Family-work conflict and work-from-home productivity: do work engagement and self-efficacy mediate?

The study examines how family-work conflict (FWC) influences work-from-home productivity (WFHP) among Taiwanese employees during the COVID-19 pandemic and investigates the mechanisms underlying this relationship. Taiwan experienced its first wave of COVID-19 around June 2021, prompting widespread work-from-home (WFH) arrangements. Building on role conflict theory and resource drain theory, the authors propose that limited personal resources (time, energy) lead to conflicts between family and work roles, which in turn reduce engagement, self-efficacy, and productivity. The research addresses gaps in prior work by: (1) testing both work engagement (WE) and self-efficacy (SE) as mediators between FWC and WFHP, (2) examining the moderating role of working hours operationalized as work-from-home days (WFHDs), and (3) focusing on SME employees in Taiwan during the pandemic. The study poses two research questions: RQ1: How does FWC affect WFHP, and what is the mechanism behind their effect? RQ2: How do working hours (WFHDs) moderate the negative association between FWC and WFHP?

The review traces WFH from a flexible, telework option to a widespread, sometimes mandatory, arrangement during COVID-19. Prior research links WFH to work-life balance, engagement, stress, and productivity, with mixed findings on gender and hours worked. Theoretical anchors are role conflict theory (Kahn et al., 1964) and resource drain theory (Rothbard & Edwards, 2003), positing finite personal resources that, when strained by competing roles, reduce engagement and performance. Hypotheses: - H1: FWC negatively affects WFHP. - H2: FWC negatively affects WE. - H3: WE positively affects WFHP. - H4: WE mediates the negative FWC→WFHP relationship. - H5: FWC negatively affects SE. - H6: SE positively affects WFHP. - H7: SE mediates the negative FWC→WFHP relationship. - H8: The negative FWC→WFHP relationship is stronger for employees with more WFHDs. Control variables examined include gender, age, work experience, work field, and number of children, given prior evidence of their potential influence on productivity.

Design and approach: Descriptive, quantitative study testing a multiple-mediator (WE, SE) and moderator (WFHDs) model of FWC→WFHP grounded in role conflict and resource drain theories. Measures: Self-report Likert (1–5) scales. - FWC: 5 items adapted from Netemeyer et al. (1996). - WE: 9 items adapted from Schaufeli et al. (2006), tailored to WFH context. - SE (occupational self-efficacy): 6 items adapted from Rigotti et al. (2008). - WFHP: 7 items adapted from Irawanto et al. (2021). Demographics: gender; age (18–30, 31–40, 41–50, >50); work experience (<2, 2–5, 6–10, >10 years); work field (production management, marketing, administrative affairs, financial accounting, other); number of children (0, 1, 2, >2); WFH days per week during 2021 level 3 alert (<2, 2–3, 4–5, >5); WFH status during the alert (yes/no, used for screening). Instrument development, validity, and reliability: Forward translation to Chinese by two English teachers; back translation by two different teachers; expert review by three HRM professionals; cognitive pretest with five employees; pilot test (n=50) with Cronbach’s alpha >0.60 and item-total correlations >0.30. For the main study, PCA and CFA confirmed construct validity; Cronbach’s alphas >0.80; CR ≥0.88; AVE ≥0.50; all standardized loadings >0.60 and significant (p<0.001). Measurement model fit: CMIN/df=2.830; CFI=0.952; GFI=0.918; IFI=0.952; TLI=0.947; RMSEA=0.048. Sampling and participants: Non-probability convenience sampling of SME employees in Taiwan (supported by Rotary International Group). Recruitment across Taipei (8 enterprises), New Taipei (7), Taichung (7), and Tainan (8). Survey administered Nov 11–Dec 29, 2021. Invitations=1307; returned=1177 (90.1% response). After excluding non-WFH cases, missing data, and outliers, final N=785. Sample profile (N=785): 45.5% female; ages 18–30 (21.7%), 31–40 (43.4%), 41–50 (25.9%), >50 (9.0%); work experience <2 (9.9%), 2–5 (29.8%), 6–10 (25.4%), >10 (34.9%); fields production management (12.7%), marketing (24.8%), administrative (23.8%), financial accounting (23.3%), others (15.3%); WFH days/week <2 (19.0%), 2–3 (12.5%), 4–5 (51.0%), >5 (17.5%); number of children 0 (8.4%), 1 (43.8%), 2 (20.4%), >2 (27.4%). Descriptives: FWC M=3.00 SD=0.87; WE M=3.67 SD=0.64; SE M=3.90 SD=0.57; WFHP M=3.49 SD=0.72. Data analysis: SPSS v22 for descriptives and preliminaries; AMOS v20 for CFA. Outliers flagged at |z|>3.29; multicollinearity checked (all VIF<5). Common method variance assessed by Harman’s single-factor test (largest factor=30.42%<50%). Hypotheses tested using PROCESS Macro Model 5 (Preacher & Hayes, 2004), estimating direct, indirect (mediation), and interaction (moderation by WFHDs) effects with bootstrapped CIs.

Measurement and validity: PCA supported construct structure (all loadings ≥0.50). CFA indicated good fit (CMIN/df=2.830; CFI=0.952; GFI=0.918; IFI=0.952; TLI=0.947; RMSEA=0.048). Reliability high (Cronbach’s alpha ≥0.88; CR ≥0.88; AVE ≥0.52). CMV not a concern (largest single factor=30.42%). Hypothesis tests (PROCESS Model 5): - H1 supported: FWC → WFHP β=-0.26, p<0.001. - H2 supported: FWC → WE β=-0.30, p<0.001. - H3 supported: WE → WFHP β=0.11, p<0.01. - H4 supported: WE mediates FWC→WFHP, indirect effect β=-0.03, 95% CI [LLCI=-0.0634, ULCI=-0.0063]. - H5 supported: FWC → SE β=-0.25, p<0.001. - H6 supported: SE → WFHP β=0.37, p<0.001. - H7 supported: SE mediates FWC→WFHP, indirect effect β=-0.09, 95% CI [LLCI=-0.1242, ULCI=-0.0621]. - H8 supported: Interaction FWC×WFHDs β=-0.07, SE=0.02, t=-2.87, p<0.01, 95% CI [LLCI=-0.1150, ULCI=-0.0216]. Conditional effects of FWC on WFHP at WFHDs: at mean β=-0.28 (p<0.001); at -1 SD β=-0.14 (p<0.01); at +1 SD β=-0.35 (p<0.001). The negative FWC→WFHP association is stronger with more WFH days. Controls: No significant WFHP differences by work experience, work field, or number of children. Significant gender difference: Men M=3.60 SD=0.62 vs Women M=3.37 SD=0.83; F(1,783)=20.478, p<0.001, η²=0.03 (higher WFHP for men). Overall: FWC reduces WFHP directly and indirectly via lower WE and SE; SE shows a relatively larger indirect effect magnitude than WE. More WFH days intensify the negative FWC→WFHP relationship.

The findings address RQ1 by demonstrating that family-work conflict undermines work-from-home productivity through both direct and indirect pathways via reduced work engagement and self-efficacy. This aligns with role conflict and resource drain theories: limited time and energy distributed across competing family and work roles diminish psychological resources (engagement, efficacy), lowering productivity. Both WE and SE partially mediate the FWC→WFHP link, with SE emerging as a notable mediator in the WFH context. Addressing RQ2, working hours measured as WFH days per week strengthen the negative association between FWC and WFHP: employees with more WFH days experience a steeper productivity decline under higher FWC, consistent with resource strain intensifying under longer at-home work exposure. Additionally, men reported higher WFHP than women, likely reflecting unequal household burdens in the Taiwanese cultural context. These results refine understanding of the conditions under which FWC most strongly impairs productivity and underscore the importance of psychological resources in remote work settings.

This study contributes by proposing and validating a multiple-mediator model linking family-work conflict to work-from-home productivity among Taiwanese SME employees during COVID-19. It provides evidence that both work engagement and self-efficacy partially mediate the negative effects of FWC on productivity, and that greater WFH days exacerbate this relationship. The inclusion of self-efficacy as a mediator fills a gap in the WFH productivity literature, and the identification of working hours as a moderator clarifies boundary conditions. Practically, the results support implementing family-friendly policies, monitoring and fostering engagement, and enhancing self-efficacy through training, mentoring, and supportive communication, as well as considering reasonable limits on WFH duration. Future research should employ probability sampling, longitudinal designs to assess causality, examine additional moderators (e.g., gender, number of children), and compare cross-country contexts to assess generalizability across different cultural and socioeconomic settings.

- Non-probability convenience sampling limits generalizability; probability sampling is recommended for future studies. - Cross-sectional design precludes causal inference; longitudinal or experimental designs are suggested. - Single-country (Taiwan) context limits external validity; multi-country comparisons are encouraged. - Potential untested moderators (e.g., gender, number of children) could further qualify effects; future work should explore these.