Tutorial SEM PLS dengan Variabel Mediasi Menggunakan SmartPLS 4 || Lengkap dengan referensi

Summary

This tutorial demonstrates how to perform a Structural Equation Modeling (SEM) analysis with mediation variables using SmartPLS 4. The video covers model setup, including variables such as price, service quality, satisfaction, and loyalty. It walks through data import, model creation, validity testing (convergent and discriminant), reliability checks, and hypothesis testing. The tutorial also explains key metrics such as R-squared values, F-squared, and significance testing, providing a comprehensive guide for users to implement SEM with mediation in their own research.

Takeaways

• 😀 The tutorial explains how to perform mediation analysis using Smart PLS 4 with variables like price, service quality, satisfaction, and loyalty.
• 😀 The model includes two independent variables (price and service quality), one intervening variable (satisfaction), and one dependent variable (loyalty).
• 😀 There are seven hypotheses in the model, five of which are direct hypotheses, and two are indirect (mediation) hypotheses.
• 😀 The data is imported from an Excel file, and the first step involves checking for missing values before building the model in Smart PLS 4.
• 😀 In Smart PLS 4, the model is created by linking the variables based on the hypotheses and adjusting them as needed.
• 😀 Convergent validity is assessed by calculating the outer loadings of each indicator, with values greater than 0.7 considered valid.
• 😀 Discriminant validity is tested using the Fornell-Larcker criterion, ensuring that the square root of AVE is higher than the correlations between constructs.
• 😀 Reliability is measured using composite reliability and Cronbach's alpha, with values above 0.7 indicating acceptable reliability.
• 😀 The model's structural (inner) evaluation involves checking R-squared values, with adjusted R-squared values of 0.542 for satisfaction and 0.778 for loyalty indicating moderate and strong models, respectively.
• 😀 Hypothesis testing is performed using bootstrapping in Smart PLS, with p-values less than 0.05 indicating significant effects. The results for direct hypotheses (H1-H5) are reported, along with the indirect (mediation) hypotheses (H6-H7).
• 😀 The Goodness of Fit (GoF) is calculated using the average AVE and R-squared values, and a GoF score greater than 0.36 is considered a good fit for the model.

Q & A

What is the focus of the tutorial provided in the video?

The tutorial focuses on conducting a mediation analysis using SmartPLS 4, exploring the relationships between independent variables (Price and Service Quality), a mediating variable (Satisfaction), and a dependent variable (Loyalty).

What software is used for the analysis in this tutorial?

The tutorial uses SmartPLS 4 for conducting the mediation analysis and creating models.

What variables are involved in the mediation analysis?

The analysis involves two independent variables (Price and Service Quality), one mediating variable (Satisfaction), and one dependent variable (Loyalty).

What is convergent validity, and how is it assessed in the tutorial?

Convergent validity is assessed by ensuring that the factor loadings of the indicators for each variable are above a threshold value. In this tutorial, a threshold of 0.7 is used to confirm convergent validity.

What is the purpose of discriminant validity, and how is it checked?

Discriminant validity ensures that each construct is distinct from the others. It is checked using the Fornell-Larcker criterion, where the square root of the AVE (Average Variance Extracted) should be greater than the correlations between constructs.

What reliability tests are used in the tutorial?

The tutorial uses composite reliability and Cronbach's alpha to test the reliability of the variables. Composite reliability is considered the most important, and values above 0.7 are preferred.

How is the structural model evaluated in the analysis?

The structural model is evaluated by checking the R-squared values for the dependent variables. The R-squared value for Satisfaction is 0.542 (moderate explanatory power), and for Loyalty, it is 0.778 (strong explanatory power).

What is the significance of the f² values in the analysis?

The f² values measure the effect size of the relationships between variables. In this tutorial, strong effects are found between Service Quality and Satisfaction, Service Quality and Loyalty, and Satisfaction and Loyalty.

What is the result of the hypothesis testing regarding direct effects?

The hypothesis testing results show that Service Quality has a significant effect on Satisfaction and Loyalty, while Price does not significantly affect Satisfaction or Loyalty.

Were the mediation hypotheses supported in the tutorial?

No, the mediation hypotheses (Price → Satisfaction → Loyalty and Service Quality → Satisfaction → Loyalty) were not supported, as the p-values for both indirect effects were greater than 0.05.

How is the Goodness of Fit (GoF) calculated, and what does it indicate?

The Goodness of Fit (GoF) is calculated by multiplying the square root of the average AVE and the average R-squared. In this tutorial, a GoF of 0.716 indicates a good fit between the outer and inner models of the research.