TUTORIAL SEM PLS DENGAN VARIABEL INTERVENING: DILENGKAPI PENJELASAN DAN KESIMPULAN

PLS-SEM Analysis Setup with Intervening Variable
📌 The tutorial demonstrates data analysis using SmartPLS version 3.0 for a study investigating the effect of work motivation (X1) and work environment (X2) on employee loyalty (Y), mediated by job satisfaction (Z).
📂 Data used must be in .txt (notepad) or .csv format for SmartPLS to read correctly.
📐 The structural model includes direct effects ($X_1 \rightarrow Y$, $X_2 \rightarrow Y$, $X_1 \rightarrow Z$, $X_2 \rightarrow Z$, $Z \rightarrow Y$) and indirect effects (mediation paths).

Measurement Model Assessment (Outer Model)
✅ Convergent Validity requires all indicator outer loadings to be greater than 0.5.
✅ Discriminant Validity (assessed via AVE) requires the AVE value for each construct to be greater than 0.5 (e.g., Job Satisfaction AVE = 0.688).
✅ Collinearity Check uses the VIF (Variance Inflation Factor), which must be less than 5 for all indicators to ensure no multicollinearity.
✅ Reliability Check assesses Composite Reliability (CR) and Rho_A, both of which must be greater than 0.7 (or 0.6 depending on the standard used) for acceptable reliability.

Structural Model Assessment (Inner Model)
📊 Model Fit ($\text{R}^2$): The $\text{R}^2$ value indicates the explained variance; $\text{R}^2 > 0.75$ is strong, $0.50 - 0.75$ is moderate. Path 2 ($\text{R}^2=0.695$) suggests a moderate ability of exogenous variables (via Z) to explain Y.
💥 Effect Size ($\text{f}^2$): Used to assess the relative impact of exogenous variables. Values of $0.02, 0.15, \text{and } 0.35+$ indicate small, medium, and large effects, respectively.
🔍 Hypothesis Testing (Direct Effects): Significance is determined by P-values $< 0.05$. The path coefficient (Original Sample) indicates directionality (positive = searah/in the same direction).

Mediation Analysis (Indirect Effects)
🔗 Indirect Effect Significance: If the P-value for the indirect effect is $< 0.05$, the mediating variable (Job Satisfaction, Z) is confirmed to play a role in the relationship between the exogenous variable (X1 or X2) and the endogenous variable (Y).
🛑 Non-Significant Mediation: The indirect effect of Work Environment ($X_2$) on Loyalty ($Y$) through Job Satisfaction ($Z$) resulted in a P-value of 0.184 ($> 0.05$), indicating that Job Satisfaction does not significantly mediate this specific relationship.
✅ Significant Mediation: The indirect effect of Work Motivation ($X_1$) on Loyalty ($Y$) through Job Satisfaction ($Z$) was significant (P-value 0.026 $< 0.05$), confirming mediation.

Key Points & Insights
➡️ Document Everything: Always screenshot and save the results from every stage (outer loadings, AVE, VIF, bootstrapping results) as they are required for the appendix/attachments section of the final research report.
➡️ Interpret Path Coefficients: A positive coefficient (e.g., $X_1 \rightarrow Y$ at 0.369) implies that an increase in the predictor variable leads to an increase in the outcome variable (direct relationship).
➡️ Bootstrapping is Essential: Bootstrapping must be performed to obtain the necessary P-values for testing the significance of direct effects, indirect effects, and collinearity checks.

📸 Video summarized with SummaryTube.com on Dec 03, 2025, 03:48 UTC