Tanya Jawab Seputar PLS SEM, Part 6, Q Square dalam PLS SEM

Summary

In this video, Sroedji explains the concept and use of Q² (predictive relevance) in PLS-SEM. He discusses the two approaches for calculating Q²—redundancy and communality—highlighting that the redundancy approach is recommended for including structural model predictions. Using SmartPLS software, Q² is computed through a blindfolding procedure to assess how well a model predicts actual data. He provides clear guidance on interpreting Q² values: values above zero indicate predictive relevance, categorized as weak, medium, or strong. An illustrative example with endogenous and exogenous latent variables demonstrates how to calculate and interpret Q² for real datasets, making the explanation practical and accessible.

Takeaways

• 😀 Chi-Square is a crucial statistic in Partial Least Squares Structural Equation Modeling (PLS-SEM), used to assess the relevance of exogenous variables in predicting endogenous ones.
• 😀 Chi-Square can be calculated using two approaches: **Cross-Braided Redundancy** and **Cross-Fade Communality**. The latter is recommended for better prediction performance.
• 😀 **SmartPLS software** uses the **blindfolding procedure** to compute Chi-Square, which helps in measuring how well the model can predict original data values.
• 😀 A Chi-Square value greater than zero indicates that exogenous variables (like X1 and X2) have a predictive relevance for endogenous variables (like Y1 and Y2).
• 😀 The predictive relevance of variables can be categorized as **weak**, **medium**, or **strong** based on Chi-Square values.
• 😀 If Chi-Square is between **0.02 and 0.15**, the predictive relevance is considered **weak**.
• 😀 If Chi-Square is between **0.15 and 0.35**, the predictive relevance is considered **medium**.
• 😀 If Chi-Square is **greater than 0.35**, the predictive relevance is considered **strong**.
• 😀 For accurate Chi-Square calculations in SmartPLS, you should select the **blindfolding** procedure and then start the calculation.
• 😀 In the given case study, the Chi-Square values for Y1 (0.457) and Y2 (0.49) are above 0.35, which indicates a **strong predictive relevance** of X1 and X2 for both Y1 and Y2.

Q & A

What is the purpose of the chi-square test in PLS-SEM?

The chi-square test in PLS-SEM is used to assess the relevance of exogenous constructs for predicting endogenous variables. It helps to measure how well the model fits and can predict the original data points.

What are the two approaches to calculating chi-square in PLS-SEM?

There are two approaches to calculating chi-square: the redundancy approach and the communality approach.

Which approach for chi-square calculation is recommended in SmartPLS software?

The redundancy approach is recommended for chi-square calculation in SmartPLS because it includes key elements like the structural model to predict eliminated data points.

What does the chi-square value indicate if it is greater than zero?

If the chi-square value is greater than zero, it indicates that the exogenous variables have a predictive relevance for the endogenous variables.

How is the relevance of prediction categorized based on chi-square values?

The predictive relevance can be categorized as weak, medium, or strong based on chi-square values: weak for values between 0.02 and 0.15, medium between 0.15 and 0.35, and strong for values above 0.35.

What is the significance of the blindfolding procedure in chi-square calculation?

Blindfolding is a procedure used to calculate chi-square in SmartPLS. It helps assess how well the model predicts actual data by leaving out data points and predicting them based on the model.

What does it mean if the chi-square value is between 0.15 and 0.35?

If the chi-square value is between 0.15 and 0.35, the predictive relevance is categorized as medium, indicating a moderate level of prediction power for the model.

What is the relationship between exogenous and endogenous variables in the context of chi-square?

Exogenous variables, which have no incoming arrows, predict endogenous variables, which have incoming arrows. The chi-square test helps determine the predictive relevance of exogenous variables for endogenous ones.

What does the chi-square value of 0.45 in the example indicate about the model?

A chi-square value of 0.45 in the example indicates that the exogenous variables (X1 and X2) have strong predictive relevance for the endogenous variable (Y1 or Y2), as it is greater than 0.35.

What should you do if you want to calculate chi-square for a specific model in SmartPLS?

To calculate chi-square in SmartPLS, you should select the blindfolding procedure from the menu and start the calculation. This will give you the chi-square values for redundancy and communality, with redundancy being the preferred approach.