<![CDATA[The quality of husband–wife interaction is important for family stability, particularly among nurses who must manage substantial occupational demands under shift and non-shift work arrangements. This study examined differences in social support, work–family conflict, family functioning, and husband–wife interaction between shift and non-shift nurses. It also investigated the associations among these variables. Using a cross-sectional design, 200 married nurses—100 shift nurses and 100 non-shift nurses—were recruited through purposive sampling. Group differences were examined using the Mann–Whitney U test, supplemented by covariate-adjusted and parametric sensitivity analyses. The proposed relationships were tested using partial least squares structural equation modeling (PLS-SEM). The primary analysis indicated that family functioning was higher among non-shift nurses (p = .009); however, this difference was no longer significant after adjustment for demographic covariates. Work–family conflict did not differ significantly between the groups in the Mann–Whitney analysis (p = .055), whereas a supplementary parametric analysis yielded p = .044, indicating that this result was sensitive to the analytical method. No significant group differences were found in social support or husband–wife interaction. The PLS-SEM results showed that social support was positively associated with family functioning (β = .660) and husband–wife interaction (β = .200) and negatively associated with work–family conflict (β = −.219). Family functioning was positively associated with husband–wife interaction (β = .416) and emerged as the primary mediating pathway. Work–family conflict was not significantly associated with either family functioning or husband–wife interaction. Overall, social support and family functioning were more closely associated with the quality of husband–wife interaction than work–family conflict. Differences between shift and non-shift nurses should be interpreted cautiously because they were not consistent across adjusted and sensitivity analyses.]]>
