Antecedents to Leadership: A CB-SEM and PLS-SEM Validation

Abstract

The main issue with this paper is to investigate the link between emotional intelligence and transformational leadership and the role of organizational culture as a moderator on that relationship by using two research methods: The covariance-based structural equation modeling (CB-SEM) and partial least squares (PLS-SEM). The study examined a complex model consisting of 60 indicators including moderator effects which used real data. This will help in understanding the respective differences of the two approaches in a setup comprising model specification and parameter estimation. The dual SEM approach represents an important contribution, permitting validation of the model’s robustness, and, thanks to the CB-SEM method, to overcome the limitations of PLS-SEM. The findings show that both methods yield similar results with minor differences that may be attributed to their respective estimation requirements including model fit and complexity issues. After considering these results and findings from studies done in this line, the researcher concludes that future studies need to observe recommendations made to focus on the phenomenon and research design aspects and, not mere modeling. A study limitation is not testing SEM boundaries with non-normal data and small sample size. The study is first to apply SEM approaches to verify results of a complex leadership model that included moderator affects. A key implication is the insight gained about the application of standards and guidelines for clarifying the interpretation of the SEM theories and models for leadership and management research. This implies the equal use of the CB-SEM and PLS-SEM for future studies, without undue bias.