Developing Instruments for Digital Talent Competence Using Partial Least Square-Based Models

Research on measuring digital talent competencies using explainable artificial intelligence is lacking in depth. The study aimed to address this gap by developing a reliable and valid measurement instrument for digital talent. SmartPLS 4.0 was employed for calculations and quality assessment. The measurement model was assessed from both reflective and formative perspectives, using Google Forms to gather responses from 600 participants. The 60 questionnaire items exhibited high validity and reliability, as evidenced by loading factors and composite reliability exceeding 0.70. Direct effects were found in various competencies: adaptability and flexibility had a 0.744 effect on learning dexterity, achievement orientation showed a 0.782 effect on innovative creation skills, customer service orientation had a 0.810 effect on digital networking, and digital communication skills exhibited a 0.824 effect on continual improvement. The measurement model met various criteria for reliability and validity, including loading factors, composite reliability, average variance extracted, cross loadings, Fornell Lacker Criterion, and Heterotrait Monotrait Ratio. These findings supported the acceptance of the measurement model’s quality and validity. SmartPLS analysis yielded an average communality of 0.624 and an average R Square of 0.592, resulting in a GoF index of 0.608. This GoF index, according to Wetzel et al. (2018), falls in the high fit category, indicating the model’s appropriateness for measuring digital talent competencies. In summary, the study successfully developed a robust measurement model for digital talent competencies using explainable artificial intelligence and demonstrated its high fit and reliability for assessing digital talent in various contexts.