The purpose of this study is to examine four specific sequential mediation models to determine the distinct roles of workplace social connectedness (WSC) and PCB in sequentially mediating the relationships between the four WHI sub-dimensions (representing both conflict and enrichment) and employee FL.

The global workplace has undergone a significant transformation since the COVID-19 pandemic. For knowledge workers, hybrid and flexible work models have become a defining feature of the ‘new normal’ (Hou & Sing, 2025), allowing work to be done from virtually anywhere. However, these models blur boundaries between professional and personal life, presenting both opportunities and challenges (Singh et al., 2022). Within the context of the 5IR, organisations must balance productivity with employee well-being (Antonaci et al., 2024). Work–home interaction remains a critical issue, as knowledge workers navigate competing demands from work and home (Song et al., 2023).

Work–family conflict is a form of interrole conflict that occurs when the demands, time, or stress from work and family roles are incompatible, so that fulfilling the requirements of one role makes it difficult to fulfil the requirements of the other (Greenhaus & Beutell, 1985). While WFC specifically captures the conflict-based or family-oriented aspect of this interaction, the broader construct of WHI encompasses both negative (interference–conflict) and positive (enrichment–facilitation) spillover between work and home domains (Geurts et al., 2005). Consequently, for the present study, WHI is adopted as the overarching construct to capture the full range of bidirectional interactions between work and home life.

Accordingly, the construct is operationalised using the Survey Work–Home Interaction Nijmegen (SWING), which measures four sub-dimensions: negative work-to-home interaction, negative home-to-work interaction, positive work-to-home interaction, and positive home-to-work interaction.

Emerging evidence links specifically WFC to adverse psychological (e.g. anxiety, emotional exhaustion), work-related (e.g. reduced innovation, thriving), and physical outcomes (e.g. fatigue, poor health) among knowledge workers (Song et al., 2023; Wang et al., 2022). Drawing on the Conservation of Resources (COR) theory (Hobfoll et al., 2018), WHI can be understood as a process of resource depletion, whereby competing demands across work and home domains are associated with reduced psychological and emotional resources, potentially undermining employees’ capacity to sustain engagement, creativity, and ultimately, flourishing (Wang et al., 2022). However, the COR theory also posits that individuals actively seek to acquire, protect, and build resources to counteract such losses. In this regard, SC and PCB represent critical relational and personal resources that may buffer the negative effects of WHI. Understanding how knowledge workers sustain flourishing despite WHI, particularly through social and personal resources such as SC and PCB, is therefore critical.

Social connectedness reflects an individual’s sense of belonging, closeness, and perceived connection to others (Lee et al., 2001). As a relational resource, SC provides emotional and instrumental support that fosters resilience and well-being (Holt-Lunstad, 2018). In workplace contexts, SC helps employees alleviate stressors and manage work–family demands by building supportive networks and trust (Winslow et al., 2019).

The rise of hybrid and remote work, however, has disrupted traditional patterns of social interaction.

Knowledge workers often report increased isolation, weaker informal connections, and fewer opportunities for spontaneous interaction with colleagues (Charalampous et al., 2019). These challenges can undermine SC, thereby reducing access to important relational resources and increasing vulnerability to stressors such as WHI. In line with the COR perspective, diminished SC may constrain individuals’ ability to access emotional and instrumental support, exacerbating the resource depletion associated with WHI (Kossek et al., 2011).

Conversely, when SC is maintained or strengthened, it is associated with greater access to support and more effective mobilisation of adaptive coping strategies. Social connectedness, therefore, plays a crucial role in how employees manage WHI and sustain flourishing in the new world of work.

Positive coping behaviour encompasses proactive cognitive and behavioural strategies that enable individuals to manage stress constructively (Coetzee et al., 2017; Marx, 2017). More broadly, PCB is conceptualised as comprising higher-order strengths: positive cognition (inventive problem-solving), positive emotion (engaging affect), intrinsic motivation (self-efficacious, goal-directed behaviour), and positive social behaviour (extroversion and social support) (Coetzee et al., 2017). Individuals who exhibit these strengths are better positioned to manage demands, respond to challenges, and align their behaviour with situational requirements (Niessen et al., 2018).

In line with the COR theory (Hobfoll et al., 2018), individuals invest personal resources, such as positive coping strategies, to protect themselves from stress and enhance their overall well-being. Positive coping behaviours, such as problem-solving, emotional regulation, and goal-directed action, have been associated with improved flourishing, resilience, and well-being (Gurvich et al., 2021; Waters et al., 2022). In the context of the ‘new normal’, where job demands are high and boundaries are blurred, PCB may play an important role in supporting resilience (Leask & Ruggunan, 2021).

Importantly, PCB does not operate in isolation but is often associated with relational resources such as SC (Gawrych, 2024). A supportive social environment is associated with greater access to resources and may be linked to the use of more adaptive coping strategies (Gawrych, 2024). Although various personal and psychological resources have been identified as buffering the effects of WHI, this study focuses on SC and PCB as they capture complementary relational and behavioural mechanisms particularly relevant to hybrid work contexts (Dailey et al., 2022). While alternative resources primarily reflect individual capacities, SC represents the social context and PCB the strategies used to manage demands, together providing an integrated social–behavioural pathway for sustaining well-being.

Although SC and PCB have been recognised individually as protective resources (Dailey et al., 2022; De Sousa et al., 2024), their sequential roles in mediating the relationship between WHI and FL remain largely unexplored. Most prior studies have examined these constructs in isolation (e.g. Brown & Leite, 2022; Harunavamwe & Ward, 2022) or as parallel mechanisms (Chen et al., 2025; Rahim et al., 2020), thereby limiting our understanding of the dynamic, process-based pathways through which WHI influences positive outcomes.

From a COR perspective, resources do not operate independently but tend to cluster and reinforce one another (Hobfoll et al., 2018). In this context, SC represents a relational resource that provides access to emotional and instrumental support, while PCB represents a personal resource associated with how individuals manage stressors. Importantly, relational resources such as SC may precede and be associated with the use of personal coping strategies by creating a supportive context in which individuals feel equipped to engage in adaptive behaviours.

Empirical evidence indicates that positive coping strategies, including problem-focused and adaptive emotion-focused approaches, are associated with improved well-being under WHI conditions (Rahim et al., 2020).

Similarly, SC has been linked to reduced strain and enhanced well-being and is associated with greater utilisation of adaptive coping strategies (Drummond et al., 2017; French et al., 2018). However, while existing studies suggest that social support may be associated with coping, these relationships are typically examined as parallel or moderated effects, rather than as part of a sequential or order-specific process (Zeng et al., 2024).

Accordingly, this study proposes that SC and PCB operate in a sequential mediation pathway, in which WHI is associated with SC, which, in turn, is associated with PCB and ultimately relates to flourishing. This ordering is theoretically meaningful, as it reflects the process through which access to relational resources may enable the mobilisation of personal coping resources, thereby helping individuals offset resource loss and maintain optimal functioning. Examining this sequence provides a comprehensive understanding of how knowledge workers sustain flourishing amid competing work and life demands and offers an empirically grounded basis for designing targeted, resource-based interventions.

Knowledge workers in the post-pandemic ‘new world of work’ face heightened risks of WHI, with significant implications for their well-being (Peters et al., 2022; Winkler-Titus et al., 2025). Although flourishing is a key positive outcome, its links to WHI remain underexplored. Social connectedness and PCB are recognised as protective resources that may mitigate the adverse effects of WHI and are theorised to operate sequentially in relation to flourishing. This ordering is theoretically meaningful, as SC represents a foundational relational resource associated with greater access to support, which in turn may facilitate the utilisation of adaptive coping strategies captured through PCB. Sequential mediation, therefore, provides a framework for examining how these resources may operate in an order-specific pathway, in which SC is associated with PCB, and both are related to flourishing.

The conceptual framework for this research (see Figure 1) illustrates the sequential mediation of WSC and PCB in the relationship between the four WHI sub-dimensions and employee FL. This integrated model captures the pathways for both conflict (NWHI, NHWI) and enrichment (PWHI, PHWI) sub-dimensions. Focusing on South African knowledge workers in hybrid and digitally mediated contexts, the study addresses calls for more holistic approaches to employee well-being. It offers insight into sustaining flourishing amid competing work–home demands.

Based on the conceptual framework (see Figure 1), the following hypotheses are proposed. Each central hypothesis (H1–H4) is tested through its direct path to flourishing, individual indirect paths via WSC and PCB, and the sequential indirect path (WHI → WSC → PCB → FL):

Both directions are included in this study to provide a comprehensive assessment of cross-domain dynamics.

Although the sources of interference or enrichment may differ across domains, both are theorised to relate to flourishing through similar underlying resource-based mechanisms (Hobfoll et al., 2018). Specifically, whether demands originate from work or home, they may influence access to relational resources (e.g. social connectedness) and the utilisation of personal coping strategies, which are associated with well-being outcomes. Including both directions, therefore, allows for a more holistic examination of how work–home dynamics are linked to flourishing through shared social and behavioural pathways.

This study adopted a quantitative, cross-sectional survey design. A quantitative approach is suitable for this research because it enables data collection to test hypotheses and establish the statistical significance of relationships between variables.

The cross-sectional survey design involves collecting data from a sample of individuals at a single point in time (Saunders et al., 2023). This approach is both practical and cost-effective for examining the proposed mediation model, as it enables analysis of direct and indirect relationships among the four WHI sub-dimensions, the mediators, and FL (see Figure 1). However, a limitation is that cross-sectional mediation is constrained to data collected at a single measurement occasion. As such, it assumes that the causal processes unfold entirely within that timeframe and are unaffected by prior states or occurrences of the variables under investigation (Cain et al., 2018).

This study utilised convenience sampling to recruit respondents from the South African labour market. The target population comprised knowledge workers with specialised expertise who engage in knowledge-intensive activities, applying and sharing both explicit and tacit knowledge to drive organisational performance, innovation, and competitive advantage (Dong et al., 2025). The inclusion criteria were permanent employment and relevant work experience; employment in knowledge-intensive roles and industries; use of specialised expertise in knowledge creation and application; and an age range of 18–65 years. Approximately 13 000 invitations were sent, resulting in 427 completed responses, which represents a response rate of 3.3%.

The research respondents were evenly distributed by gender, with 49.6% female and 49.6% male respondents.

A small percentage (0.8%) did not specify their gender. The majority of respondents were in the older age brackets, with 42.6% aged 46–55 years, followed by 25.5% in the 36–45 years age group. Together, these two age groups accounted for over two-thirds of the total sample. In terms of professional standing, the respondents predominantly held high-level positions, including senior management (37%) and top management/executive roles (29.7%), accounting for over 60% of the sample. The sample was concentrated in a few broad industry sectors, with the largest representation from manufacturing, engineering, and industrial sectors (n = 109), followed by consumer-facing sectors such as retail, media, hospitality, and food services (n = 54), primary and natural resource industries including agriculture and mining (n = 53), and financial services (n = 49). A smaller but still notable representation was observed from construction and infrastructure, healthcare and life sciences, and technology and telecommunications sectors. Overall, the sample comprised primarily experienced, high-level professionals, with an even gender split, a concentration in middle-to-older age groups, and broad representation across diverse industries.

The following psychometrically sound measurement instruments were used to collect data:

Following the identification of a suitable gatekeeper, the research questionnaires were uploaded onto LimeSurvey, which automatically generated a URL for access. This URL, accompanied by a participant information sheet, was then emailed to the gatekeeper. The gatekeeper distributed the invitation to participate via email to prospective participants who met the predetermined inclusion criteria. Respondents could access the survey directly via the provided URL, where their responses were captured in a secure online environment. Data collection spanned a 3-month period, during which two reminder emails were sent to increase response rates.

The data analysis was conducted using IBM Statistical Package for Social Science (SPSS) Statistics, version 30.0 (IBM, 2025, Armonk, New York, United States), with the statistical significance set at p ≤ 0.05.

The data were screened in accordance with standard survey research protocols, including checks for missing values, unresponsive cases, and distributional characteristics such as skewness and kurtosis (Pallant, 2020).

No missing values or unresponsive cases were identified in the dataset. Outliers were assessed using Mahalanobis distance, with particular attention to natural breaks in the D² values, where extreme cases would typically appear distinctly separate from the remaining observations (Byrne, 2010). No cases were identified as extreme.

Given that all variables were measured using self-report Likert-type scales, the potential for common method bias (CMB) was assessed. Harman’s single-factor test was conducted by performing an exploratory factor analysis on all measured items to determine whether a single factor accounted for the majority of the variance (Podsakoff et al., 2024). The results indicated that the first factor accounted for 23.14% of the total variance, which is below the commonly accepted threshold of 50%, suggesting that CMB is unlikely to pose a significant concern in this study. Descriptive statistics summarised participant demographics and item-level responses.

Confirmatory factor analysis was conducted in Analysis of Moment Structures (AMOS) using the maximum likelihood estimator, which is considered robust under conditions of approximate normality and moderate-to-large sample sizes (Byrne, 2010). Confirmatory factor analysis was used to assess construct validity and model fit for three competing structures: Model 1 (unidimensional), Model 2 (first-order), and Model 3 (second-order). Model fit was assessed using multiple indices: the Chi-square statistic (χ²), Comparative Fit Index (CFI), Tucker–Lewis Index (TLI), Root Mean Square Error of Approximation (RMSEA), and Standardised Root Mean Square Residual (SRMR).

Acceptable model fit was indicated by CFI and TLI values ≥ 0.90, while values ≥ 0.95 suggested a good fit (Hair et al., 2019). Root Mean Square Error of Approximation and SRMR values ≤ 0.08 were considered indicative of a reasonable model fit, with values ≤ 0.05 representing a good fit (Hu & Bentler, 1999).

The internal consistency of the scales was assessed using Cronbach’s alpha (α), with coefficients ≥ 0.70 indicating acceptable reliability (DeVellis & Thorpe, 2022). Pearson’s product-moment correlation coefficient (r) was calculated to investigate the bivariate relationships between the variables. The strength of the correlations was interpreted in line with Cohen’s (1988) guidelines: coefficients between r = 0.10 and 0.29 were considered small, between r = 0.30 and 0.49 medium, and r ≥ 0.50 large.

Lastly, a serial multiple mediation analysis was conducted using Hayes’s PROCESS macro (Model 6) to test the hypothesised causal chain (Hayes, 2018). This model was used to investigate whether the independent variable (WHI) predicts (or explains the variation in) the dependent variable (FL) through a sequential pathway involving two mediators: WSC (M₁) and PCB (M₂).

Indirect effects were assessed using bias-corrected bootstrapping (5000 resamples), with significance determined if the 95% confidence interval (lower-level confidence interval [LLCI]–upper-level confidence interval [ULCI]) did not include zero (Hayes, 2018).

The study was conducted in accordance with established ethical guidelines, and formal ethical clearance was obtained from the research institution (reference number: 2021_CRERC_016(FA)). Participation was entirely voluntary, and participants retained the right to withdraw at any stage without consequence. Informed consent was obtained from all participants. To safeguard participant rights, strict measures were implemented to ensure confidentiality and support anonymity. The anonymous nature of the online data collection process guaranteed that no personally identifiable information could be linked to individual responses.

The descriptive statistics for the four constructs and their sub-dimensions are summarised in Table 1.

Firstly, low mean scores for Negative Home–Work Interaction (WI_NHWI) (M = 1.46; SE = 0.029) and WI_NWHI (M = 2.06; SE = 0.036) indicate that the respondents perceived low levels of negative spillover between their home and work domains. Conversely, the higher mean scores on the Positive Work-Home Interaction (WI_PWHI) (M = 2.50; SE = 0.039) and Positive Home-Work Interaction (WI_PHWI) (M = 2.68; SE = 0.045) subscales indicate that positive interactions and enrichment between the work and home domains were perceived as more prevalent than negative spillover.

Secondly, the overall mean score for the Informal Workplace Social Interactions (WSC) scale was 3.77 (maximum 5; SE = 0.038), indicating that the respondents generally engage in and experience positive feelings from informal social interactions at work. The mean score of 3.73 (SE = 0.047) on the behavioural (WSC_B) subscale suggests that respondents frequently participate in informal social interactions with colleagues. The even higher mean of 3.81 (SE = 0.039) for the affective (WSC_A) subscale indicates that the respondents not only engage in these interactions but also experience strong positive feelings as a result. These results suggest that these informal social bonds are a significant source of emotional well-being and positive affect in the workplace.

The analysis of the PCBI indicates that the respondents generally have favourable self-evaluations of their coping behaviours (maximum 7; M = 4.77; SE = 0.046). Specifically, the highest mean score was for intentional coping (PCBI_INT) (M = 4.97; SE = 0.050), suggesting participants frequently use deliberate, purposeful behaviours to manage stress. High scores were also observed for inventive (PCBI_INV) (M = 4.85; SE = 0.051) and influential coping (PCBI_INF) (M = 4.67; SE = 0.057), indicating that participants are proactive in using creative problem-solving skills and actively shaping their environment. While slightly lower, the mean for engaging coping (PCBI_ENG) (M = 4.60; SE = 0.058) suggests frequent use of strategies such as seeking social support and actively confronting problems (Coetzee et al., 2017).

Lastly, the FL shows a very high mean score (maximum 7; M = 5.77; SE = 0.050), indicating that, on average, respondents report high levels of psychological flourishing.

A CFA was conducted to assess construct validity and model fit before testing the mediation models. Given the multidimensional nature of the scales and the strong intercorrelations among first-order dimensions, a second-order factor structure (Model 3) was prioritised as a theoretically coherent and parsimonious representation of the data. The underlying dimensions reflect related facets of broader latent constructs, and prior literature similarly conceptualises them as higher-order domains. Thus, the second-order model was selected to retain multidimensionality while reducing redundancy among highly related subdimensions.

Although the initial first-order model showed acceptable fit in some areas, inspection of modification indices and standardised residuals identified a limited number of items across the SWING, WSC, and PCBI scales that demonstrated persistent localised misfit (standardised residuals > 2.50). Item refinement was guided by both empirical evidence and theoretical judgement, with only a few items showing persistent misfit, conceptual redundancy, or limited contribution to the construct’s core meaning considered for removal. Care was taken to ensure that theoretically central dimensions remained adequately represented, thereby preserving acceptable construct coverage.

Following refinement, the retained items continued to reflect the constructs’ central content domains, and all final scales demonstrated acceptable reliability and composite reliability (CR). Potential bias from item deletion was also evaluated through changes in factor structure, standardised loadings, and construct interpretation, with no substantive distortion observed.

The final second-order model (Model 3) demonstrated good fit: χ²/df = 2.22, RMSEA = 0.05, SRMR = 0.06, CFI = 0.91, and TLI = 0.90, providing a robust basis for the subsequent sequential mediation analysis (Table 2).

The psychometric properties of the scales were assessed by examining their internal consistency and CR. Cronbach’s alpha (α) and CR values for all constructs exceeded the recommended threshold of 0.70 (DeVellis & Thorpe, 2022; Hair et al., 2019), indicating acceptable to excellent reliability. Specifically, the WI_NWHI and WI_NHWI subscales demonstrated strong reliability (α = 0.92, CR = 0.92; α = 0.83, CR = 0.83), while the WI_PWHI and WI_PHWI subscales also showed satisfactory reliability (α = 0.78, CR = 0.78; α = 0.89, CR = 0.89). The WSC scale demonstrated good reliability (α = 0.79, CR = 0.79), the PCBI showed excellent reliability (α = 0.93, CR = 0.93), and the FL scale exhibited excellent reliability (α = 0.94, CR = 0.94).

To determine the magnitude and direction of the bivariate relationships between the study variables, Pearson’s product-moment correlations were calculated. Given acceptable skewness and kurtosis values and a large sample size, the assumptions for parametric correlation analysis were considered met (Field, 2024). The results of the correlational analysis are presented in Table 3.

As hypothesised, significant negative correlations were observed between negative WHI and several key variables. Specifically, the WI_NWHI correlated negatively with WSC (r = −0.16, small practical effect size, p < 0.01) and PCB (r = −0.11, small practical effect size, p < 0.01), while WI_NHWI also had a significant negative correlation with PCB (r = −0.15, small practical effect size, p < 0.01). The relatively stronger (though still small) negative associations were observed with FL, where the WI_NWHI (r = −0.26, small practical effect size, p < 0.01) and the WI_NHWI (r = −0.28, small practical effect size, p < 0.01) both showed moderate negative correlations.

Conversely, WI_PWHI was positively associated with WSC (r = 0.18, small practical effect size, p < 0.01) while WI_PHWI was non-significant (r = 0.07, small practical effect size, p = 0.06). Small, but significant positive relationships were observed between WI_PWHI and PCB (r = 0.08, small practical effect size, p < 0.05), as well as between WI_PHWI and PCB (r = 0.08, small practical effect size, p < 0.05). Furthermore, both WI_PWHI and WI_PHWI were positively associated with FL (WI_PWHI: r = 0.17, small practical effect size, p < 0.01; WI_PHWI: r = 0.17, small practical effect size, p < 0.01). Lastly, WSC correlated positively with PCB (r = 0.14; small practical effect size; p < 0.01) and FL (r = 0.25; small practical effect size; p < 0.01).

Overall, the correlations were small in magnitude, indicating that although the relationships are statistically significant, their practical effects may be modest and should be interpreted with caution.

A serial multiple mediation analysis (Model 6) was conducted to examine whether SC (M₁) and PCB (M₂) sequentially mediate the effect of WHI (X) on flourishing (Y). As the SWING instrument captures four distinct types of WHI, four separate mediation models were estimated using the PROCESS macro. The corresponding results are summarised in Table 4 and Table 5.

For WI_NWHI, the direct effect on flourishing was statistically significant (B = –0.18, SE = 0.06, p = 0.004, 95% CI [–0.31, –0.06]). This indicates that higher levels of WI_NWHI were associated with lower FL. Although the individual effect sizes were relatively small, the overall model explained a moderate proportion of variance in FL (R² = 0.29, F[6, 420] = 28.36, p < 0.001). The total indirect effect of WI_NWHI via WSC and PCB was not significant (b = –0.05, BootSE = 0.03, 95% CI [–0.12, 0.01]), as the confidence interval included zero. However, upon examining the specific pathways, a significant negative indirect path was identified through WSC (Ind1: WSC → FL, b = −0.03, 95% CI [−0.06, −0.00]), suggesting that negative WHI was associated with lower WSC, which in turn was associated with slightly lower flourishing.

For WI_NHWI, the direct effect was also statistically significant (B = –0.27, SE = 0.08, p < 0.001, 95% CI [–0.42, –0.12]), indicating that WI_NHWI was associated with lower levels of FL. This model again explained a moderate proportion of variance (R² = 0.29, F[6, 420] = 28.36, p < 0.001).

The total indirect effect was not significant. However, the analysis of the specific paths revealed a significant negative indirect effect through PCB (Ind2: PCB → FL, b = −0.08, 95% CI [−0.154, −0.003]), suggesting that negative WHI was associated with lower PCB, which in turn was associated with lower levels of FL. Notably, the hypothesised sequential pathway (Ind3: WSC → PCB → FL) was not significant for either negative subdimension (NWHI and NHWI).

For WI_PWHI, the total indirect effect was not significant (b = 0.06, 95% CI [−0.013, 0.129]), yet two specific indirect paths were significantly positive. The single-step path via WSC was significant (Ind1: WSC → FL, b = 0.04, 95% CI [0.009, 0.077]), indicating that when work positively impacts home life, it promotes WSC, which then contributes to FL. Crucially, the sequential mediation pathway (Ind3: WSC → PCB → FL) was also found to be significantly positive (b = 0.01, 95% CI [0.001, 0.023]), although the effect size was small. This suggests that WH_PWHI was associated with higher WSC, which in turn was associated with PCB and, subsequently, with FL. This effect was specific to this subdimension and should be interpreted with caution.

Finally, for WI_PHWI, none of the indirect pathways tested, including the sequential pathway, were found to be statistically significant, indicating that WSC and PCB do not serve as channels through which positive home experiences translate into work-related flourishing.

Based on these findings, H₃ is supported because the specific serial hypothesis for PWHI was confirmed. H₁, H₂ and H₄ were not supported (see Figure 2).

The results revealed several key patterns. Firstly, the respondents reported relatively low levels of negative WHI (WI_NWHI: M = 1.46) and comparatively higher levels of positive interactions (PHWI: M = 2.68). They also frequently engaged in informal workplace social interactions (WSC_B: M = 3.73) and experienced strong positive emotions from these interactions (WSC_A: M = 3.81), highlighting WSC as an essential relational resource of well-being. In addition, respondents reported favourable coping behaviours (PCBI: M = 4.77), particularly intentional coping (PCBI_INT: M = 4.97), alongside favourable levels of flourishing (M = 5.77).

This pattern indicates a generally resource-rich sample, which may partly explain the relatively small effect sizes observed across the analyses.

Secondly, both WI_NWHI and WI_NHWI were negatively associated with flourishing (r = −0.26 and r = −0.28, respectively), supporting prior evidence that conflict between work and home domains erodes well-being (Bian & Mohd Sukor, 2024; Darouei & Pluut, 2021). However, these effects were small, suggesting that although such interference is detrimental, many knowledge workers retain resources that buffer against its full impact. Furthermore, WSC and PCB were positively associated with each other (r = 0.14) and with flourishing (r = 0.26 and r = 0.43, respectively), establishing the necessary bivariate links for mediation.

Thirdly, the findings provided partial support for the hypothesised sequential mediation model, specifically confirming the hypothesised pathway for WI_PWHI (H₃). This suggests that when positive experiences spill over from work into the home domain, they are associated with higher levels of SC, which in turn is associated with coping behaviour and, ultimately, flourishing. However, the magnitude of this indirect effect was small (b = 0.01), and the pathway was not observed for other WHI dimensions. As such, this finding should be interpreted as context-specific rather than indicative of a generalisable mechanism.

While the sequential hypothesis was supported for WI_PWHI, it was not supported for the remaining WHI sub-dimensions. For the two negative sub-dimensions (WI_NWHI and WI_NHWI), the findings indicate that conflict-related effects on flourishing operate through more immediate, single-step mechanisms rather than the proposed sequential process. These single-step pathways are summarised below:

Finally, none of the tested indirect pathways, including the sequential model, were significant for PHWI. This suggests that positive spillover from the home domain into the work domain may operate through mechanisms not captured by the present mediators (WSC and PCB), or that its positive influence on flourishing is primarily direct.

At a theoretical level, this study contributes to the literature by examining the potential interplay between relational and personal resources within a single framework. While prior research has often considered these constructs independently, the present findings suggest that their sequential operation may be more limited and context-dependent than theoretically assumed. This highlights the importance of adopting a more nuanced perspective when applying resource-based frameworks, such as the COR theory, to the work–home interface.

At a practical level, this study offers several recommendations. However, given the small effect sizes and limited mediation effects observed, these should be interpreted as potential areas for consideration rather than definitive prescriptions, and as part of broader organisational efforts to support employee well-being:

For organisational well-being initiatives, these findings suggest that, in addition to measuring flourishing, organisations may benefit from also monitoring key associated resources such as WSC and PCB. Tracking these resources may help practitioners identify potential areas of vulnerability and inform proactive support strategies aimed at maintaining employee well-being.

Several limitations are acknowledged. Firstly, the study utilised a convenience sampling approach with a relatively low response rate (3.3%), which may introduce self-selection and non-response bias. Participants may differ systematically from non-participants, potentially affecting the representativeness of the sample and the observed relationships. In addition, the sample consisted predominantly of knowledge workers, limiting generalisability to broader occupational groups (Andrade, 2021), particularly those in less autonomous roles who may experience WHI differently.

Secondly, the use of self-report measures may introduce common method variance (Podsakoff et al., 2024) and social desirability bias (Durmaz et al., 2020), potentially inflating associations between constructs. Although Harman’s single-factor test suggested that CMB was not a substantial concern, this approach has recognised limitations, and the potential for method-related variance cannot be entirely excluded.

Thirdly, refining the measurement model by removing several items while improving model fit may affect content validity. Although both statistical and theoretical considerations guided this process, the potential reduction in construct coverage should be considered when interpreting the findings.

A further limitation concerns the alignment between the conceptual and operational definitions of key constructs. While WSC and PCB are broadly defined, the instruments used primarily capture interaction-based connectedness and cognitive–behavioural coping. Accordingly, the findings should be interpreted within the scope of these measured dimensions rather than the broader constructs.

Future research should build on this study by employing longitudinal or experimental designs to examine causal and dynamic relationships among the study variables. Investigating moderators, such as leadership style, organisational support, boundary management preferences, and demographic or cultural factors, may clarify the conditions under which social and personal resources are most relevant. Expanding mediators (e.g. resilience, mindfulness) and outcomes (e.g. performance, commitment) may further enrich the model, while cross-cultural and comparative studies could assess generalisability. Additionally, examining technological influences, including digital support systems, remote work intensity, and technostress, may provide further insight into flourishing in digitally mediated work environments. Finally, future studies may consider evaluating targeted organisational interventions aimed at strengthening social and personal resources.