Development of a two-way mentorship scale focusing on next-generation core competencies

The paper addresses how rapid technological and societal changes are shifting the core competencies required of new employees from primarily knowledge and skills toward attributes such as technology sensitivity, continuous learning, and teamwork. Traditional one-way mentorships may no longer capture these evolving needs. The study proposes two-way mentorship—integrating traditional and reverse mentoring—as a better-aligned model where mentors and protégés learn mutually. The research aims to develop and validate a two-way mentorship scale grounded in next-generation core competencies, to assess mentorship functions and better predict workplace performance and organizational needs.

The review traces competency theory from McClelland to contemporary applications, defining core competencies as dynamic, organization-wide abilities that evolve with environmental changes. It highlights employer prioritization of internal qualities (e.g., proactive attitude, adaptability, communication) and technology-related skills (e.g., working with AI, interdisciplinary collaboration). Traditional mentorship (Kram, Scandura) comprises career development and psychosocial functions with role modeling; it supports socialization, performance, and organizational competitiveness. Reverse mentoring inverts roles, enabling younger or less-experienced employees to teach senior staff about new technologies and ideas, fostering cross-generational learning, leadership development, and cultural connection. Two-way mentorship integrates both models, emphasizing bidirectional learning aligned with digital-era work. The review organizes two-way mentorship functions into three constructs: career development (instruction, protection, challenge arrangements, exposure, visibility, with reverse-mentoring considerations around management and creative challenge), psychosocial support (acceptance, teaching, consultation, friendship, teamwork rooted in social support and exchange theories), and role modeling (trust, respect, and observational learning, with cross-generational role identity considerations). It also summarizes reported benefits across domains including healthcare and education, and notes the emergence of e-mentoring and social media in mentorship networks. Table 1 outlines theoretical sources underpinning two-way mentorship and next-generation competencies.

Design: Instrument development and validation study to construct a two-way mentorship scale integrating next-generation core competencies with three constructs: career development, psychosocial support, and role modeling. Item generation drew on Kram (1983), Scandura (1992), Murphy (2012), and Chen (2014, 2018), and on collated enterprise core competencies. Response format: 6-point Likert (1=strongly disagree to 6=strongly agree). Expert review: Experts (including supervisor, trainer, experienced and inexperienced employees; profiles reported for five experts spanning IC design HR head, technology manufacturing coordinator, associate professor, semiconductor senior manager, and information services executive assistant) reviewed items for clarity and content relevance; iterative revisions were made. Pilot scale: 37 items (career development 16, psychosocial support 12, role modeling 9). Purposive sampling recruited employees in mentorship programs across northern, central, southern, and eastern Taiwan; n=303 responses. Item analysis used upper/lower 27% grouping and t-tests; items with nonsignificant CR, item–total correlation <0.30, or improving alpha upon deletion were removed (e.g., High standard a). EFA: KMO=0.95; Bartlett’s χ2=269.22, df=35, p<0.001; extraction via principal axis factoring (fm="pa"), promax rotation. Initial factor eigenvalues ~7.01, 6.12, 3.78; items failing loading/complexity criteria were dropped (e.g., Management a/b, Information a/b, Learning resource a, Identification, Supersession a/b, High standard b, Communication a/b). Second EFA: KMO=0.92; Bartlett’s χ2=157.03, df=20, p<0.001; eigenvalues 8.747, 1.685, 1.327; three factors labeled career development, psychosocial support, role modeling. Reliability after refinement: overall α=0.91; subscales α=0.892, 0.878, 0.763. Proposed scale validation sample: Purposive sampling across Taiwan; n=504 participants engaged in mentorship programs. Descriptives: 316 female, 188 male; education mainly university (n=259) and graduate+ (n=208); mentorship roles included 326 never mentors (267 current mentees), 178 with mentor experience (91 former, 87 current), employed across diverse industries (academia/education/counseling most represented). Item analysis confirmed means between 4–5 and item–total correlations >0.30; overall α=0.94 and deletion of any item reduced α. EFA on n=504: fm="pa", promax. First EFA eigenvalues 9.615, 1.515, 1.229; items generally loaded >0.50; Team a had loading <0.50 and was removed. Second EFA: KMO=0.94; Bartlett’s χ2=245.9, df=19, p<0.001; eigenvalues 8.75, 1.69, 1.33; final retained items loaded on three factors consistent with theory. CFA: Conducted in R (lavaan) on n=504 to test the three-factor model. Fit: χ2=583.093, df=167, p<0.001 (noted as sample-size sensitive); RMSEA=0.07; SRMR=0.044; CFI=0.921; NNFI (non-normal fit index)=0.911; all met recommended thresholds, supporting construct validity. Convergent validity (Fornell-Larcker): standardized loadings >=0.50; AVE/CR: career development AVE=0.486, CR=0.911; psychosocial support AVE=0.497, CR=0.850; role modeling AVE=0.342, CR=0.591 (lower, attributed to fewer items after deletions). Discriminant validity: sqrt(AVE) for career development=0.697 and psychosocial support=0.704 exceeded inter-construct correlations; role modeling sqrt(AVE)=0.584 did not exceed correlations with other constructs, indicating weaker discriminant validity for role modeling.

- Developed a two-way mentorship scale with three constructs: career development, psychosocial support, role modeling, aligned with next-generation core competencies. - Reliability: Overall Cronbach’s α≈0.94; subscales α=0.911 (career development), 0.879 (psychosocial support), 0.791 (role modeling). - Factor structure: All retained items had factor loadings >0.50; EFA and CFA supported a robust three-factor model. CFA fit indices: RMSEA=0.07, SRMR=0.044, CFI=0.921, NNFI=0.911; χ2=583.093, df=167. - Convergent validity: Loadings ≥0.50; AVE/CR ranged 0.342–0.497 and 0.591–0.911, respectively. Role modeling showed lower AVE and CR due to fewer items after deletion. - Discriminant validity: Supported for career development and psychosocial support; weaker for role modeling (sqrt(AVE) 0.584 below some inter-factor correlations). - Descriptives (N=504): Overall scale mean=4.739 (SD=0.661); subscale means: career development 4.808 (SD=0.684), psychosocial support 4.624 (SD=0.828), role modeling 4.749 (SD=0.838).

Findings demonstrate that a two-way mentorship model—integrating traditional and reverse mentoring—can be operationalized with a psychometrically sound scale aligned to current organizational competency needs. The scale captures bidirectional features of mentorship, particularly in career development (instruction, protection, challenge, exposure, visibility) and psychosocial support (acceptance, teaching, consultation, friendship, teamwork), consistent with prior theory and evidence on mentorship’s benefits for performance, socialization, and satisfaction. Role modeling in a two-way context appears to hinge primarily on trust and respect, reflecting evolving cross-generational role identities; however, its construct distinctness is less robust and warrants further refinement. Compared with established instruments (e.g., Scandura & Ragins; MFQ-9), incorporating next-generation core competencies and two-way learning enriches content relevance for today’s workplaces, better aligning measurement with organizations’ emphasis on internal attributes, technology sensitivity, and collaborative capacities.

This study contributes a validated two-way mentorship scale reflecting contemporary mentorship dynamics and next-generation core competencies across three constructs. Psychometric analyses (EFA, CFA, reliability, convergent and discriminant validity) support the instrument’s use to assess mentorship relationships in organizations, inform training and development, and align talent practices with evolving competency demands. The scale can aid HR in recruitment, development, and retention by identifying mentorship dynamics that foster performance, learning, and collaboration. Future research should: expand samples across industries and roles; enrich methodology with qualitative interviews and field observations; refine and expand the role modeling subscale to improve convergent and discriminant validity; and test criterion-related validity by linking scale scores to outcomes such as job performance, engagement, and innovation.

The role modeling subscale contained few items post-item reduction, resulting in lower AVE, CR, and weaker discriminant validity relative to other constructs. The purposive sampling within Taiwan and sector distributions may limit generalizability. The study relied on self-report survey methods without triangulating with qualitative or behavioral data. CFA indicated room for model adjustment (chi-square significant), albeit with acceptable alternative fit indices.