3.3. Model and analysis
Given the existence of an exogenous latent variable (transformational leadership [ξ1]), a first-grade endogenous latent variable (organizational learning [η1]) and second-grade endogenous latent variables (organizational innovation [η2] and organizational performance [η3]), the study analyzes the data using structural equations modeling (LISREL 8.30 program) to establish causal relationships between these variables. This procedure translates the theoretical construction into mathematical models in order subsequently to estimate and evaluate them empirically (Jöreskog and Sörbom, 1996).
Fig. 1 presents the theoretical model. The findings give the hypotheses concrete form. The study uses a recursive non-saturated model.
Structural equation modeling takes into account errors in measurement, variables with multiple indicators and multiple-group comparisons (Koufteros et al., 2009).