Methods

This research was descriptive-correlational. The statistical population included military men and women at risk of emotional divorce. Based on Green’s formula, the sufficient sample size for modeling was calculated as 234 people; however, considering possible flawed questionnaires and to improve external generalizability, the sample size was set at 300. With the cooperation of a clinic affiliated with the Iranian Psychological and Counseling Organization, the required sample was selected via purposive convenience sampling from clients whose individual treatments had ended. The city of Tehran was divided into regions, three regions were selected randomly, and one clinic from each region was chosen randomly (the list of Tehran clinics was obtained from the Iranian Psychological and Counseling Organization).
The research instruments included the Emotional Divorce Questionnaire, Basic Empathy Scale, Cognitive Flexibility Inventory, Cognitive Emotion Regulation Questionnaire, and Connor–Davidson Resilience Scale [13-21].
Data analysis used descriptive statistics with AMOS and SPSS 24. To assess the fit of the measurement model, convergent validity, discriminant validity, and instrument reliability were examined using three criteria: factor load coefficients, Cronbach’s alpha coefficients, and composite reliability (CR). First, the measurement model was evaluated via confirmatory factor analysis and indices of convergent and discriminant validity. Subsequently, the structural model (path model) was fitted to test the research hypotheses and examine the mediating role of emotion regulation. Model fit was assessed using indices such as χ²/df, CFI, GFI, RMSEA, and SRMR. The structural model fit was evaluated based on significance coefficients (T-values), the coefficient of determination (R²), and the predictive relevance coefficient (Q²).

Results

In this study, there were 129 men (46.1%) and 151 women (53.9%). The highest frequency was observed in the 41–50 age group, comprising 99 individuals (35.4%). This was followed by the 30–40 age group with 76 individuals (27.1%), the over-50 group with 66 individuals (23.6%), and the under-30 group with 39 individuals (13.9%).
The mean score for emotional divorce is 15.30 ± 2.37. For resilience, the highest mean is for the component "spiritual influences" (1.94) and the lowest is for "perception of individual competence" (1.84). These means are below the scale average, indicating below-average resilience among the participants. For emotional empathy, the highest mean is for "emotional response" (3.02) and the lowest is for "parental empathy" (2.36). For cognitive emotion regulation, the component "positive refocusing" has the highest mean (3.41), while "catastrophizing" and "self-blame" have the lowest (3.32).
All factor loadings exceeded 0.4, indicating adequate relationships between indicators and constructs. The Average Variance Extracted (AVE) and composite reliability values exceeded the recommended thresholds, confirming a good fit, and all paths were statistically significant. All variables had Cronbach’s alpha and composite reliability values greater than 0.7, indicating good internal consistency and reliability. AVE values greater than 0.5 confirmed convergent validity. Thus, the measurement tools demonstrated sufficient accuracy and validity.
The correlation coefficients between the research variables were calculated. The path coefficients and model fit were assessed using T-values and R² values. The T-value for the effect of resilience on emotional divorce was 2.19, for emotional empathy 3.53, and for cognitive flexibility 3.99. Since all these values exceed 1.96, all paths are statistically significant, confirming the hypotheses.
Based on the R² values, emotional divorce had a coefficient of 0.636 and emotion regulation 0.416. According to standard benchmarks (weak = 0.19, moderate = 0.33, strong = 0.67), the explanatory power for emotional divorce is relatively strong, and for emotion regulation is moderate-to-high. Therefore, the structural model demonstrates good explanatory power.
The Q² value for emotional divorce and emotion regulation was 0.609 and 0.301, resepectively. Per predictive relevance benchmarks (weak = 0.02, moderate = 0.15, strong = 0.35), the model has strong predictive power for emotional divorce and moderate predictive power for emotion regulation. This indicates that the model has a good fit from a predictive perspective.

Discussion and Conclusion

The primary objective of the research was to examine and model the relationships between psychological variables influencing emotional divorce in military families; with a particular emphasis on the roles of resilience, emotional empathy, and cognitive flexibility, while considering the mediating role of emotion regulation. Results showed the structural equation modeling of emotional divorce based on these variables with emotion regulation mediation has good fit in couples at risk. This finding is consistent with the results of previous studies. [22-27].
The study found that resilience, emotional empathy, and cognitive flexibility are each directly and significantly related to emotional divorce. Couples with higher resilience better withstand relationship pressures and are therefore less likely to experience emotional separation. Emotional empathy fosters deeper emotional connection—partners who understand and share each other’s feelings are more able to prevent breakdowns in critical moments. Cognitive flexibility—the capacity to reframe thoughts, accept different perspectives, and overcome cognitive biases—improves interactions and reduces internal conflict, which also lowers the risk of emotional divorce.
A key finding was that emotion regulation mediates the relationships between those predictors and emotional divorce. In other words, the ability to identify, process, and manage emotions amplifies the protective effects of resilience, empathy, and cognitive flexibility. Couples who combine these positive psychological traits with effective emotion regulation are more successful at maintaining their emotional bond and preventing gradual deterioration.
These results align with prior research highlighting the importance of resilience and emotion regulation for mental health and interpersonal functioning, the role of empathy in marital quality, and the contribution of cognitive flexibility to problem solving and adjustment [28, 29]. The model showed good fit and coherently explained how these psychological factors interact to influence emotional divorce.
Implications: interventions that build resilience, teach empathy skills, enhance cognitive flexibility, and strengthen emotion-regulation abilities may effectively reduce emotional divorce and could be used by family counselors, psychologists, and public health planners.
Limitations: The sample consisted solely of military families, who differ socially, culturally, and psychologically from civilian populations; therefore, generalizations should be made with caution. Data were self-reported and may be biased, particularly if military family members underreport negative emotions or problems. Although the model examined key factors, it did not control for variables such as marital satisfaction, attachment styles, conflict resolution, psychiatric history, or social support, any of which could moderate or mediate the observed relationships.
Recommendations: Replicate the model in non‑military families to enable direct comparison and investigate the role of organizational culture in emotional divorce. Use longitudinal designs to capture dynamic relationships over time—for example, by tracking changes in emotion regulation and empathy among at‑risk couples. Future studies should include additional variables (attachment style, marital conflict, social support, religious beliefs, sexual quality, etc.) to assess their effects on the primary constructs.

Ethical Considerations

Participants were informed about research objectives and questionnaire purpose. Identity information remained confidential. The study has been approved by Islamic Azad University - North Tehran Branch ethics committee with code IR.IAU.TNB.REC.1404.077.

Funding

There is no funding support.

Authors’ Contribution

The first author handled data collection and initial drafting, the second led conceptualization, analysis, and editing, the third contributed to analysis and editing. All authors contributed equally to conceptualization and writing. Each reviewed and approved the final manuscript, agreeing on all aspects.

Conflict of Interest

Authors declared no conflict of interest.

Acknowledgments

We sincerely thank all the participants who participated as subjects in this study and supported its implementation.