What can talent management do about involuntary remote working in the post-COVID-19 era? Spanish IT employees’ organizational commitment

The COVID-19 pandemic accelerated the adoption of telework, creating new challenges and opportunities for talent management (TM). This study examines how professional isolation (PI) and work-family balance (WFB) relate to organizational commitment (OC) among Spanish IT employees who were forced to telework, considering the mediating role of job satisfaction (JS) and the potential moderating roles of gender and time spent teleworking. The research aims to: (1) understand PI and WFB as talent retention levers through their effects on OC and JS; (2) test moderation by gender and telework intensity on the PI–OC link; and (3) assess the predictive capability of an integrative model in a forced teleworking context. Grounded in organizational support theory, need-to-belong theory, and self-determination theory, the study addresses an underexamined setting—forced telework in IT firms—and informs TM strategies to sustain commitment in the post-pandemic era.

Theoretical framing integrates organizational support theory (Rhoades & Eisenberger, 2002), need-to-belong theory (Baumeister & Leary, 1995), and self-determination theory (Deci & Ryan, 2012). Telework can reduce face-to-face interaction, hindering relatedness and trust, potentially increasing PI and diminishing JS and OC. OC is conceptualized across affective, normative, and continuance dimensions and is positively associated with JS. PI encompasses psychological and physical isolation among teleworkers and is linked to lower JS and OC. Perceived organizational support for WFB is associated with higher JS and OC. Based on prior evidence, the study hypothesizes: H1 JS→OC (+); H2 PI→JS (−); H3 PI→OC (−); H4 WFB→JS (+); H5 WFB→OC (+); H6 JS mediates PI→OC; H7 JS mediates WFB→OC; H8 gender moderates PI→OC; H9 time spent teleworking moderates PI→OC.

Design: Quantitative, cross-sectional survey (post-positivist). Analytical approach: PLS-SEM emphasizing prediction. Software: SmartPLS 4.0.8.7. Sampling and data collection: Target population—employees teleworking in Spain’s technology sector. Survey distributed via email using snowball sampling; initial convenience sample of 30 teleworkers from companies listed in the Spanish Association of Technology Parks directory (computer science and telecommunications, 1,074 companies). Data collected January–March 2021. Of 297 responses, 3 multivariate outliers (Mahalanobis distance) were removed, yielding N=294 valid teleworkers (each from a different company). Mean age 36.42 (SD=8.39); 63.61% male; 48.64% teleworked >50% of their workday. Sampling error estimated at 4.87% (95% CI; p=q=0.5) when considering companies as the population unit. A priori power analysis (G*Power) indicated a minimum of 119 cases for power=0.95, effect size=0.15, three predictors. Common method bias: Assessed via full collinearity approach with a random latent variable (CMB) regressed on all constructs; VIFs <3.3 indicated absence of CMB. Measures: All constructs reflective; 7-point Likert (1=strongly disagree; 7=strongly agree). Scales adapted and translated to Spanish: PI (4 items from Golden et al., 2008; reduced due to redundancy in translation), JS (5 items from Pond & Geyer, 1991), OC (4 items from Meyer & Allen, 1991), WFB (5 items from Thompson et al., 1999). Expert review and pretest (n=30 students with telework experience); Cronbach’s alpha >0.80. Confirmatory tetrad analysis (CTA-PLS) supported reflective specification of WFB and JS. Measurement model: After dropping three items (JS4, OC3, PI3; loadings <0.70), reliability was acceptable: CA, rho_A, CR >0.80; AVE >0.50. Discriminant validity met via Fornell-Larcker and HTMT (<0.90). Structural model: Bootstrapping with 10,000 subsamples. Collinearity acceptable (inner VIFs <3.3). Predictive relevance: R2 JS=0.685; R2 OC=0.811; Q2predict positive for endogenous constructs. Out-of-sample prediction (PLSpredict; 10-fold, 10 repetitions) for OC showed low predictive performance (only 1/3 OC items MAE_PLS < MAE_LM). Unobserved heterogeneity assessed via FIMIX-PLS (1–3 segments); fit indices supported a one-segment solution. Importance-performance map analysis (IPMA) conducted with OC as target and PI, JS, WFB as predecessors.

• Direct effects: WFB→JS β=0.478, p<0.001; WFB→OC β=0.360, p<0.001; PI→JS β=−0.413 (reported also as −0.470 in text), p<0.001; PI→OC β=−0.070, p<0.05; JS→OC β=0.508, p<0.001 (H1–H5 supported).
• Indirect effects: JS mediates WFB→OC (β=0.243, p<0.001) and PI→OC (β=−0.210, p<0.001) (H6–H7 supported).
• Moderation: Gender and time spent teleworking did not moderate PI→OC (H8, H9 rejected).
• Model fit and prediction: R2 JS=0.685; R2 OC=0.811; Q2predict positive; PLSpredict indicated low out-of-sample predictive performance relative to a linear model baseline.
• IPMA: Importance for OC—WFB 0.603 (highest), JS 0.508, PI −0.280 (lowest). Performance—WFB 64.725, JS 68.585, PI 32.634 (lowest performance, indicating room for improvement). Overall: WFB and JS are the strongest positive determinants of OC; PI has a negative but comparatively smaller effect. Neither gender nor telework intensity conditions the PI–OC relationship.

Findings confirm that professional isolation undermines job satisfaction and organizational commitment, while perceived support for work-family balance enhances both. Job satisfaction plays a central mediating role, translating the effects of PI and WFB into OC. Contrary to expectations, gender and the intensity of telework do not alter the negative association between PI and OC, suggesting that PI’s detrimental impact on commitment is robust across these subgroups. Practically, organizations should proactively reduce perceived isolation (e.g., mentoring, fostering informal interactions) and prioritize WFB policies to bolster satisfaction and commitment. The IPMA underscores that improving WFB and JS will yield the greatest gains in OC, whereas PI, although negative, is less consequential in determining OC performance in this IT telework context. The integrative, multi-theoretical model helps reconcile mixed evidence on telework outcomes by demonstrating how support and satisfaction can offset isolation’s downsides.

This study advances understanding of talent management in forced telework by integrating professional isolation, work-family balance, job satisfaction, and organizational commitment into a single model with mediating and moderating mechanisms. Among Spanish IT teleworkers, WFB and JS emerge as primary drivers of OC, while PI exerts a smaller negative effect; JS mediates both WFB→OC and PI→OC. Gender and telework intensity do not moderate the PI–OC link. The results suggest that organizations can maintain or enhance commitment in remote settings by emphasizing WFB-supportive cultures and initiatives that elevate JS, alongside efforts to monitor and mitigate isolation. Future research should extend to other sectors and geographies, employ longitudinal designs to strengthen causal inference, consider alternative theoretical lenses, and explore additional moderators and controls to refine talent management strategies in hybrid and remote work environments.

• Design: Cross-sectional survey limits causal inference; longitudinal approaches (e.g., PLSe2) are recommended.
• Sampling: Snowball, non-random sampling due to lack of a teleworker census may limit generalizability; future studies should pursue random sampling if feasible.
• Perspective and data sources: The paper notes the importance of gathering input from workers in addition to managerial perspectives; multi-informant designs and mixed methods (documents, interviews) could provide a more comprehensive picture.
• Theoretical scope: The study uses three theories (organizational support, need-to-belong, self-determination); future research could test alternative frameworks.
• Moderators and context: Gender and time spent teleworking did not moderate PI→OC here; effects should be analyzed in other contexts (sectors, geographies, post-lockdown periods).
• Controls: Future work should consider job type, company type, work experience, education, and marital status to account for heterogeneity.
• Prediction: PLSpredict indicated low out-of-sample predictive performance; model refinement and inclusion of additional predictors may improve predictive validity.