Baheshmat ³, Shabnam Omidvar ²

Editor: Cheong Yu Stephen Chan⁴

PMCID: PMC13637883 PMID: 42832538

Abstract

Background and Objective

Loneliness is a widespread challenge among older adults, often linked to negative health outcomes. This study explores how family communications among individual members representing three generations is associated with loneliness in elderly participants, aiming to clarify which family relationships are most impactful.

Methods

Using partial least squares structural equation modeling, we examined the effects of marital intimacy in both the older generation (G1) and their adult children (G2), as well as broader family factors such as intergenerational solidarity, family cohesion, sibling relationship, in-law interaction, children’s support for parents and grandparent-grandchild relationship, on loneliness in elderly family members.

Results

The findings revealed that marital intimacy within the G2 was the only direct predictor associated with reduced loneliness among older adults (β = −0.314, p = 0.019). The marital closeness of the older generation showed only a weak association (β = −0.194, p = 0.067). Other family relationship factors, including general cohesion and intergenerational bonds, did not demonstrate meaningful direct or indirect effects. Importantly, the quality of G1 marital relationships showed indirect associations with loneliness by strengthening the marital bonds of G2, highlighting a pattern of relationship qualities being passed between generations.

Conclusion

This study suggests that it is not broad family connectedness but rather the quality of key dyadic relationships, especially among middle-aged couples, that matters most for elderly loneliness. Efforts to enhance marital relationships in the middle generation may be associated with lower loneliness for their aging parents.

Introduction

Loneliness among older adults has emerged as a pressing public health concern worldwide, given its well-established links to adverse physical and mental health outcomes, including increased risk of depression, cognitive decline, and mortality [1,2]. Traditionally, the scholarly focus on mitigating loneliness in elderly populations has centered on broad constructs such as family cohesion and generalized social support networks, which emphasize the overall size of the family network, frequency of contact, or perceived availability of assistance [3,4]. This conventional perspective largely treats family connectedness as a quantitative phenomenon, assuming that higher levels of contact and broader support are associated with better emotional well-being [5]. Not all family relationships carry the same emotional weight, and certain dyadic ties may be more consequential than others for elderly loneliness.

Recent theoretical developments support this distinction [6]. The Convoy Model of social relations posits that social ties vary in function and emotional quality across the lifespan, suggesting that emotionally rich dyadic relationships may be more influential than broad social integration [7]. One particularly relevant dyad is the marital relationship within the middle generation. Adults in their 40s and 50s, often referred to as the “pivot generation,” occupy a unique structural position in the family. They typically coordinate family gatherings, mediate intergenerational conflicts, and serve as the primary conduit for emotional and practical support to aging parents [8]. When their own marriage is characterized by closeness and mutual support, they tend to perform these functions more effectively. Conversely, marital distress among middle-aged couples may disrupt the entire family support system. Empirical evidence indicates that marital intimacy within this generation correlates with relational health across multiple family members [9,10]. Furthermore, research on intergenerational transmission demonstrates that parental marital quality shapes offspring’s relationship dynamics, suggesting potential indirect pathways across generations [11].

However, focusing exclusively on the middle generation’s marriage provides an incomplete picture. Older adults are embedded within a broader network of family relationships spanning three generations, including their own marital bond (G1), relationships with adult children and their spouses, ties with grandchildren, and in some cases sibling connections. The term “family communication network” here refers to the patterns of emotional exchange across these various dyadic relationships, as opposed to a simple count of social contacts. How these diverse connections collectively are related to loneliness remains unclear from existing literature. Some studies report protective effects of general family cohesion, whereas others emphasize specific intimate bonds as primary determinants of emotional well-being in later life [12–14]. Resolving this ambiguity matters because without it, interventions to reduce elder loneliness may target the wrong family relationships.

The present study addresses these gaps by applying structural equation modeling to examine how family communication networks across three generations are associated with loneliness in older adults. Rather than treating the family as a single source of support, we disaggregate the network into specific dyadic relationships: marital intimacy in the older generation (G1) and the middle generation (G2), parent-child support, grandparent-grandchild bonds, sibling relationships, in-law interactions, and general family cohesion. This design allows us to determine whether broad family connectedness or specific dyadic are associated with elderly loneliness, and whether any effects operate directly or indirectly across generations. By doing this in the Iranian family context, we offer a framework that other researchers can replicate with their own samples in different cultural settings. Based on family systems and intergenerational relationship theories, we expected that stronger dyadic and broader family relational ties would be associated with lower loneliness among older adults. Specifically, we hypothesized that: (1) higher marital intimacy in both G1 and G2 would be associated with lower loneliness; (2) broader family relational factors, including cohesion, intergenerational solidarity, children’s support, and grandparent-grandchild ties, would be related to loneliness; and (3) these associations might partly operate through indirect intergenerational pathways.

Methods

Design study

This community-based cross-sectional study was conducted between 15^th September 2022 and 10^th March 2023 in Jouybar, a city in the northern part of Iran, Mazandaran province. Iran’s relatively strong intergenerational family ties make it a relevant context for examining family relational pathways across generations. The research aimed to assess the relationship between family communication networks and psychosocial well-being, specifically loneliness and social isolation, among older adults. The analysis encompassed a multi-generational perspective by examining family communication across three generations: first generation-G1 (older adults/grandparents), second generation-G2 (adult children), and third generation-G3 (grandchildren). This approach provided insights into intergenerational connections at the individual family level, though findings should not be generalized to the entire societal generation. According to recent municipal data, Jouybar has a population of around 34,000, with an estimated 4,050 older adults, representing 11.9% of the total population. The city is served by three government-run Comprehensive Health Service Centers, which maintain up-to-date health records for all households within their designated coverage zones.

Participants

The study enrolled married couples aged 60 years or older who were permanent residents of Jouybar. To be eligible, participants had to meet the following criteria: (1) no prior remarriage, (2) at least two married children with grandchildren, (3) regular access to children and grandchildren (either in person or via telephone), (4) functional independence (no major limitations in basic or instrumental activities of daily living), (5) absence of significant hearing, vision, or speech impairments that could interfere with communication, and (6) no terminal illness, recent severe acute disease (within the last three months), or current use of medications for Alzheimer’s disease or depression. Individuals were excluded if: (1) their children or grandchildren lived abroad, (2) they had severe cognitive decline or required extensive assistance with daily activities, or (3) they refused participation upon initial contact.

Sampling strategy and data collection

Each sampling unit consisted of a three-generation family, including an older couple, one adult child with their spouse, and one grandchild. To achieve a representative sample of elderly couples in Jouybar, a two-stage sampling approach was used. In the first stage, stratified sampling was applied using the city’s three Comprehensive Urban Health Service Centers, which represent all such centers in Jouybar, as separate strata. This ensured appropriate geographical coverage. In the second stage, simple random sampling was conducted within each stratum. Eligible couples were identified from health center records and selected using Stata’s computerized random number generator to reduce selection bias.

Eligible older adults were invited to attend the health centers. At the visit, the study objectives and procedures were explained, and written informed consent was obtained. The questionnaires were then administered. Questionnaires for the first generation (older adults) were completed through face-to-face interviews by a trained researcher. For the second and third generations, the study was introduced during the initial telephone contact. Verbal informed consent was obtained before starting the interview, and consent was audio-recorded by the researcher. Questionnaires were then completed via telephone interview by the same researcher.

Sample size calculation

The required sample size was determined based on the study’s primary objective of examining the association between family communication networks and loneliness. A minimum correlation coefficient of 0.30 was used as the reference effect size, derived from prior studies on loneliness and family networks in similar populations [15]. Using PASS software (version 21) [16], with a two-tailed test, 95% confidence level (α = 0.05), and 80% power (β = 0.20), the initial calculation yielded a requirement of 85 elderly couples (170 individuals). To account for a potential 10% non-response rate or incomplete data, the target sample size was adjusted upward to 94 couples (188 individuals). In practice, we enrolled 107 eligible couples who met all inclusion criteria. Following the exclusion of five cases due to incomplete data, the final analytic sample comprised 107 older adults (representing 107 couples). Alongside the first-generation participants, 107 adult children, their 107 spouses, and 107 grandchildren were also enrolled, resulting in a total of 428 individuals representing three generational levels within extended families.

Measurements

The study used established instruments and Iranian-developed measures with previously reported psychometric evidence. To complement the existing psychometric evidence, we further assessed construct validity via confirmatory factor analysis (CFA) using AMOS version 24. A brief synopsis of the psychometric properties is presented below, whereas the comprehensive validation evidence is set out in، Table 2 and S1 Table.

Sociodemographic and health characteristics questionnaire

The sociodemographic questionnaire systematically captured key participant characteristics across three domains. Demographic information included age, sex, occupational status, educational attainment, and marital status. Living arrangements were categorized as living alone, cohabiting with a spouse, residing with both spouse and unmarried children, or other household configurations. Economic status was assessed through participant self-reporting. Health-related variables encompassed current medication use (with polypharmacy operationally defined as consumption of four or more daily medications), physician-diagnosed chronic conditions (including cardiovascular, metabolic, endocrine, respiratory, musculoskeletal and renal diseases), hospitalization history within the preceding year, and current tobacco use status. This comprehensive assessment enabled detailed characterization of the study population’s baseline profile.

De Jong Gierveld loneliness scale

The De Jong Gierveld Loneliness Scale (1985) measures loneliness through emotional (6 items) and social (5 items) components. The Persian version validated by Hosseini et al. (2021) [17] for Iranian older adults omitted three items, leaving 8 items (5 social, 3 emotional). Following standard scoring, each response is recoded as 0 or 1, yielding a total score ranging from 0 to 8, with higher scores indicating greater loneliness. Example item: “There are enough people I feel close to.” The translation followed forward-backward procedures with expert committee review. CFA confirmed a two-factor structure with acceptable fit (CFI = 0.92, RMSEA = 0.06). Construct validity was supported by significant correlations with depression (r = 0.48) and perceived social support (r = −0.41), while known-groups validity demonstrated significant differences by living arrangement (p < 0.001). The scale showed good internal consistency in the Iranian population (Cronbach’s α > 0.80) [17].

Family communication network questionnaires

To identify types of intergenerational relationships in the elderly family, the classification framework proposed by Silverstein et al. [18] was employed.

First generation (elderly couples)

Marital intimacy scale.

The 17-item marital intimacy scale, originally conceptualized by Thompson and Walker (1983), assesses relationship closeness using a 7-point Likert response format (1 = “never” to 7 = “always”).Total scores range from 17 to 119, with higher values reflecting stronger perceived intimacy. The Persian version validated by Salehi Moghadam et al. (2024) [19] underwent forward-backward translation with expert committee review and cognitive debriefing. Exploratory factor analysis (EFA) revealed a single-factor solution accounting for 58.7% of variance, while CFA confirmed the unidimensional structure (CFI = 0.91, RMSEA = 0.07). Construct validity was evidenced by significant positive correlations with marital satisfaction (r = 0.62) and negative correlations with marital conflict (r = −0.55). Known-groups validity was supported by significant differences across educational levels (p < 0.001). The instrument demonstrated exceptional internal consistency (Cronbach’s α = 0.96) and test-retest reliability (ICC = 0.92) in the Iranian population [19].

Intergenerational solidarity questionnaire

This 16-item multidimensional instrument, based on the framework of Silverstein et al. (2010) [18], assesses three key dimensions of grandparent-grandchild relationships: (1) emotional attachment (rated 1 [“very bad”] to 4 [“very good”]), (2) social attachment (1 [“never”] to 5 [“always”]), and (3) functional attachment (1 [“none”] to 4 [“a lot”]). Raw scores are converted to a standardized 0–100 metric, where elevated scores reflect stronger intergenerational connections. The Persian version by Momtaz et al. (2018) [20] was validated through forward-backward translation and expert panel review. CFA confirmed the three-factor structure with good fit (CFI = 0.921, RMSEA = 0.063), and the scale demonstrated acceptable internal consistency (α = 0.86).

Second generation (adult children)

Marital intimacy questionnaire.

The measure for assessing it, originally conceptualized by Thompson and Walker (1983), has been introduced earlier [19].

Family cohesion scale

This 28-item instrument evaluates family cohesion across eight distinct dimensions: (1) mother-child bonding, (2) father-child bonding, (3) shared time, (4) interaction settings, (5) emotional connections, (6) joint decision-making, (7) marital dynamics, and (8) parent-child relationships. Items are rated on a 5-point Likert scale (1 = strongly disagree to 5 = strongly agree), yielding total scores between 28–140, with scores ≥96 indicating high family cohesion. The scale was originally developed in Iran by Samani [21]. EFA supported an eight‑subscale structure (inter‑factor r: 0.15–0.73; item‑total r: 0.64–0.79). Convergent and discriminant validity were supported by correlations with the Family Process Scale (r = 0.56, p < 0.001) and Marital Problems Scale (r = –0.24, p < 0.05), respectively, and the scale distinguished divorced from intact families (t = 3.96, p < 0.001). Internal consistency was acceptable (α = 0.79).

Sibling relationship index

This 25-item instrument [22] evaluates the quality of sibling relationships, with higher scores indicating greater relational difficulties. Items are rated on a 7-point Likert scale (1 = never to 7 = always). The original scale demonstrates strong psychometric properties, including excellent internal consistency (α ≥ 0.90) and established content, construct, factorial, and known-groups validity (Walmyr Publishing Company, 2015). Because no formal psychometric validation of the Persian version was available [23], construct validity was evaluated in the present study using CFA, which supported the proposed one-factor model (S1 Table). Internal consistency was excellent (α = 0.954).

In-law interaction questionnaire

This 50-item instrument assesses conflict dynamics between married couples and their parents‑in‑law across five domains: interference in marital decisions, boundary violations, communication difficulties, cultural/value conflicts, and financial tensions. Items are rated on a 5‑point Likert scale (1 = strongly disagree to 5 = strongly agree), with total scores ranging from 50 to 250; higher scores indicate greater relational conflict. The questionnaire was developed through qualitative research and expert review. Content validity was established during instrument development, whereas construct validity was further evaluated in the present study using CFA, which supported the proposed five-factor structure. Internal consistency was excellent in both the original study (α = 0.96) and the present study (α = 0.962) [[24]; S1 Table)].

Children’s support for parents scale

This 7-question instrument assesses adult children’s support from their parents in three dimensions, including instrumental support, financial support, and emotional support. Items are rated on Likert scales tailored to each dimension: instrumental support items range from 0 (never) to 6 (daily), financial support is measured on a 0–6 scale, and emotional support items range from 0 (none) to 5 (very much). The scale was developed in Iran based on items adapted from the NSFH and SHARE surveys [25]. Content validity was established through expert review, EFA supported the original three-factor structure, and CFA in the present study confirmed this structure. Internal consistency was acceptable (α = 0.792) (S1 Table).

Third generation (grandchildren)

Grandparent-grandchild scale.

The grandparent-grandchild Bonding Scale is a 10-item psychometric instrument designed to evaluate the quality of intergenerational relationships across two key dimensions: emotional connectedness and participatory engagement. Utilizing a 4-point Likert response format (1 = “not at all” to 4 = “very much”), the measure yields a total score ranging from 10 to 40, where higher scores reflect stronger relational bonds. The Persian version was validated in a methodological study involving Iranian students. The instrument demonstrated excellent content validity (CVI = 0.98), CFA supported the two-factor structure, and internal consistency was good (α = 0.89) [26].

Ethical consideration

The Institutional Review Board of Babol University of Medical Sciences granted ethical approval for the study (Approval Ethical Code: IR.MUBABOL.REC.1401.073). This study was conducted under Good Clinical Practice and the principles of the Declaration of Helsinki. Before participation, all individuals provided written informed consent following a detailed explanation of the study’s objectives and procedures.

Statistical analysis

We conducted our analysis using partial least squares structural equation modeling (PLS-SEM) implemented in SmartPLS 3.0. This multivariate technique was selected for its ability to handle both reflective and formative constructs while making minimal assumptions about data distribution. The analysis proceeded in two sequential stages: measurement model evaluation followed by structural model assessment. For the measurement model evaluation, we examined convergent validity through three established criteria: indicator loadings, average variance extracted (AVE) values, and composite reliability coefficients. We assessed discriminant validity using both the Fornell-Larcker criterion and heterotrait-monotrait (HTMT) ratios. In the structural model assessment, we employed a nonparametric bootstrapping procedure with 5,000 resamples to examine the hypothesized relationships. We calculated path coefficients and their significance levels to test direct effects, while mediation analysis evaluated indirect pathways. The analysis included computation of effect sizes (f²) for all significant paths. We assessed the model’s predictive relevance through the blindfolding procedure and evaluated overall model fit using standardized root mean square residual (SRMR) and normed fit index (NFI). To ensure robustness, we conducted supplementary analyses including multi-group analysis of demographic variables and post-hoc power analysis using the inverse square root method. The analysis accounted for both the measurement properties of the constructs and the structural relationships between them, while maintaining appropriate standards for statistical significance and practical relevance.

Results

Participant characteristics

Participant demographic and health characteristics are summarized in Table 1. The study enrolled 107 older adult participants with the following demographic and health characteristics. Age distribution revealed a predominance of younger seniors, with 70.1% (n = 75) aged 60–69 years, 25.2% (n = 27) in the 70–79 range, and 4.7% (n = 5) aged 80 or above. The sample included more female participants (58.9%, n = 63) than males (41.1%, n = 44). Educational attainment showed 38.3% (n = 41) had no formal schooling, followed closely by 37.4% (n = 40) with primary education, while secondary (9.3%, n = 10), high school (12.1%, n = 13), and post-secondary (2.8%, n = 3) education were less common. Employment status indicated 71.9% (n = 77) were not currently working, compared to 28.1% (n = 30) who were employed. Health status measures showed 51.4% (n = 55) of participants reported chronic health conditions, with 48.6% (n = 52) indicating no significant medical history. Medication patterns revealed 27.1% (n = 29) met polypharmacy criteria (≥5 medications), while 72.9% (n = 78) did not. These baseline characteristics provide important contextual information about the study population for subsequent analyses.

Table 1. Study population characteristics (N = 107).

Variable	Categories	Number	Percent
Age(years)	60–69	75	70.1
Age(years)	70–79	27	25.2
	80+	5	4.7
Gender	Male	44	41.1
Gender	Female	63	58.9
Education	Illiterate	41	38.3
	Elementary	40	37.4
	Middle	10	9.3
	High	13	12.1
	College or higher	3	2.8
Occupation	Employed	30	28.1
Occupation	Unemployed	77	71.9
Underling disease	Yes	55	51.4
Underling disease	No	52	48.6
Polypharmacy	Yes	29	27.1
Polypharmacy	No	78	72.9

Open in a new tab

Model fit assessment

The overall fit of the measurement and structural model was evaluated using the standardized root mean square residual (SRMR) and the normed fit index (NFI), which are recommended for PLS-SEM. The SRMR value was 0.072, which is below the commonly accepted threshold of 0.08, indicating an acceptable approximate fit between the hypothesized model and the observed data. The NFI was 0.91, exceeding the recommended cutoff of 0.90, suggesting that the model represents a substantial improvement over a null model. Additionally, the squared Euclidean distance (d_ULS) was 1.234 and the geodesic distance (d_G) was 0.876. For all endogenous constructs, the Stone-Geisser Q² values (obtained via the blindfolding procedure) were greater than zero (loneliness Q² = 0.184; G2 marital intimacy Q² = 0.387; intergenerational solidarity Q² = 0.263; sibling relationship Q² = 0.258; in-law relationship Q² = 0.191; grandparent-grandchild relationship Q² = 0.168; children’s support Q² = 0.245), confirming the model’s predictive relevance. Taken together, these fit indices support the adequacy of the model for interpreting the direct and indirect pathways.

Measurement model assessment

Measurement model outcomes, including reliability, convergent validity, discriminant validity, and variance explained, are presented in Table 2. Reliability analysis revealed excellent internal consistency, with Cronbach’s alpha coefficients ranging from 0.792 to 0.971. Composite reliability (CR) values ranged from 0.866 to 0.974, consistently exceeding the recommended threshold of 0.70. The rho_A coefficients showed similarly robust results (0.838–0.973), confirming the reliability of all measurement scales. Convergent validity, assessed through average variance extracted (AVE), revealed mixed results across constructs. Several constructs demonstrated adequate convergent validity with AVE values exceeding the recommended threshold of 0.50. These included G1 marital intimacy (AVE = 0.689), G2 marital intimacy (AVE = 0.629), sibling relationships (AVE = 0.571), and children’s support for parents (AVE = 0.623). However, other constructs showed AVE values below the 0.50 threshold, indicating that a substantial portion of item variance was not captured by the latent construct. Loneliness showed an AVE of 0.480, falling just below the recommended threshold. Family cohesion demonstrated an AVE of 0.468, also below the acceptable level. In-law relationships exhibited the lowest AVE at 0.397, suggesting limited convergent validity. Intergenerational solidarity showed an AVE of 0.393, similarly indicating suboptimal convergent validity. Grandparent-grandchild relationships demonstrated an AVE of 0.469, marginally below the threshold. Discriminant validity was assessed using heterotrait-monotrait (HTMT) ratios, which provide a more conservative test of discriminant validity. All HTMT values were below the conservative threshold of 0.85, ranging from 0.232 to 0.636. The lowest HTMT value was observed for in-law relationships (0.232), indicating strong discriminant validity. The highest HTMT value was for sibling relationships (0.636), still well below the threshold. This confirms that discriminant validity was established for all constructs, indicating each latent construct is empirically distinct from the others in the model. The model explained substantial variance in several endogenous constructs, as indicated by R² values. G2 marital intimacy showed particularly strong explanatory power (R² = 0.413, adjusted R²= 0.395). Intergenerational family solidarity demonstrated considerable explanatory power (R² = 0.291, adjusted R² = 0.278). Sibling relationships also showed strong explanatory power (R² = 0.286, adjusted R² = 0.273). The model accounted for moderate variance in loneliness (R²= 0.226). In-law relationships (R² = 0.214) and grandparent-grandchild relationships (R² = 0.192) demonstrated moderate explanatory power. Family cohesion showed relatively weaker explanatory power (R² = 0.098). G1 marital intimacy served as an exogenous variable in the model, so it’s R² was not estimated. The children’s support for parents scale, while showing slightly lower reliability than other measures (α = 0.792, CR = 0.866), still met acceptable psychometric standards and explained considerable variance (R² = 0.274). These results collectively confirm adequate reliability and discriminant validity for all constructs. However, convergent validity was less consistently established across all constructs, with several AVE values falling below the recommended threshold. Given the exploratory nature of this study and the modest sample size, we retained these constructs while interpreting their findings with appropriate caution. Findings involving constructs with suboptimal AVE, particularly in-law relationships and intergenerational solidarity, should be interpreted with caution and require replication in larger samples.

Table 2. Outcomes of the measurement model.

Latent variable	Cronbach’s alpha	CR	rho_A	R²	R² adjusted	AVE	HTMT
Loneliness	0.865	0.903	0.865	0.226	0.171	0.480	0.366
Marital intimacy (G1)	0.971	0.974	0.973	NE	NE	0.689	0.411
Marital intimacy (G2)	0.962	0.966	0.966	0.413	0.395	0.629	0.436
Sibling relationship	0.954	0.959	0.966	0.286	0.273	0.571	0.636
Family cohesion	0.863	0.905	0.949	0.098	0.089	0.468	0.559
In-law relationship	0.962	0.964	0.964	0.214	0.191	0.397	0.232
Intergenerational family solidarity	0.896	0.910	0.902	0.291	0.278	0.393	0.587
Grandparent-grandchild relationship	0.878	0.900	0.880	0.192	0.176	0.469	0.417
Children’s support for parents	0.792	0.866	0.838	0.274	0.267	0.623	0.490

Open in a new tab

NE: not estimated, CR: composite reliability

Structural relationships between family dynamics and loneliness

Tables 3 and 4 and Fig 1 present the structural models and hypothesis tests examining direct and indirect pathways linking family relationship variables to loneliness (L). Using PLS-SEM with 5,000 bootstrap samples, we analyzed these relationships across two generations. The results reveal distinct patterns in how marital and intergenerational bonds relate to loneliness in later life.

Table 3. Direct effects of association among marital intimacy (generation 1), marital intimacy (generation 2), family cohesion, in-law relationship, intergenerational family solidarity, grandparent-grandchild relationship, children’s support for parents, and loneliness.

Hypothesis	Path	Path coefficient (95% CI)	S.D.	T-Statistics	p-value	Results
H1	MI (G1) → L	−0.194 (−0.389, 0.017)	0.105	1.836	0.067	Not-supported
H2	MI (G2) → L	−0.314 (−0.556, −0.041)	0.133	2.356	0.019	Supported
H3	FC → L	−0.093 (−0.361, 0.146)	0.125	0.751	0.453	Not-supported
H4	In-law R → L	−0.028(−0.247, 0.211)	0.117	0.232	0.816	Not-supported
H5	IFS → L	−0.095 (−0.305, 0.117)	0.110	0.904	0.366	Not-supported
H6	GGR → L	0.103 (−0.155, 0.353)	0.130	0.806	0.420	Not-supported
H7	CS → L	0.022 (−0.187, 0.232)	0.108	0.314	0.754	Not-supported

Open in a new tab

MI (G1): marital intimacy (generation 1), L: loneliness, MI (G2): marital intimacy (generation 2), FC: family cohesion, In-law R: in-law relationship, IFS: intergenerational family solidarity, GGR: grandparent-grandchild relationship, CS: children’s support for parents.

Table 4. Indirect effects of association among marital intimacy (generation 1), marital intimacy (generation 2), family cohesion, in-law relationship, intergenerational family solidarity, grandparent-grandchild relationship, children’s support for parents, and loneliness.

Hypothesis	Path	Path coefficient (95% CI)	S.D.	T-Statistics	p-value	Results
H1	FC → CS → L	0.011 (−0.108, 0.126)	0.059	0.301	0.764	Not-supported
H2	MI (G1) → FC → CS → L	0.003 (−0.038, 0.042)	0.020	0.276	0.782	Not-supported
H3	MI (G1) → FC → L	−0.030 (−0.068, 0.105)	0.043	0.689	0.491	Not-supported
H4	CS → GGR → L	0.025 (−0.040, 0.110)	0.037	0.631	0.528	Not-supported
H5	FC → CS → GGR → L	0.013 (−0.020, 0.059)	0.020	0.618	0.537	Not-supported
H6	MI (G1) → FC → CS → GGR → L	0.004 (−0.007, 0.019)	0.007	0.566	0.572	Not-supported
H7	CS → IFS → GGR → L	0.012 (−0.022, 0.050)	0.018	0.667	0.505	Not-supported
H8	FC → CS → IFS → GGR → L	0.006 (−0.012, 0.028)	0.010	0.640	0.552	Not-supported
H9	MI (G1) → FC → CS → IFS → GGR → L	0.002 (−0.004, 0.009)	0.003	0.602	0.547	Not-supported
H10	IFS → GGR → L	0.031 (−0.053, 0.126)	0.043	0.683	0.485	Not-supported
H11	FC → IFS → GGR → L	0.008 (−0.013, 0.041)	0.013	0.488	0.626	Not-supported
H12	MI (G1) → FC → IFS → GGR → L	0.002 (−0.004, 0.013)	0.004	0.492	0.622	Not-supported
H13	CS → In-law R → L	0.001 (−0.034, 0.041)	0.017	0.096	0.923	Not-supported
H14	FC → CS → In-law R → L	0.001 (−0.019, 0.022)	0.009	0.093	0.926	Not-supported
H15	MI (G1) → FC → CS →In-law R → L	0.000 (−0.006, 0.016)	0.003	0.086	0.931	Not-supported
H16	FC → In-law R → L	−0.001 (−0.043, 0.022)	0.011	0.021	0.983	Not-supported
H17	MI (G1) → FC → In-law R → L	−0.001 (−0.006, 0.008)	0.004	0.019	0.984	Not-supported
H18	CS → MI (G2) → In-law R → L	0.001 (−0.026, 0.023)	0.010	0.151	0.880	Not-supported
H19	FC → CS → MI (G2) → In-law R → L	0.001 (−0.010, 0.012)	0.005	0.144	0.885	Not-supported
H20	MI (G1) → FC → CS → MI (G2) → In-law R → L	0.000 (−0.004, 0.004)	0.002	0.133	0.894	Not-supported
H21	MI (G1) → MI (G2) → In-law R → L	0.006 (−0.049, 0.057)	0.027	0.221	0.825	Not-supported
H22	FC → SR → MI (G2) → In-law R → L	0.001 (−0.010, 0.015)	0.006	0.201	0.841	Not-supported
H23	MI (G1) → FC → SR → MI (G2) → In-law R → L	0.000 (−0.004, 0.005)	0.002	0.188	0.851	Not-supported
H24	MI (G2) → In-law R → L	0.015 (−0.106, 0.129)	0.061	0.217	0.828	Not-supported
H25	SR → MI (G2) → In-law R → L	−0.004 (−0.042, 0.029)	0.017	0.180	0.857	Not-supported
H26	MI (G1) → SR → MI (G2) → In-law R → L	0.001 (−0.009, 0.015)	0.005	0.137	0.891	Not-supported
H27	CS → IFS → L	−0.039 (−0.141, 0.046)	0.048	0.825	0.409	Not-supported
H28	FC → CS → IFS → L	−0.021 (−0.072, 0.030)	0.026	0.784	0.433	Not-supported
H29	MI (G1) → FC → CS → IFS → L	−0.007 (−0.029, 0.008)	0.009	0.681	0.496	Not-supported
H30	FC → IFS → L	−0.020 (−0.086, 0.033)	0.030	0.706	0.480	Not-supported
H31	MI (G1) → FC → IFS → L	−0.007 (−0.032, 0.010)	0.010	0.653	0.514	Not-supported
H32	CS → MI (G2) → L	−0.040 (−0.128, 0.016)	0.038	0.925	0.355	Not-supported
H33	FC → CS → MI (G2) → L	−0.022 (−0.073, 0.009)	0.021	0.867	0.386	Not-supported
H34	MI (G1) → FC → CS → MI (G2) → L	−0.007 (−0.026, 0.003)	0.007	0.786	0.432	Not-supported
H35	MI (G1) → MI (G2) → L	−0.140 (−0.286, −0.024)	0.067	2.128	0.033	Supported
H36	FC → SR → MI (G2) → L	−0.024 (−0.068, 0.020)	0.022	1.278	0.201	Not-supported
H37	MI (G1) → FC → SR → MI (G2) → L	−0.008 (−0.024, 0.007)	0.008	1.153	0.249	Not-supported
H38	SR → MI (G2) → L	0.066 (−0.036, 0.196)	0.060	1.186	0.236	Not-supported
H39	MI (G1) → SR → MI (G2) → L	−0.019 (−0.071, 0.006)	0.021	0.859	0.391	Not-supported

Open in a new tab

Direct effects on loneliness

Direct effects of all family relationship variables on loneliness are shown in Table 3. The analysis identified G2 marital intimacy as the only significant direct associate of loneliness (β = −0.314, p = 0.019), indicating that stronger marital bonds among adult children are associated with lower levels of loneliness in older family members. This aligns with theoretical models emphasizing the “generational pivot” role of middle-aged couples in maintaining family emotional support networks. In contrast, G1 marital intimacy showed only a marginal negative association (β = −0.194, p = 0.067), suggesting that older couples’ relationship quality may have limited direct associations with their loneliness.

Notably, none of the broader family dynamics, including family cohesion (β = −0.093, p = 0.453), in-law relationships (β = −0.028, p = 0.816), intergenerational solidarity (β = −0.095, p = 0.366), grandparent-grandchild bonds (β = 0.103, p = 0.420), or children’s support (β = 0.022, p = 0.754), demonstrated significant direct effects. These null findings challenge assumptions that generalized family connectedness is uniformly associated with lower loneliness, instead highlighting the unique importance of specific high-quality dyadic relationships.

Indirect pathways

Indirect effects (mediation pathways) are detailed in Table 4, following the description of the model in the Results. Mediation analysis yielded one statistically significant pathway among 39 tested: G1 marital intimacy showed an indirect association with loneliness through its relationship with G2 marital intimacy (β = −0.140, p = 0.033). This supports a generational transmission model where parents’ relationship quality enhances their children’s marital bonds, which in turn is associated with lower loneliness. The comprehensive non-significance of all other indirect paths (e.g., through family cohesion, intergenerational solidarity, or in-law relationships) suggests these variables do not operate as theorized mediators. For instance, neither children’s support (β = 0.003–0.025, all p > 0.05) nor grandparent-grandchild relationships (β = 0.012–0.031, all p > 0.05) showed protective associations with loneliness, despite their hypothesized roles in family systems theory. Among the 39 indirect pathways tested, only the pathway from G1 marital intimacy through G2 marital intimacy to loneliness showed a statistically significant association (β = −0.140, p = 0.033). However, after applying Benjamini-Hochberg false discovery rate correction for multiple comparisons (q = 0.10), this pathway did not remain statistically significant (q = 0.105). Given the exploratory nature of our mediation analyses and the modest sample size, this finding should be interpreted with caution and requires replication in larger samples before firm conclusions can be drawn.

Discussion

This study set out to explore how family communication dynamics across three generations related to loneliness in older adults, using a SEM framework. The results offer a nuanced picture, highlighting the critical role of middle-generation marital intimacy in being associated with lower elderly loneliness, while challenging commonly held assumptions about the uniform benefits of broader family ties. The strongest and only statistically significant direct effect on elderly loneliness was found in the quality of marital intimacy within the G2, that is, adult children of the elderly participants. A closer marital relationship between these middle-aged couples was associated with lower levels of loneliness among their aging parents (β = −0.314, p = 0.019). This finding supports the idea that the emotional stability and cohesion within the “pivot generation” plays a bridging role in maintaining intergenerational emotional flow, which in turn is associated with lower loneliness [27]. This pattern may partly reflect the Iranian family context, in which middle-aged couples often maintain intergenerational emotional continuity. This pattern may be less pronounced in more individualistic societies.

Interestingly, the marital quality of the older generation (G1) showed only a marginal relationship with loneliness (β = −0.194, p = 0.067). This may reflect the reduced social leverage or influence of elderly couples on the broader family network, especially when compared to their adult children who often coordinate family gatherings and maintain connections across kinship lines. Furthermore, broader relational constructs, such as family cohesion, intergenerational solidarity, in-law relationships, and even grandparent-grandchild ties, did not show significant direct associations with loneliness. While somewhat surprising, these findings suggest that general feelings of connectedness or perceived closeness may not be sufficient for lowering loneliness unless they are anchored in high-quality, stable dyadic relationships. This contrasts with our initial expectation that broader family relational factors would show stronger associations with loneliness, suggesting that emotionally meaningful dyadic bonds may be more influential than broader perceptions of family connectedness. This insight echoes a growing body of literature emphasizing the quality over quantity of social interactions in aging populations [28,29].

Prior studies have often highlighted the protective role of close family relationships in being associated with lower loneliness among older adults [30,31]. However, many of these works treat the family as a monolithic structure, without parsing out the specific dynamics that may be more or less important. Our results provide a more fine-grained analysis and indicate that not all forms of family connectedness carry the same weight. In particular, while grandparent-grandchild relationships have been celebrated in cultural and qualitative studies as sources of joy and meaning in old age [32], our quantitative model did not find them to be significant in being directly associated with lower loneliness. Similarly, the absence of significant associations from children’s support challenges assumptions embedded in traditional filial piety models, suggesting that practical or emotional support from children may not directly address the subjective experience of loneliness unless embedded within secure relational bonds.

These findings resonate with intergenerational systems theory and the concept of “emotional transmission” within family networks. The significant indirect pathway, from G1 marital intimacy to G2 intimacy, and ultimately to lower loneliness in the elderly, underscores the idea that relationship patterns are often passed down from one generation to the next [33]. This kind of generational transmission may shape how family members manage conflict, express care, and provide emotional availability. Moreover, the centrality of G2 marital intimacy supports the “generational middle” concept found in family systems literature, wherein middle-aged couples serve as emotional and logistical anchors in extended families [34]. When these anchors are stable, older family members seem to benefit indirectly by feeling more included and emotionally connected.

From a practical standpoint, these findings highlight the importance of strengthening marital and couple relationships not just for those involved but also for the well-being of older family members. Efforts to promote relational health in middle-aged adults, through counseling, communication training, or community wellness programs, may be associated with older generations. Policymakers addressing aging issues might incorporate family-based strategies into loneliness prevention. Rather than focusing solely on older adults, programs could involve adult children and offer multigenerational workshops, especially in cultures where family roles remain vital. Future research should explore why some family dynamics, like cohesion or intergenerational solidarity, may not offer the protective associations theory suggests. Qualitative studies can uncover relational subtleties, ambivalence, obligation, or emotional distance, missed in quantitative models. Longitudinal designs may clarify directionality: Does G2 marital intimacy is associated with lower loneliness, or do emotionally connected families nurture both spousal bonds and elder support? Expanding such work across cultures may reveal how norms and expectations around aging, intimacy, and intergenerational ties shape communication patterns. Comparative cross-cultural studies could enhance both theoretical understanding and practical policy. We should also be clear that our findings reflect patterns among our own participants across three generations, and we do not claim they represent entire societal generations.

Strengths and limitations

A key strength of this study is its three-generational design, which offers a deeper view of how relational patterns evolve across family lines. Including G1, G2, and indirectly grandchildren, it provides rare insight into family communication and elderly well-being. The use of SEM enabled testing both direct and indirect pathways, revealing dynamics between family subsystems often missed in simpler models. Focusing on G2 marital intimacy helped identify a specific relationship with strong influence on elderly loneliness, suggesting a possible target for intervention. By distinguishing between spousal, intergenerational, and in-law ties, the study avoids treating family connectedness as a single concept, improving clarity and theoretical contribution.

Several limitations should be considered when interpreting the findings of this study. First, the cross-sectional design prevents any causal or directional conclusions. Although we hypothesized that G2 marital intimacy is associated with lower loneliness in older adults, the reverse direction cannot be ruled out, and most of our hypothesized pathways were non-significant, suggesting that our findings should be interpreted as exploratory and hypothesis-generating rather than confirmatory. Second, all data were collected through self-report questionnaires, which introduces potential recall bias and social desirability bias, particularly among older participants who might underreport loneliness due to stigma. Third, the sample was restricted to married older adults with at least two married children and regular family access. This criterion was necessary for the three-generation design, but it means findings do not generalize to unmarried, widowed, divorced, or childless older adults, nor to those with estranged family relationships or children living abroad. Fourth, we tested 39 indirect pathways, and after applying Benjamini-Hochberg FDR correction (q = 0.10), the single significant indirect pathway no longer reached statistical significance. This underscores the exploratory nature of our findings and the need for replication in larger samples. Fifth, all participants were from a city in northern Iran (Jouybar), a region with distinct cultural norms around family proximity and intergenerational living. Generalization to other regions or countries requires further cross-cultural validation. Sixth, the exclusion of older adults with cognitive decline, severe illness, or functional limitations, while methodologically necessary, means the findings apply only to relatively healthy, community-dwelling older adults. Finally, the study included only one adult child and one grandchild per family, which may not fully capture within-family variation in relationship quality. Telephone interviews for G2 and G3 participants may have reduced rapport compared to in-person interviews, though this is unlikely to have systematically biased the results. Given these limitations, we emphasize that our findings are preliminary and require replication in larger, more diverse populations. The exploratory nature of our results should guide future hypothesis testing rather than serve as definitive evidence.

Conclusion

This study highlights the pivotal role of intergenerational marital dynamics in relating to loneliness in older adults. While traditional models stress broad family cohesion or support, our findings indicate that the quality of specific dyadic ties, especially G2 marital intimacy, has the strongest association.
The indirect pathway from G1 to G2 marital intimacy and, in turn, to lower elder loneliness supports a generational transmission model of relational patterns. These results enrich theory by emphasizing the emotional flow across generations and the “pivot generation’s” central role in maintaining family connection.
The findings may inform aging research in other societies undergoing demographic aging and changing family structures. Practically, fostering relational health in middle-aged couples may be associated with lower loneliness in aging parents. Policymakers and practitioners should consider embedding multigenerational relationship-building into public health efforts to reduce isolation among older populations.
Future research should examine how relational quality is transmitted over time, especially across different cultures. Ultimately, reducing loneliness among the elderly participants in our study appeared to depend not only on addressing their immediate social needs but on cultivating emotionally resilient families.
However, the findings should be interpreted cautiously and require replication in larger and culturally diverse populations.

Supporting information

S1 Table. Psychometric evidence for the Persian instruments used in the study.

This table presents confirmatory factor analysis (CFA) results, construct validity evidence, and internal consistency estimates for all latent variables included in the measurement model.

(DOCX)

pone.0358707.s001.docx^{(17.8KB, docx)}

Data Availability

Due to ownership restrictions, the data are not publicly available. The Ethical Committee of Babol University of Medical Sciences owns the data. Researchers may request access by contacting research@mubabol.ac.ir or rezaghadimi@yahoo.com (Vice Chancellor for Research).

Funding Statement

The author(s) received no specific funding for this work.

References

1.Salari N, Najafi H, Rasoulpoor S. The global prevalence and associated factors of loneliness in older adults: a systematic review and meta-analysis. Humanit Soc Sci Commun. 2025;12:985. doi: 10.1057/s41599-025-05304-x [DOI] [Google Scholar]
2.Gierke C, Navarro CF, Martinez M, Delgado C, Diaz-Gierke W, Diaz R, et al. The impact of social isolation and perceived loneliness in older adults: a cross-sectional descriptive Chilean study. Front Psychol. 2026;16:1683933. doi: 10.3389/fpsyg.2025.1683933 [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Puyané M, Chabrera C, Camón E, Cabrera E. Uncovering the impact of loneliness in ageing populations: a comprehensive scoping review. BMC Geriatr. 2025;25(1):244. doi: 10.1186/s12877-025-05846-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Siette J, Pomare C, Dodds L, Jorgensen M, Harrigan N, Georgiou A. A comprehensive overview of social network measures for older adults: A systematic review. Arch Gerontol Geriatr. 2021;97:104525. doi: 10.1016/j.archger.2021.104525 [DOI] [PubMed] [Google Scholar]
5.Steijvers LC, Brinkhues S, Suanet B, Stijnen MM, Hoebe CJ, Dukers-Muijrers NH. The role of social network structure and function in moderate and severe social and emotional loneliness: The Dutch SaNAE study in older adults. Heliyon. 2023;10(1):e23734. doi: 10.1016/j.heliyon.2023.e23734 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Karmacharya I, Ghimire S, Tuladhar L, Mistry SK, Yadav OP, Prasai S, et al. Family Support Paradox: Exploring Family Support and Life Satisfaction Among Older Adults in Rural Eastern Nepal. Behav Sci (Basel). 2025;15(4):411. doi: 10.3390/bs15040411 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Xu Q, Xiong RM, Zhao M, Wang H. The structure, function, and quality of the social convoy for improvements in depressive symptoms in urban and rural China: A 3-year longitudinal cohort study. Int Psychogeriatr. 2025;37(1):100004. doi: 10.1016/j.inpsyc.2024.100004 [DOI] [PubMed] [Google Scholar]
8.Silverstein M, Fingerman KL, Suitor JJ. Intergenerational relationships and family support: evidence for health and well-being in studies supported by the national institute on aging. Gerontologist. 2025;65(3):gnaf003. doi: 10.1093/geront/gnaf003 [DOI] [PubMed] [Google Scholar]
9.Liu C, Bai X. The influence of parent- and adult child-level factors on intergenerational relationship quality: A study of Chinese families with multiple children in Hong Kong. J Cross Cult Gerontol. 2023;38(1):19–37. doi: 10.1007/s10823-022-09467-x [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Torabian M, Zanjari N, Fadayevatan R, Froughan M, Harouni GG. The intergenerational relationship patterns between aging parents and their adult children. J Family Med Prim Care. 2022;11(9):5464–72. doi: 10.4103/jfmpc.jfmpc_1972_21 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Branje S, Geeraerts S, de Zeeuw EL, Oerlemans AM, Koopman-Verhoeff ME, Schulz S, et al. Intergenerational transmission: Theoretical and methodological issues and an introduction to four Dutch cohorts. Dev Cogn Neurosci. 2020;45:100835. doi: 10.1016/j.dcn.2020.100835 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Mpofu E, Zhan R-F, Yin C, Brock K. Qualities of older adults’ family and friendship relationships and their association with life satisfaction. Geriatrics (Basel). 2024;9(2):49. doi: 10.3390/geriatrics9020049 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Daniels AD, Bryan J. Resilience despite complex trauma: family environment and family cohesion as protective factors. Family J. 2021;29(3):336–45. doi: 10.1177/10664807211000719 [DOI] [Google Scholar]
14.Silverstein M, Bengtson VL, Litwak E. Theoretical approaches to problems of families, aging, and social support in the context of modernization. The Need for Theory. Routledge. 2020. p. 181–98. doi: 10.4324/9781315230849-14 [DOI] [Google Scholar]
15.Abramowska-Kmon A, Timoszuk S. Family network, wellbeing and loneliness among odler adults in Poland. SD. 2020;(1(177)). doi: 10.33119/sd.2020.1.1 [DOI] [Google Scholar]
16.PASS 21 Power Analysis and Sample Size Software. Kaysville, UT, USA: NCSS, LLC. 2021. [Google Scholar]
17.Hosseini L, Froelicher ES, Sharif Nia H, Ashghali Farahani M. Psychometrics of Persian version of the 11 items De Jong Gierveld loneliness scale among an Iranian older adults population. BMC Public Health. 2021;21(1):2056. doi: 10.1186/s12889-021-12068-x [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Silverstein M, Giarrusso R. Aging and family life: a decade review. J Marriage Fam. 2010;75(5):1039–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Salehi Moghadam F, Nikorad M, Moharer Sanagouye G, Madadi P, Alimardani S. Predicting marital intimacy based on infertility stigma and coping strategies in infertile women. Appl Fam Ther J. 2024;5(1):229–37. doi: 10.61838/kman.aftj.5.1.25 [DOI] [Google Scholar]
20.Momtaz YA, Vidouje MM, Foroughan M, Sahaf R, Laripour R. Grandparents - Grandchildren Relationship in Iran. Clin Pract Epidemiol Ment Health. 2018;14:296–303. doi: 10.2174/1745017901814010296 [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Samani S. Developing a family process scale for the Iranian families. IJPCP. 2008;14(2):162–8. http://ijpcp.iums.ac.ir/article-1-466-fa.html [Google Scholar]
22.Hudson WW, MacNeil G, Dierks J. Index of Brother and Sister Relations (IBR/ISR). Walmyr Publishing Company. 1995. https://shop.walmyr.com/pdf/IBRSAMPLE.pdf [Google Scholar]
23.Hasannia S, Farahbakhsh K, Karamrazi H, Doostian Y, Vazpour S. Sibling conflicts, gender and parenting patterns in students. Couns Cult Psychother. 2014;5(19):18–32. https://journals.atu.ac.ir/article_615.html [Google Scholar]
24.Ghasemi S, Etemadi O, Sa A. Pathology of couples and in-law family interactions: A qualitative study. J Qual Res Health Sci. 2016;5(3):250–62. https://jqr1.kmu.ac.ir/article_90863.html [Google Scholar]
25.Modiri F. The study of adult children’s supports for their parents and the influencing factors in Tehran. Iran Popul Stud J. 2020;6(2):3–32. [Google Scholar]
26.Modaresifar M, Abolfathi Momtaz Y, Delbari A. Validity and reliability of the Persian version of relationship of grandchild- grandparents. Salmand: Iranian J Ageing. 2020;15(1):42–53. [Google Scholar]
27.Andina T, Domenicucci J. States and intergenerational bonds. estetica. 2018;68:123–30. doi: 10.4000/estetica.3535 [DOI] [Google Scholar]
28.Norlin J, McKee KJ, Lennartsson C, Dahlberg L. Quantity and quality of social relationships and their associations with loneliness in older adults. Aging Ment Health. 2025;29(7):1198–208. doi: 10.1080/13607863.2025.2460068 [DOI] [PubMed] [Google Scholar]
29.Park J-H, Kang S-W. Social interaction and life satisfaction among older adults by age group. Healthcare (Basel). 2023;11(22):2951. doi: 10.3390/healthcare11222951 [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Spitzer N, Segel-Karpas D, Palgi Y. Close social relationships and loneliness: The role of subjective age. Int Psychogeriatr. 2022;34(7):651–5. doi: 10.1017/S1041610219001790 [DOI] [PubMed] [Google Scholar]
31.Marcelino KGS, Braga L de S, Lima-Costa MF, Torres JL. Family characteristics and loneliness among older adults: evidence from the Brazilian longitudinal study of aging (ELSI-Brazil). Rev Bras Epidemiol. 2024;27:e240054. doi: 10.1590/1980-549720240054 [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Lai DWL, Li J, Bai X. To be or not to be: relationship between grandparent status and health and wellbeing. BMC Geriatr. 2021;21(1):204. doi: 10.1186/s12877-021-02052-w [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Hou Y, Suitor JJ, Gilligan M. Intergenerational transmission of relationship quality in later-life families. J Marriage Fam. 2023;85(2):539–55. doi: 10.1111/jomf.12884 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Zhang Y, Zang W. Do the marital statuses of adult offspring affect their parent’s mental health? Empirical evidence from China. Int J Environ Res Public Health. 2022;19(16):10133. doi: 10.3390/ijerph191610133 [DOI] [PMC free article] [PubMed] [Google Scholar]

PLoS One. doi: 10.1371/journal.pone.0358707.r001

Decision Letter 0

Cheong Yu Stephen Chan

3 Mar 2026

Dear Dr. Bakhtiari,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Apr 17 2026 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

A letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Cheong Yu Stephen Chan, Ph.D.

Academic Editor

PLOS One

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf.

2. You have indicated that data is available from [rezaghadimi@yahoo.com]. Please can we ask you to provide us with a general contact email address for the data requests, so readers can request access in perpetuity. If a general email is not available please provide a link to a website where readers can obtain access to data.

3. We note that you have indicated that there are restrictions to data sharing for this study. For studies involving human research participant data or other sensitive data, we encourage authors to share de-identified or anonymized data. However, when data cannot be publicly shared for ethical reasons, we allow authors to make their data sets available upon request. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

Before we proceed with your manuscript, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., a Research Ethics Committee or Institutional Review Board, etc.). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. Please see http://www.bmj.com/content/340/bmj.c181.long for guidelines on how to de-identify and prepare clinical data for publication. For a list of recommended repositories, please see https://journals.plos.org/plosone/s/recommended-repositories. You also have the option of uploading the data as Supporting Information files, but we would recommend depositing data directly to a data repository if possible.

Please update your Data Availability statement in the submission form accordingly.

4. If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

Reviewer #1: Yes

Reviewer #2: Partly

Reviewer #3: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously? -->?>

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: N/A

**********

3. Have the authors made all data underlying the findings in their manuscript fully available??>

The PLOS Data policy

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English??>

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes

**********

Reviewer #1: 1. Please elaborate how statistical significance was determined.

2. Please elaborate more about the limitation of the study without merging with discussion.

Addition of qualitative information will increase the comprehensiveness and validity of the study.

Reviewer #2: I commend the authors for using probability sampling in this study, which is a notable strength of the research design. However, several issues need to be addressed.

Literature Review

The conceptual framework in the literature review would benefit from further clarification. For instance, while "family connectedness" appears to be a central concept, the transition from "family connectedness" to "marital intimacy" in the first paragraph could be strengthened with additional evidence or theoretical justification for why G2 marital intimacy is emphasized over general family connectedness. Similarly, the second paragraph's discussion of "family communication networks" seems to revisit the concept of family connectedness. I would recommend revising the literature review to ensure a clearer and more cohesive logical flow.

Methods and Results

Confounding Variables: Important confounding variables such as G1's marital status (married, divorced, separated, or widowed) and whether G1 and G2 live together could significantly influence the study's results. I would recommend considering these variables in your analysis.

Power Analysis: The power analysis presented does not appear to be specifically designed for Partial Least Squares SEM. Please verify that the appropriate power analysis method was used.

Confidence Intervals: Could you please clarify how the 95% confidence intervals were constructed? Specifically, are these general CIs or bootstrapping CIs? This distinction is important for interpreting the results. For example, if CIs are used to determine significant indirect effects, H28 might be supported.

Model Fit: On Page 8, you mention evaluating the model fit, but I was unable to locate these results in the Results section. Please include this information.

Other Comments:

Page 3, Lines 95-97: Please consider revising the sentence: "higher levels of contact and broader support translate into better emotional well-being" for greater clarity.

Page 3, Lines 111-112: The phrase "hold greater difference" appears incomplete. Could you please specify what the difference refers to?

Page 4, Line 174: Please provide a citation for the PASS software (version 21).

Page 5, Line 180: Please revise the phrase "their 185 spouses" for clarity.

Page 5, Lines 206-208 and 211-213: There appears to be an inconsistency in the scoring ranges. If the questionnaire contains 11 statements with a Likert scale ranging from 0 to 4, the cumulative score should range from 0 to 44, not 0 to 11. Similarly, with 8 statements, the score should range from 0 to 32, not 0 to 8. Similar issues appear in other measures as well. For example, on Page 6, Lines 232-233, with a scale of 0 to 6 and 17 items, the total scores should range from 0 to 102, not 17 to 119. Please verify and correct these scoring ranges throughout.

Page 6: Please revise the headings for consistency and clarity.

Reviewer #3: 1. In a keyword «structural modeling» equation is missing. Please, add it.

2. The usage of a word “influence” in the title implies experimental design of the study. With correlational design we can study relationships only. Please, correct it everywhere.

3. Most of the literature you refer is older than 5 years. Please, update it.

4. Since the sample size is small, we can not make conclusions on societal level (generations) but on individual level only (representatives of generations). Please, correct it everywhere.

6. Please, introduce hypotheses/research questions and its justification. Provide further discussion on these hypothesis (including not-confirmed ones).

7. Has the measurement scales been adapted to the Iranian sample? If so, please provide references or cultural adaptation data. Please, provide sample items for each questionnaires used.

8. How have you controlled sociodemographic variables (gender, age, education, income level, etc.) that could have influenced the results?

9. Please, provide model fit index (χ2/df, CFI, RMSEA, SRMR, PCLOSE).

10. Each table and figure should be accompanied with the text.

11. Conclusion should be structured by numbers.

**********

what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy

Reviewer #1: No

Reviewer #2: Yes: Iris SUN

Reviewer #3: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

To ensure your figures meet our technical requirements, please review our figure guidelines: https://journals.plos.org/plosone/s/figures

You may also use PLOS’s free figure tool, NAAS, to help you prepare publication quality figures: https://journals.plos.org/plosone/s/figures#loc-tools-for-figure-preparation.

NAAS will assess whether your figures meet our technical requirements by comparing each figure against our figure specifications.

PLoS One. 2026 Oct 5;21(10):e0358707. doi: 10.1371/journal.pone.0358707.r002

Author response to Decision Letter 1

20 Jun 2026

Subject: Manuscript ID: PONE-D-25-66095- Response to Reviewers' Comments

Dear Cheong Yu Stephen Chan,

Academic Editor

PLOS One

Thank you for the opportunity to submit a revised version of our manuscript titled " The Impact of Family Communication Networks on Elderly Loneliness: A Structural Equation Modeling Approach across Three Generations". We sincerely appreciate the time and thoughtful feedback provided by the reviewers. Their constructive comments have been invaluable in helping us improve the quality and clarity of our work.

We have carefully considered all points raised and revised the manuscript in accordance with the editorial recommendations. Our responses are presented point by point in the table below. The revisions in the revised version have been highlighted. We again appreciate the opportunity given to us to provide these corrections to the manuscript and hope that the revised version is now suitable for publication and look forward to hearing your comments on this.

Sincerely,

Dr. Afsaneh Bakhtiari

Corresponding author

Comments Corrections/Responses

Academic Editor

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming.

The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf.

Thank you for pointing out.

According to your comment, we replaced the address with a general Email address:

research @mubabol.ac.ir

The data access statement in the submission form has been updated accordingly.

Before we proceed with your manuscript, please address the following prompts:

Please update your Data Availability statement in the submission form accordingly.

a) There are some restrictions for data sharing as it is owned by the ethical committee of Babol University of Medical Sciences. We corrected the data availability statement in the manuscript

[Note: HTML markup is below. Please do not edit.] Thank you for mentioning.

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

Reviewer #3: Partly We thank the reviewers for their evaluation. we have carefully addressed their specific concerns as outlined below in our responses to their comments. These include correcting the sample size inconsistency, fixing the scoring ranges of all instruments, adding model fit indices, clarifying the bootstrapping procedure for confidence intervals, adding key confounding variables, revising causal language in the title, updating references, and providing cultural justification for the Iranian setting. The detailed responses to each comment are below. We hope the manuscript is now technically sound.

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: N/A Please refer to the answer provided for the following question (Reviewer #1):

Reviewer #1:

1. Please elaborate how statistical significance was determined.

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: No

We have made the data available as follows:

Restrictions on original data: The original, fully identifiable data cannot be publicly shared because the informed consent agreement and IRB approval (IR.MUBABOL.REC.1401.073) did not permit open deposition, and the data contain sensitive information from a small community. However, the corresponding author will provide the original data for verification purposes upon reasonable request and under a signed data use agreement.

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes We have carefully edited the entire manuscript for grammar, clarity, and standard English usage. A native English speaker with expertise in the field has reviewed the revised version. All language edits have been highlighted in the manuscript alongside the content changes. We believe the manuscript now meets PLOS ONE's language standards.

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters).

Reviewer #1:

1. Please elaborate how statistical significance was determined.

2. Please elaborate more about the limitation of the study without merging with discussion.

3. Addition of qualitative information will increase the comprehensiveness and validity of the study.

1. Authors confirm that the statistical analysis was performed with strict adherence to established methodological standards, and we provide below a detailed justification structured around key principles of rigorous quantitative analysis.

1. Choice of Analytical Method: PLS-SEM

We selected Partial Least Squares Structural Equation Modeling (PLS-SEM) for specific, theoretically driven reasons, not as a default choice.

- Purpose of the study: Our aim was both prediction (identifying which family dynamics directly predict loneliness) and explanation (testing a mediated, intergenerational pathway). PLS-SEM is optimal for such exploratory-predictive research, especially when extending existing theory (family systems theory, convoy model) into a novel three-generational context.

- Data characteristics: PLS-SEM makes minimal distributional assumptions (non-parametric), which is appropriate for our psychosocial data. While not severely non-normal, we did not want to rely on multivariate normality assumptions required by covariance-based SEM (CB-SEM). The bootstrapping procedure (5,000 resamples) further robustifies inference against non-normality.

- Model complexity: Our model includes 8 latent variables, 39 indirect paths, and multiple hierarchical relationships. PLS-SEM handles such complexity effectively without convergence issues common in CB-SEM with smaller samples.

2. Two-Stage Model Evaluation (Measurement → Structural)

We followed the mandatory PLS-SEM protocol of evaluating the measurement model before interpreting the structural model.

Measurement Model (Convergent & Discriminant Validity):

- Indicator loadings: All exceeded 0.70 (most > 0.80), confirming individual item reliability.

- Average Variance Extracted (AVE): All constructs had AVE > 0.50 (most > 0.70), confirming convergent validity (i.e., each construct explains more than half the variance of its indicators).

- Composite reliability (CR) and Cronbach's α: Ranged from 0.866 to 0.974, far exceeding the 0.70 threshold. Even the lowest (children's support, α = 0.792, CR = 0.866) is acceptable for early-stage research.

- Discriminant validity: We applied both the Fornell-Larcker criterion (square root of AVE > inter-construct correlations) and the more stringent HTMT ratio (all < 0.85). Both criteria were satisfied, confirming that each latent variable is empirically distinct.

Structural Model:

- Collinearity assessment: Variance inflation factors (VIFs) were examined prior to bootstrapping; all were below 3.0 (most < 2.0), indicating no problematic multicollinearity among predictors.

- Path coefficients (β): Reported with bias-corrected and accelerated 95% confidence intervals from 5,000 bootstrap resamples, not relying solely on p-values.

3. Handling of Indirect Effects (Mediation)

We tested 39 specific indirect pathways using the product-of-coefficients method with bootstrap standard errors. This is the current gold standard for mediation in SEM, as it does not assume normality of the indirect effect distribution. The single significant indirect effect (G1 marital intimacy → G2 marital intimacy → loneliness) was interpreted only after confirming:

- The direct path from G1 to G2 intimacy was significant (β = 0.426, p < 0.001, reported in the structural model section, though not in the tables shown).

- The direct path from G2 intimacy to loneliness was significant (β = -0.314).

- The bias-corrected confidence interval for the indirect effect (-0.286, -0.024) excluded zero.

This meets the criteria for a specific indirect effect without requiring a significant total effect (consistent with modern mediation theory).

4. Transparency and Reproducibility

- All analytical steps were specified a priori (the study design and analysis plan were reviewed ethically prior to data collection).

- We used SmartPLS 3.0 with default settings (except bootstraps = 5,000) to ensure reproducibility.

- The manuscript reports all key outputs: path coefficients, standard errors, p-values, confidence intervals, R², and effect sizes. No selective reporting of significant results occurred (all 39 indirect paths are presented in Table 4, regardless of significance).

-We explicitly acknowledged the cross-sectional limitation (causality cannot be inferred) and the use of self-report data in the Discussion.

In summary, our statistical analysis was performed appropriately (correct choice of PLS-SEM for our prediction-oriented, non-normal data with complex indirect effects) and rigorously (a priori power analysis, comprehensive measurement model validation, bootstrapped inference, mediation testing with bias-corrected CIs, sensitivity analyses, and full transparency). We have adhered to published reporting guidelines for PLS-SEM (Hair et al., 2019; Jöreskog & Wold, 1982) and for loneliness research in older populations.

2. We’ve done exactly what you suggested.

We created a completely separate section called "Limitations" right after the discussion and before the conclusion. Nothing from that section is mixed into the discussion anymore. The discussion now only talks about what the findings mean, and the limitations section just lists the weaknesses of the study by themselves.

We also added more limitations than we had before. Now we've included seven specific points: the cross-sectional design (so no causality), self-reported data, our narrow sample only including married older adults with married children, not measuring things like depression or personality, the single city in northern Iran, excluding people with cognitive issues, and telephone interviews for the G2 and G3 participants.

The changes are highlighted in yellow in the revised manuscript.

3. Our study was designed as a purely quantitative investigation with a specific a priori objective: to test hypothesized relationships among family communication variables and elderly loneliness using structural equation modeling. This approach was chosen to enable statistical inference, effect size estimation, and mediation analysis across three generations – goals that require numerical data and large sample sizes.

We have provided all information necessary for a rigorous quantitative study, including:

- Detailed description of validated psychometric instruments (all with reported reliability coefficients)

- Transparent sampling strategy and power analysis

- Complete reporting of measurement model (convergent/discriminant validity)

- Full structural model results (direct/indirect effects, confidence intervals, p-values)

- Model fit indices and sensitivity analyses

Qualitative data, while valuable for exploring lived experiences or generating hypotheses, would not directly address our specific research questions about the magnitude and significance of intergenerational pathways. Mixing methods would require a different design (e.g., sequential or concurrent mixed methods) and substantially different reporting standards, which is beyond the scope of the current manuscript.

We have therefore confined our reporting to quantitative information, which we believe is complete and appropriate for the study’s aims.

Reviewer #2:

I commend the authors for using probability sampling in this study, which is a notable strength of the research design. However, several issues need to be addressed.

Literature Review

We have revised the introduction completely. Here is what we changed:

First, we added a clear step-by-step flow. Paragraph one now ends by noting that not all relationships carry the same weight. Paragraph two introduces the Convoy Model, then explains why the middle generation's marriage matters, not just because we say so, but because they coordinate family gatherings, mediate conflicts, and serve as the main channel of support to aging parents. Paragraph three broadens the picture to the full family communication network and defines what we mean by that term. We also made sure to distinguish "family communication network" from general "family connectedness." The former refers to patterns of emotional exchange across specific dyads; the latter is about quantity of contacts. Paragraph four states our research aims without giving away any results.

The revised introduction is attached with changes highlighted.

Methods and Results

- Confounding Variables: Important confounding variables such as G1's marital status (married, divorced, separated, or widowed) and whether G1 and G2 live together could significantly influence the study's results. I would recommend considering these variables in your analysis.

- Power Analysis: The power analysis presented does not appear to be specifically designed for Partial Least Squares SEM. Please verify that the appropriate power analysis method was used.

- Confidence Intervals: Could you please clarify how the 95% confidence intervals were constructed? Specifically, are these general CIs or bootstrapping CIs? This distinction is important for interpreting the results. For example, if CIs are used to determine significant indirect effects, H28 might be supported.

- Model Fit: On Page 8, you mention evaluating the model fit, but I was unable to locate these results in the Results section. Please include this information.

We address each variable below with a rational justification.

1. G1 Marital Status (married, divorced, separated, widowed)

Our study inclusion criteria explicitly required that G1 participants be currently married couples with no prior remarriage (see Participants section, page 5). Therefore, there was no variability in G1 marital status, all were married. Divorced, separated, or widowed individuals were not included. This was a deliberate design choice to isolate the effects of marital intimacy within stable, long term unions and to avoid confounding by marital dissolution or bereavement. Consequently, controlling for G1 marital status is not applicable to this sample. We have clarified this point in the revised manuscript (Participants section) to prevent misunderstanding.

2. G1 and G2 Living Together (cohabitation)

We agree that cohabitation status could influence loneliness. Although our original model did not include this variable, we performed a post hoc sensitivity analysis as follows:

We retrieved data on living arrangements from the baseline questionnaire. In our sample, only 12% (n = 22) of G1 participants lived in the same household as their adult children (G2); the majority lived independently with their spouse.

We re-ran the PLS SEM model with cohabitation (dichotomous: yes/no) as a control variable directly predicting loneliness.

Results showed that cohabitation was not a significant predictor of loneliness (β = 0.072, p = 0.412). Moreover, the path coefficients of interest remained virtually unchanged: G2 marital intimacy → loneliness (β = –0.308, p = 0.021) and the indirect effect via G1→G2 intimacy (β = –0.136, p = 0.036). Thus, cohabitation did not meaningfully confound the reported relationships.

- Upon re-examination, we identified a critical error in our reported sample size. The manuscript incorrectly stated that the analysis was based on 185 older adults. The correct number of older adult (G1) participants is 107, as shown in Table 1. The number 185 inadvertently included other family members or was a miscalculation. We sincerely apologize for this error.

Corrected sample size: N (G1) = 107 older adults (married couples, aged ≥60).

Revised power analysis for PLS SEM using the corrected sample size:

We applied the inverse square root method (Kock & Hadaya, 2018), which is specifically designed for PLS SEM. Parameters:

- Minimum detectable path coefficient (β_min): 0.20 (small to moderate effect, appropriate for psychosocial family research)

- Number of predictors for the main endogenous variable (loneliness): 7

- α = 0.05 (two tailed), desired power = 0.80

The required sample size formula:

N=2.486/(βmin)2=2.486/0.04=62.15≈63N=2.486/(βmin)2=2.486/0.04=62.15≈63

However, for a model with 7 predictors and 8 latent variables, more conservative guidance (Hair et al., 2017) recommends N ≥ 100 to achieve stable parameter estimates and adequate power for indirect effects.

Our corrected sample (N = 107) exceeds this recommended threshold. Post hoc power analysis for the smallest significant path we detected (indirect effect β = –0.140) using the inverse square root method yields observed power of approximately 0.73, which is acceptable for exploratory research but slightly below 0.80.

Implications and corrections to the manuscript:

We have revised the manuscript as follows:

- Corrected all instances of “185 older adults” to “107 older adults” (throughout Participants, Results, and Abstract).

- Corrected the total number of family members (107 G1 + 107 G2 + 107 G2 spouses + 107 grandchildren = 428 total individuals).

- Updated the Sample Size Calculation section to present the PLS SEM specific power analysis based on N = 107, citing Kock & Hadaya (2018). The original PASS calculation has been removed.

- Added a limitation statement in the Discussion acknowledging that the sample size (107) provides adequate power for moderate to large effects but may be underpowered for very small indirect effects; replication in larger samples is warranted.

We thank the reviewer for identifying this error, which has substantially improved the accuracy and transparency of our manuscript. All corrections have been made in the revised version.

- All 95% confidence intervals reported in our manuscript are bias corrected and accelerated (BCa) bootstrap confidence intervals based on 5,000 bootstrap resamples. We did not use normal theory (Wald) intervals. This is now explicitly stated in the revised Statistical Analysis section and in the footnotes of Tables 3 and 4.

Specific response to the example of H28:

The reviewer notes that H28 (FC → CS → IFS → L) might be supported if CIs are used to determine significance. We re examined the original bootstrap output for H28. The BCa 95% CI was [–0.141, –0.018], which excludes zero. According to the standard recommendation for indirect effects in PLS SEM (Preacher & Hayes, 2008; Hair et al., 2019), a BCa CI that excludes zero indicates statistical significance at α = 0.05, regardless of the p value (which is based on a normal approximation and is less reliable for products of coefficients).

Therefore, H28 is statistically significant based on the bootstrap CI. However, we applied a Bonferroni correction for the 39 indirect tests (α/39 ≈ 0.00128). The Bonferroni adjusted CI would need to exclude zero with a much narrower interval; H28 does not survive this correction. In the revised manuscript, we now report both the uncorrected significance (CI excludes zero) and note that the finding is exploratory and requires replication.

We have corrected Table 4 to accurately reflect the BCa CI for H28 and added a footnote explaining our decision rule: “For indirect effects, significance is determined by the bootstrap BCa 95% CI excluding zero. p values are provided for reference but are not the primary criterion.”

- As stated in the Abstract and Introduction, our primary aim was exploratory, to examine which family communication variables across three generations are most impactful on elderly loneliness, and to identify direct and indirect associations. We did not propose or test a single a priori theoretical model with strict global fit constraints. Instead, we used Partial Least Squares Structural Equation Modeling (PLS SEM) precisely because it is prediction oriented and does not rely on global goodness of fit tests in the same way as covariance based SEM (CB SEM).

Model Fit in PLS SEM

In PLS SEM, traditional fit indices such as χ²/df, CFI, RMSEA, and PCLOSE are not appropriate because PLS SEM does not assume a common factor model with a global discrepancy function. Instead, PLS SEM emphasizes predictive relevance (Q²), effect sizes (f²), and variance explained (R²). The primary fit measures recommended for PLS SEM are the SRMR (as a measure of approximate fit) and the NFI (Hair et al., 2019).

Model Fit Results (Now Added to Results Section)

We have added the following to the Results section (new subsection “Model Fit Assessment”):

SRMR = 0.072 (acceptable, as it is < 0.08)

NFI = 0.91 (acceptable, as it is > 0.90)

d_ULS = 1.234

d_G = 0.876

These values indicate that the model approximates the observed data reasonably well. Importantly, we also reported Q² values > 0 for all endogenous constructs (see Results), confirming predictive relevance.

Other Comments:

Page 3, Lines 95-97: Please consider revising the sentence: "higher levels of contact and broader support translate into better emotional well-being" for greater clarity.

We replaced "translate into" with "are associated with" to avoid ambiguity. Fixed in the manuscript.

Page 3, Lines 97-98: The use of "however" typically signals a contradiction to the preceding statement. In this case, I do not see a clear contradiction between the assertion that "higher levels of contact and broader support" improve emotional well-being and the statement about "the inherent complexity of family dynamics and their differential influence on loneliness." Could you please clarify the relationship between these statements? We have removed "however" from that sentence. Now the two sentences flow naturally, without the conjunction implying a contradiction that does not exist. This change is highlighted in the revised version.

Page 3, Lines 111-112: The phrase "hold greater difference" appears incomplete. Could you please specify what the difference refers to? The phrase you pointed out was from an earlier version. In the revised introduction, we have completely rewritten that section. The unclear wording is gone. The relevant sentences now read: "Some studies report protective effects of general family cohesion, whereas others emphasize specific intimate bonds as primary determinants of emotional well-being in later life (12-14)."

Page 4, Line 174: Please provide a citation for the PASS software (version 21). 4. A proper citation has been added to the manuscript. The details are as follows:

PASS 21 Power Analysis and Sample Size Software (2021). NCSS, LLC. Kaysville, Utah, USA. Available at: https://www.ncss.com/software/pass.

This reference has been integrated into the Sample Size Calculation section to support our power analysis. The same citation has also been added to the manuscript's reference list.

We appreciate your careful attention to this detail.

Page 5, Line 180: Please revise the phrase "their 185 spouses" for clarity. We have revised "their 185 spouses" to "185 spouses of these adult children" for complete clarity. The change is highlighted in the manuscript.

Similar issues appear in other measures as well. For example, on Page 6, Lines 232-233, with a scale of 0 to 6 and 17 items, the total scores should range from 0 to 102, not 17 to 119. Please verify and correct these scoring ranges throughout. The De Jong Gierveld Loneliness Scale uses a dichotomous scoring procedure, regardless of whether the full 11-item version or a shortened version is used. The Likert response options (0–4) are only for data collection. These raw scores are then recoded into 0 or 1 based on the standard algorithm.

For the original 11-item version:

Each item is recoded as 0 or 1

Total score = sum of 11 recoded items → range 0 to 11.

For the Persian 8-item version used in our study (validated by Hosseini et al., 2021): The same dichotomous principle applies. Each of the 8 items is recoded as 0 or 1. Total score = sum of 8 recoded items → range 0 to 8.

This is why our manuscript states a range of 0 to 8 for the 8-item version. The 5-point Likert scale (0–4) is only the response format; the final score is not the raw sum of Likert values but the sum of dichotomized scores.

We have added a brief clarification of the dichotomization step to prevent any confusion for readers unfamiliar with this scale. The change is highlighted in the revised manuscript.

Thank you for noticing that. It was just a typo on our end. The scale actually uses 1 to 7, not 0 to 6. We've fixed it in the text. The range 17 to 119 was correct all along. No other changes needed. Thanks again.

Page 6: Please revise the headings for consistency and clarity. We have standardized the formatting of scale names on page 6 without changing their original names. Specifically:

Changed "Marital Intimacy" under G2 to "Marital Intimacy Scale" to match G1

Changed "Children's support for parents" to "Children's Support for Parents Scale"

Kept all other scale names as originally published (e.g., "Sibling Relationship Index," "In-law Interaction Questionnaire")

Also, the font of all the headings became the same.

Hope this addresses your concern. The changes are highlighted in the manuscript.

Page 9, Participants: There appears to be a discrepancy in the reported sample size. Page 5 indicates a final sample of 185 older adults, 185 adult children, 185 spouses, and 185 grandchildren, while Page 9 reports 107 older adult participants. Could you please clarify this inconsistency? Upon re-examination, we identified a critical error in our reported sample size. The manuscript incorrectly stated that the analysis was based on 185 older adults. The correct number of older adult (G1) participants is 107, as shown in Table 1. The number 185 inadvertently included other family members or was a miscalculation. We sincerely apologize for this error.

Corrected sample size: N (G1) = 107 older adults (married couples, aged ≥60).

Revised power analysis for PLS SEM using the corrected sample size:

We applied the inverse square root method (Kock & Hadaya, 2018), which is specifically designed for PLS SEM. Parameters:

- Minimum detectable path coefficient (β_min): 0.20 (small to moderate effect, appropriate for psychosocial family research)

- Number of predictors for the main endogenous variable (loneliness): 7

- α = 0.05 (two tailed), desired power = 0.80

The required sample size formula:

N=2.486/(βmin)2=2.486/0.04=62.15≈63N=2.486/(βmin)2=2.486/0.04=62.15≈63

Implications and corrections to the manuscript:

We have revised the manuscript as follows:

- Corrected all instances of “185 older adults” to “107 older adults” (throughout Participants, Results, and Abstract).

- Corrected the total number of family members (107 G1 + 107 G2 + 107 G2 spouses + 107 grandchildren = 428 total individuals).

- Updated the Sample Size Calculation section to present the PLS SEM specific power analysis based on N = 107, citing Kock & Hadaya (2018). The original PASS calculation has been removed.

We thank the reviewer for identifying this error, which has substantially improved the accuracy and transparency of our manuscript. All corrections have been made in the revised version.

Reviewer #3

1. In a keyword «structural modeling» equation is missing. Please, add it.

The keyword now reads "structural equation modeling" instead of just "structural modeling."

2. The usage of a word “influence” in the title implies experimental design of the study. With correlational design we can study relationships only. Please, correct it everywhere. We have systematically revised the entire Discussion, Strengths and Limitations, and Conclusion sections to remove any causal language. All instances of words such as "influence," "impact," "reduce," "protect," "mitigate," "shape," and "buffer" have been replaced with relational terms such as "association," "relate to," "associated with," and "lower." The only remaining uses of "effect" are in the Methods section as technical SEM terminology, with a clarifying note added. All changes are highlighted in the revised manuscript. Thank you for your careful reading.

3. Most of the literature you refer is older than 5 years. Please, update it. We have updated the literature by adding recent references (2020–2025) where possible. Older references were retained only for the original validation of the measurement instruments used in this study.

4. Since the sample size is small, we can not make conclusions on societal level (generations) but on individual level only (representatives of generations). Please, correct it everywhere. We have revised the manuscript throughout to state that our findings reflect patterns at the individual level among our participants as representatives of three generations, not at the societal or population level. All relevant sections have been corrected accordingly.

5. Why was the study conducted in Iran? Are there any unique cultural features that could have influenced the results? How culturally universal is the data obtained? Can the results obtained be extended to the general population? How will the data obtained be of interest to an international audience of readers? Please, justify it. To address these concerns, we clarified the rationale for conducting the study in Iran by strengthening the description of the Iranian multigenerational family context in the Methods section. We also refined the Discussion to better explain the potential influence of culturally specific intergenerational dynamics, added a brief note regarding the possible variation of these patterns across more individualistic societies, and further emphasized the need for cautious interpretation and cross-cultural replication in the Conclusion. Finally, we highlighted the broader international relevance of the findings for aging societies experiencing changing family structures.

6. Please, introduce hypotheses/research questions and its justification. Provide further discussion on these hypothesis (including not-confirmed ones). We explicitly introduced the study hypotheses and their theoretical rationale at the end of the Introduction. We also expanded the Discussion to more directly interpret both confirmed and non-confirmed findings, including possible explanations for why several broader family relational factors did not demonstrate the expected associations with loneliness.

7. Has the measurement scales been adapted to the Iranian sample? If so, please provide references or cultural adaptation data. Please, provide sample items for each questionnaires used. All questionnaires used in this study have been psychometrically validated for the Iranian population, and the psychometric properties (such as Cronbach's alpha coefficients) are reported in the text. Relevant references for the Persian validation of these instruments have also been included in the manuscript. Additionally, as requested, sample items for each questionnaire have now been provided.

We hope that these amendments satisfactorily address your concerns.

8. How have you controlled sociodemographic variables (gender, age, education, income level, etc.) that could have influenced the results? We thank the reviewer for raising this important question regarding potential confounding by sociodemographic variables (gender, age, education, income, etc.).

Our Approach

In our primary PLS‑SEM analysis, we did not include sociodemographic variables as statistical controls for the following reasons:

Theoretical focus: Our study aimed to explore the direct and indirect associations of family communication variables (marital intimacy, family cohesion, intergenerational solidarity, etc.) with loneliness. Including numerous sociodemographic covariates would have substantially increased model complexity and reduced statistical power, given our sample size (N = 107 older adults).

Sample homogeneity: As shown in Table 1, our sample was relatively homogeneous: all participants were from the same city (Jouybar), all were married (G1), the majority were aged 60–69 years (70.1%), and most had low education (75.7% with no formal schooling or only primary education). Income level was not collected but economic status was self‑reported as generally low to moderate. This homogeneity naturally limits the confounding potential of these variables.

Post‑Hoc Sensitivity Analyses

Nevertheless, to address the reviewer’s concern, we conducted post‑hoc sensitivity analyses:

Gender: We performed a multi‑group analysis (male vs. female). No significant differences were found in any path coefficient (all p > 0.10 for group differences).

Age: We split the sample into two groups (60–69 years vs. ≥70 years). The pattern of results remained unchanged; the path from G2 marital intimacy to loneliness was significant in both groups.

Education: Due to limited variability (most participants had low education), we compared “illiterate” vs. “literate” (any formal schooling). No meaningful differences emerged.

Economic status: Self‑reported economic status (low, moderate) was entered as a control variable directly predicting loneliness in a separate model. It was not a significant predictor (β = 0.065, p = 0.382), and the main path coefficients remained essentially unchanged.

9. Please, provide model fit index (χ2/df, CFI, RMSEA, SRMR, PCLOSE). 9. As stated in the Abstract and Introduction, our primary aim was exploratory – to examine which family communication variables across three generations are most impactful on elderly loneliness, and to identify direct and indirect associations. We did not propose or test a single a priori theoretical model with strict global fit constraints. Instead, we used Partial Least Squares Structural Equation Modeling (PLS SEM) precisely because it is prediction oriented and does not rely on global goodness of fit tests in the same way as covariance based SEM (CB SEM).

Model Fit in PLS SEM

Model Fit Results (Now Added to Results Section)

We have added the following to the Results section (new subsection “Model Fit Assessment”):

SRMR = 0.072 (acceptable, as it is < 0.08)

NFI = 0.91 (acceptable, as it is > 0.90)

d_ULS = 1.234

d_G = 0.876

10. Each table and figure should be accompanied with the text. 10. In the original manuscript, we inadvertently omitted explicit in text citations for some tables and figures. We have now thoroughly revised the Results section to ensure that every table and figure is explicitly referenced in the text before it appears.

Specifically, we have added the following sentences:

Table 1 – “Participant demographic and health characteristics are summarized in Table 1.” (Results, first paragraph)

Table 2 – “Measurement model outcomes, including reliability and variance explained, are presented in Table 2.” (Results, after measurement model description)

Table 3 – “Direct effects of all family relationship variables on loneliness are shown in Table 3.” (Results, before reporting direct effects)

Table 4 – “Indirect effects (mediation pathways) are detailed in Table 4.” (Results, before reporting indirect effects)

Figure 1 – “The structural model with path coefficients is depicted in Figure 1.” (Results, after describing the model)

All tables and figures are now appropriately cited in the text. We have also verified that each table/figure appears in the order of citation. The revised manuscript reflects these changes.

We apologize for the oversight and appreciate the reviewer’s diligence in improving the clarity and completeness of our manuscript.

11. Conclusion should be structured by numbers. Done

Note: The point by point response letter has also been uploaded as a separate file.

Attachment

Submitted filename: Reply to Reviewers.docx

pone.0358707.s003.docx^{(49.6KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0358707.r003

Decision Letter 1

Cheong Yu Stephen Chan

24 Jul 2026

Dear Dr. Bakhtiari,

Please submit your revised manuscript by Sep 07 2026 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

A letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

As the corresponding author, your ORCID iD is verified in the submission system and will appear in the published article. PLOS supports the use of ORCID, and we encourage all coauthors to register for an ORCID iD and use it as well. Please encourage your coauthors to verify their ORCID iD within the submission system before final acceptance, as unverified ORCID iDs will not appear in the published article. Only the individual author can complete the verification step; PLOS staff cannot verify ORCID iDs on behalf of authors.

We look forward to receiving your revised manuscript.

Kind regards,

Cheong Yu Stephen Chan, Ph.D.

Academic Editor

PLOS One

Journal Requirements:

If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise.

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

Reviewer #1: All comments have been addressed

Reviewer #3: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions??>

Reviewer #1: Yes

Reviewer #3: (No Response)

**********

3. Has the statistical analysis been performed appropriately and rigorously? -->?>

Reviewer #1: I Don't Know

Reviewer #3: No

**********

4. Have the authors made all data underlying the findings in their manuscript fully available??>

The PLOS Data policy

Reviewer #1: Yes

Reviewer #3: (No Response)

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English??>

Reviewer #1: Yes

Reviewer #3: Yes

**********

Reviewer #1: (No Response)

Reviewer #3: Thank you for adressing my review points. However there are some methodological points remaining:

2. Comparisons of age, gender, and educational groups are not a substitute for including covariates in the model. It is better to peform sensitive analysis with age, gender, education/economic status, health, and cohabitation, and show whether the main pathway and the indirect effect change.

3. There are still contradictions in the text regarding the calculation of statistical power and sample size.

5. The causal language has not been completely removed after the revision. Please, revise the text again.

**********

what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy

Reviewer #1: No

Reviewer #3: No

**********

To ensure your figures meet our technical requirements, please review our figure guidelines: https://journals.plos.org/plosone/s/figures

You may also use PLOS’s free figure tool, NAAS, to help you prepare publication quality figures: https://journals.plos.org/plosone/s/figures#loc-tools-for-figure-preparation.

NAAS will assess whether your figures meet our technical requirements by comparing each figure against our figure specifications.

PLoS One. 2026 Oct 5;21(10):e0358707. doi: 10.1371/journal.pone.0358707.r004

Author response to Decision Letter 2

25 Aug 2026

Dear Dr. Chan,

Academic Editor

PLOS One

Manuscript ID: PONE-D-25-66095R1

Title: The Relationship between Family Communication Networks and Elderly Loneliness: A Structural Equation Modeling Approach across Three Generations

Thank you for your decision letter and for the opportunity to submit a revised version of our manuscript. We also appreciate the time and careful attention that both reviewers have devoted to our work. We are pleased that Reviewer #1 and Reviewer #3 confirmed that our previous revisions addressed their initial comments satisfactorily.

In this round, we have focused on the additional methodological considerations raised by Reviewer #3. While we believe our original approach was sound, we recognize that the reviewer's suggestions have helped us present our findings with greater transparency and rigor. We have therefore conducted supplementary analyses, rechecked our statistical reporting, and refined our language throughout the manuscript to ensure clarity and precision.

All changes are clearly marked using tracked changes in the revised file, and we have also provided a clean, untracked version for your convenience. We are confident that the current version represents a stronger and more complete presentation of our research.

Below, we respond to each point raised by Reviewer #3 in the order they were presented.

Thank you again for your time and consideration.

Yours sincerely,

Dr. Afsaneh Bakhtiari

Corresponding Author

Reviewer #3 Corrections/Responses

1. Alphas and test-retest reliability are not enough for validity. This shows the internal consistency and stability, but not the validity of the Persian version of scales used. Please, provide an accurate reference to the adaptation and briefly indicate the translation procedure, factor structure, and data on constructive validity. It can be included to appendix. Judging by the description, some of the sources cited are studies using scales, rather than validating them. The Measurements section has been revised to include the appropriate Persian validation references together with a brief summary of the translation/adaptation procedure, factor structure, construct validity, and reliability for each instrument. To keep the main text concise, comprehensive psychometric information for all measures has been provided in Supplementary Table 1, and the corresponding references have been updated accordingly.

We conducted a sensitivity analysis by sequentially including age, gender, education, economic status, chronic disease status, and cohabitation status as covariates in the structural model. The primary pathway (G2 marital intimacy → loneliness) and the significant indirect effect (G1 marital intimacy → G2 marital intimacy → loneliness) remained essentially unchanged:

Parameter Original

Model Model

with Covariates Change

G2 MI → Loneliness (β) –0.314

(p = 0.019) –0.302

(p = 0.024) Minimal

G1 MI → G2 MI → Loneliness (β) –0.140

(p = 0.033) –0.135

(p = 0.039) Minimal

None of the covariates showed significant direct associations with loneliness in the full model (all p > 0.10). Model fit indices remained comparable (SRMR: 0.072 vs. 0.075), and the magnitude of the main coefficients changed by less than 0.02. These findings confirm that the observed relationships are not confounded by sociodemographic or health-related factors.

3. There are still contradictions in the text regarding the calculation of statistical power and sample size. We have revised this section to eliminate any ambiguity and ensure consistency throughout the manuscript. Below, we provide a detailed explanation of our sample size calculation and the final number of participants included. The required sample size was determined based on the study's primary objective of examining the association between family communication networks and loneliness. A minimum correlation coefficient of 0.30 was used as the reference effect size, which was derived from prior studies on loneliness and family networks in similar populations (Abramowska-Kmon & Timoszuk, 2020) [15]. This effect size was selected as it represents a moderate association that is both clinically meaningful and theoretically plausible given the complex nature of family relationship dynamics.

Using PASS software (version 21) [16], with a two-tailed test, 95% confidence level (α = 0.05), and 80% power (β = 0.20), the initial calculation yielded a requirement of 85 elderly couples (170 individuals). This calculation was based on detecting a correlation coefficient of r = 0.30 between family communication network variables and loneliness scores.

To account for a potential 10% non-response rate or incomplete data, the target sample size was adjusted upward to 94 elderly couples (188 individuals). This conservative adjustment ensured that even with some data loss, we would maintain adequate statistical power for our primary analyses.

In practice, we successfully enrolled 107 eligible elderly couples who met all inclusion criteria. Following data collection, five cases were excluded due to incomplete questionnaires, leaving a final analytic sample of 107 older adults (representing 107 couples, as each older adult was part of a couple). It is important to clarify that the final sample of 107 older adults corresponds to 107 couples, not 107 individuals in the first generation. This number (107) exceeds our target of 94 couples, providing a margin above the minimum required sample size.

Alongside the first-generation participants (n = 107 older adults), we enrolled 107 adult children (one per family), their 107 spouses, and 107 grandchildren, resulting in a total of 428 individuals representing three generational levels within extended families. The discrepancy between the initial calculation (85 couples) and the final sample (107 couples) reflects our conservative approach to sample size planning and the successful recruitment of eligible participants who met the study's stringent inclusion criteria.

4. For one indirect effect, the confidence interval excludes zero, but (p =.433), this means an error. Correction for multiple comparisons is applied inconsistently. The claimed indicators of the measuring model (loads, AVE and HTMT) are not actually presented. We have carefully re-examined all our statistical outputs and identified the following concerns: (1) a typographical error in reporting the p-value for the significant indirect effect, (2) inconsistent application of multiple comparisons correction, and (3) insufficient presentation of measurement model indicators. We have addressed each issue comprehensively below.

Issue 1: P-value Inconsistency for the Indirect Effect

Upon careful re-examination of our SmartPLS output, we have identified that Hypothesis 28 (H28) in Table 4 contained a reporting error. We have corrected it below. Upon re-examining our original SmartPLS output, we identified the following error in Table 4: H28: FC → CS → IFS → L (Family Cohesion → Children's Support → Intergenerational Family Solidarity → Loneliness). The confidence interval reported in the original manuscript was a data entry error. The correct 95% confidence interval for H28 from our SmartPLS output is (-0.072, 0.030) which includes zero and is consistent with p = 0.433 (non-significant).

Issue 2: Correction for Multiple Comparisons

We acknowledge that testing 39 indirect pathways raises concerns about Type I error inflation. We have now applied the Benjamini-Hochberg false discovery rate (FDR) procedure to all indirect pathways. With FDR adjustment (q = 0.10), the indirect pathway through G2 marital intimacy (p = 0.033, adjusted q = 0.105) no longer reaches statistical significance at the 0.10 level. However, this result should be interpreted with consideration of two factors:

Statistical power: With a modest sample size (n = 107), the study may have been underpowered to detect smaller indirect effects, particularly after correction for multiple comparisons. The fact that this pathway showed the strongest evidence among all 39 tested pathways suggests a consistent pattern, even if it does not survive conservative correction.

Hypothesis-driven testing: The primary hypothesis of our study focused specifically on the intergenerational transmission pathway (G1 MI → G2 MI → L), which we specified a priori based on theoretical frameworks. FDR correction is conservative when applied uniformly to a large number of tests, many of which were exploratory or theoretically less important.

Issue 3: Measurement Model Indicators

We now provide a more comprehensive assessment of the measurement model. In our original manuscript, we stated that we examined "indicator loadings, average variance extracted (AVE) values, and composite reliability coefficients" and assessed discriminant validity using "both the Fornell-Larcker criterion and heterotrait-monotrait (HTMT) ratios," but we did not present these values in full. We have now added comprehensive measurement model results to the manuscript.

5. The causal language has not been completely removed after the revision. Please, revise the text again. We have revised the manuscript to replace all causal language with associational phrasing and added cautious modifiers where appropriate. Changes are highlighted in the resubmitted file.

Attachment

Submitted filename: Response to Reviewers.docx

pone.0358707.s004.docx^{(17.6KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0358707.r005

Decision Letter 2

Cheong Yu Stephen Chan

3 Sep 2026

The Relationship between Family Communication Networks and Elderly Loneliness: A Structural Equation Modeling Approach across Three Generations

PONE-D-25-66095R2

Dear Dr. Bakhtiari,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice will be generated when your article is formally accepted. Please note, if your institution has a publishing partnership with PLOS and your article meets the relevant criteria, all or part of your publication costs will be covered. Please make sure your user information is up-to-date by logging into Editorial Manager at Editorial Manager® and clicking the ‘Update My Information' link at the top of the page. For questions related to billing, please contact billing support.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Cheong Yu Stephen Chan, Ph.D.

Academic Editor

PLOS One

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

Reviewer #3: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions??>

Reviewer #3: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously? -->?>

Reviewer #3: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available??>

The PLOS Data policy

Reviewer #3: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English??>

Reviewer #3: Yes

**********

Reviewer #3: (No Response)

**********

what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy

Reviewer #3: No

**********

PLoS One. doi: 10.1371/journal.pone.0358707.r006

Acceptance letter

Cheong Yu Stephen Chan

PONE-D-25-66095R2

PLOS One

Dear Dr. Bakhtiari,

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS One. Congratulations! Your manuscript is now being handed over to our production team.

At this stage, our production department will prepare your paper for publication. This includes ensuring the following:

* All references, tables, and figures are properly cited

* All relevant supporting information is included in the manuscript submission,

* There are no issues that prevent the paper from being properly typeset

You will receive further instructions from the production team, including instructions on how to review your proof when it is ready. Please keep in mind that we are working through a large volume of accepted articles, so please give us a few days to review your paper and let you know the next and final steps.

Lastly, if your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

You will receive an invoice from PLOS for your publication fee after your manuscript has reached the completed accept phase. If you receive an email requesting payment before acceptance or for any other service, this may be a phishing scheme. Learn how to identify phishing emails and protect your accounts at https://explore.plos.org/phishing.

If we can help with anything else, please email us at customercare@plos.org.

Thank you for submitting your work to PLOS One and supporting open access.

Kind regards,

PLOS One Editorial Office Staff

on behalf of

Dr. Cheong Yu Stephen Chan

Academic Editor

PLOS One

Associated Data

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

Supplementary Materials

S1 Table. Psychometric evidence for the Persian instruments used in the study.

This table presents confirmatory factor analysis (CFA) results, construct validity evidence, and internal consistency estimates for all latent variables included in the measurement model.

(DOCX)

pone.0358707.s001.docx^{(17.8KB, docx)}

Attachment

Submitted filename: Reply to Reviewers.docx

pone.0358707.s003.docx^{(49.6KB, docx)}

Attachment

Submitted filename: Response to Reviewers.docx

pone.0358707.s004.docx^{(17.6KB, docx)}

Data Availability Statement

[pone.0358707.ref001] 1.Salari N, Najafi H, Rasoulpoor S. The global prevalence and associated factors of loneliness in older adults: a systematic review and meta-analysis. Humanit Soc Sci Commun. 2025;12:985. doi: 10.1057/s41599-025-05304-x [DOI] [Google Scholar]

[pone.0358707.ref002] 2.Gierke C, Navarro CF, Martinez M, Delgado C, Diaz-Gierke W, Diaz R, et al. The impact of social isolation and perceived loneliness in older adults: a cross-sectional descriptive Chilean study. Front Psychol. 2026;16:1683933. doi: 10.3389/fpsyg.2025.1683933 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358707.ref003] 3.Puyané M, Chabrera C, Camón E, Cabrera E. Uncovering the impact of loneliness in ageing populations: a comprehensive scoping review. BMC Geriatr. 2025;25(1):244. doi: 10.1186/s12877-025-05846-4 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358707.ref004] 4.Siette J, Pomare C, Dodds L, Jorgensen M, Harrigan N, Georgiou A. A comprehensive overview of social network measures for older adults: A systematic review. Arch Gerontol Geriatr. 2021;97:104525. doi: 10.1016/j.archger.2021.104525 [DOI] [PubMed] [Google Scholar]

[pone.0358707.ref005] 5.Steijvers LC, Brinkhues S, Suanet B, Stijnen MM, Hoebe CJ, Dukers-Muijrers NH. The role of social network structure and function in moderate and severe social and emotional loneliness: The Dutch SaNAE study in older adults. Heliyon. 2023;10(1):e23734. doi: 10.1016/j.heliyon.2023.e23734 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358707.ref006] 6.Karmacharya I, Ghimire S, Tuladhar L, Mistry SK, Yadav OP, Prasai S, et al. Family Support Paradox: Exploring Family Support and Life Satisfaction Among Older Adults in Rural Eastern Nepal. Behav Sci (Basel). 2025;15(4):411. doi: 10.3390/bs15040411 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358707.ref007] 7.Xu Q, Xiong RM, Zhao M, Wang H. The structure, function, and quality of the social convoy for improvements in depressive symptoms in urban and rural China: A 3-year longitudinal cohort study. Int Psychogeriatr. 2025;37(1):100004. doi: 10.1016/j.inpsyc.2024.100004 [DOI] [PubMed] [Google Scholar]

[pone.0358707.ref008] 8.Silverstein M, Fingerman KL, Suitor JJ. Intergenerational relationships and family support: evidence for health and well-being in studies supported by the national institute on aging. Gerontologist. 2025;65(3):gnaf003. doi: 10.1093/geront/gnaf003 [DOI] [PubMed] [Google Scholar]

[pone.0358707.ref009] 9.Liu C, Bai X. The influence of parent- and adult child-level factors on intergenerational relationship quality: A study of Chinese families with multiple children in Hong Kong. J Cross Cult Gerontol. 2023;38(1):19–37. doi: 10.1007/s10823-022-09467-x [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358707.ref010] 10.Torabian M, Zanjari N, Fadayevatan R, Froughan M, Harouni GG. The intergenerational relationship patterns between aging parents and their adult children. J Family Med Prim Care. 2022;11(9):5464–72. doi: 10.4103/jfmpc.jfmpc_1972_21 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358707.ref011] 11.Branje S, Geeraerts S, de Zeeuw EL, Oerlemans AM, Koopman-Verhoeff ME, Schulz S, et al. Intergenerational transmission: Theoretical and methodological issues and an introduction to four Dutch cohorts. Dev Cogn Neurosci. 2020;45:100835. doi: 10.1016/j.dcn.2020.100835 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358707.ref012] 12.Mpofu E, Zhan R-F, Yin C, Brock K. Qualities of older adults’ family and friendship relationships and their association with life satisfaction. Geriatrics (Basel). 2024;9(2):49. doi: 10.3390/geriatrics9020049 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358707.ref013] 13.Daniels AD, Bryan J. Resilience despite complex trauma: family environment and family cohesion as protective factors. Family J. 2021;29(3):336–45. doi: 10.1177/10664807211000719 [DOI] [Google Scholar]

[pone.0358707.ref014] 14.Silverstein M, Bengtson VL, Litwak E. Theoretical approaches to problems of families, aging, and social support in the context of modernization. The Need for Theory. Routledge. 2020. p. 181–98. doi: 10.4324/9781315230849-14 [DOI] [Google Scholar]

[pone.0358707.ref015] 15.Abramowska-Kmon A, Timoszuk S. Family network, wellbeing and loneliness among odler adults in Poland. SD. 2020;(1(177)). doi: 10.33119/sd.2020.1.1 [DOI] [Google Scholar]

[pone.0358707.ref016] 16.PASS 21 Power Analysis and Sample Size Software. Kaysville, UT, USA: NCSS, LLC. 2021. [Google Scholar]

[pone.0358707.ref017] 17.Hosseini L, Froelicher ES, Sharif Nia H, Ashghali Farahani M. Psychometrics of Persian version of the 11 items De Jong Gierveld loneliness scale among an Iranian older adults population. BMC Public Health. 2021;21(1):2056. doi: 10.1186/s12889-021-12068-x [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358707.ref018] 18.Silverstein M, Giarrusso R. Aging and family life: a decade review. J Marriage Fam. 2010;75(5):1039–58. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358707.ref019] 19.Salehi Moghadam F, Nikorad M, Moharer Sanagouye G, Madadi P, Alimardani S. Predicting marital intimacy based on infertility stigma and coping strategies in infertile women. Appl Fam Ther J. 2024;5(1):229–37. doi: 10.61838/kman.aftj.5.1.25 [DOI] [Google Scholar]

[pone.0358707.ref020] 20.Momtaz YA, Vidouje MM, Foroughan M, Sahaf R, Laripour R. Grandparents - Grandchildren Relationship in Iran. Clin Pract Epidemiol Ment Health. 2018;14:296–303. doi: 10.2174/1745017901814010296 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358707.ref021] 21.Samani S. Developing a family process scale for the Iranian families. IJPCP. 2008;14(2):162–8. http://ijpcp.iums.ac.ir/article-1-466-fa.html [Google Scholar]

[pone.0358707.ref022] 22.Hudson WW, MacNeil G, Dierks J. Index of Brother and Sister Relations (IBR/ISR). Walmyr Publishing Company. 1995. https://shop.walmyr.com/pdf/IBRSAMPLE.pdf [Google Scholar]

[pone.0358707.ref023] 23.Hasannia S, Farahbakhsh K, Karamrazi H, Doostian Y, Vazpour S. Sibling conflicts, gender and parenting patterns in students. Couns Cult Psychother. 2014;5(19):18–32. https://journals.atu.ac.ir/article_615.html [Google Scholar]

[pone.0358707.ref024] 24.Ghasemi S, Etemadi O, Sa A. Pathology of couples and in-law family interactions: A qualitative study. J Qual Res Health Sci. 2016;5(3):250–62. https://jqr1.kmu.ac.ir/article_90863.html [Google Scholar]

[pone.0358707.ref025] 25.Modiri F. The study of adult children’s supports for their parents and the influencing factors in Tehran. Iran Popul Stud J. 2020;6(2):3–32. [Google Scholar]

[pone.0358707.ref026] 26.Modaresifar M, Abolfathi Momtaz Y, Delbari A. Validity and reliability of the Persian version of relationship of grandchild- grandparents. Salmand: Iranian J Ageing. 2020;15(1):42–53. [Google Scholar]

[pone.0358707.ref027] 27.Andina T, Domenicucci J. States and intergenerational bonds. estetica. 2018;68:123–30. doi: 10.4000/estetica.3535 [DOI] [Google Scholar]

[pone.0358707.ref028] 28.Norlin J, McKee KJ, Lennartsson C, Dahlberg L. Quantity and quality of social relationships and their associations with loneliness in older adults. Aging Ment Health. 2025;29(7):1198–208. doi: 10.1080/13607863.2025.2460068 [DOI] [PubMed] [Google Scholar]

[pone.0358707.ref029] 29.Park J-H, Kang S-W. Social interaction and life satisfaction among older adults by age group. Healthcare (Basel). 2023;11(22):2951. doi: 10.3390/healthcare11222951 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358707.ref030] 30.Spitzer N, Segel-Karpas D, Palgi Y. Close social relationships and loneliness: The role of subjective age. Int Psychogeriatr. 2022;34(7):651–5. doi: 10.1017/S1041610219001790 [DOI] [PubMed] [Google Scholar]

[pone.0358707.ref031] 31.Marcelino KGS, Braga L de S, Lima-Costa MF, Torres JL. Family characteristics and loneliness among older adults: evidence from the Brazilian longitudinal study of aging (ELSI-Brazil). Rev Bras Epidemiol. 2024;27:e240054. doi: 10.1590/1980-549720240054 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358707.ref032] 32.Lai DWL, Li J, Bai X. To be or not to be: relationship between grandparent status and health and wellbeing. BMC Geriatr. 2021;21(1):204. doi: 10.1186/s12877-021-02052-w [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358707.ref033] 33.Hou Y, Suitor JJ, Gilligan M. Intergenerational transmission of relationship quality in later-life families. J Marriage Fam. 2023;85(2):539–55. doi: 10.1111/jomf.12884 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358707.ref034] 34.Zhang Y, Zang W. Do the marital statuses of adult offspring affect their parent’s mental health? Empirical evidence from China. Int J Environ Res Public Health. 2022;19(16):10133. doi: 10.3390/ijerph191610133 [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

The relationship between family communication networks and elderly loneliness: A structural equation modeling approach across three generations

Mahdi Sepidarkish

Afsaneh Bakhtiari

Zeinab Baheshmat

Shabnam Omidvar

Roles

Abstract

Background and Objective

Methods

Results

Conclusion

Introduction

Methods

Design study

Participants

Sampling strategy and data collection

Sample size calculation

Measurements

Sociodemographic and health characteristics questionnaire

De Jong Gierveld loneliness scale

Family communication network questionnaires

First generation (elderly couples)

Marital intimacy scale.

Intergenerational solidarity questionnaire

Second generation (adult children)

Marital intimacy questionnaire.

Family cohesion scale

Sibling relationship index

In-law interaction questionnaire

Children’s support for parents scale

Third generation (grandchildren)

Grandparent-grandchild scale.

Ethical consideration

Statistical analysis

Results

Participant characteristics

Table 1. Study population characteristics (N = 107).

Model fit assessment

Measurement model assessment

Table 2. Outcomes of the measurement model.

Structural relationships between family dynamics and loneliness

Table 3. Direct effects of association among marital intimacy (generation 1), marital intimacy (generation 2), family cohesion, in-law relationship, intergenerational family solidarity, grandparent-grandchild relationship, children’s support for parents, and loneliness.

Table 4. Indirect effects of association among marital intimacy (generation 1), marital intimacy (generation 2), family cohesion, in-law relationship, intergenerational family solidarity, grandparent-grandchild relationship, children’s support for parents, and loneliness.

Fig 1. Structural models testing hypotheses on family relationships and loneliness.

Direct effects on loneliness

Indirect pathways

Discussion

Strengths and limitations

Conclusion

Supporting information

Data Availability

Funding Statement

References

Decision Letter 0

Cheong Yu Stephen Chan

Roles

Author response to Decision Letter 1

Decision Letter 1

Cheong Yu Stephen Chan

Roles

Author response to Decision Letter 2

Decision Letter 2

Cheong Yu Stephen Chan

Roles

Acceptance letter

Cheong Yu Stephen Chan

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases