Structural Equation Modeling in Behavioral Science

Theoretical Model & Research Setup

An organizational psychology doctoral researcher developed an integrated model testing how ethical leadership and psychological empowerment mediate the impact of perceived organizational justice on employee innovative work behavior across 480 knowledge workers.

Statistical Challenges Faced

• Sub-optimal initial measurement model fit (CFI < 0.88, RMSEA > 0.085).
• Multicollinearity between closely related psychological sub-dimensions.
• Need for non-parametric bootstrapping to confirm indirect mediation effects with bias-corrected 95% confidence intervals.

Stalynus Research Intervention

1. Data Screening & Normality: Evaluated univariate and multivariate skewness/kurtosis (Mardia's coefficient) and screened for multivariate outliers via Mahalanobis d-squared.
2. Measurement Model (CFA): Conducted Confirmatory Factor Analysis in IBM SPSS AMOS, verifying convergent validity (Average Variance Extracted > 0.50) and discriminant validity using the Fornell-Larcker criterion and HTMT ratios.
3. Structural Model & Fit Indices: Optimized model fit indices achieving CFI = 0.962, TLI = 0.954, SRMR = 0.041, and RMSEA = 0.048.
4. Mediation Bootstrapping: Executed 5,000 bootstrap resamples in AMOS, establishing statistically significant indirect mediation effects without relying on the restrictive Sobel test.

Result

The candidate received commendations from their internal and external examiners for rigorous statistical hygiene and flawless APA 7th presentation of structural path models.