The assessment of the performance of covariance-based structural equation modeling and partial least square path modeling
Structural equation modeling (SEM) is the second generation statistical analysis technique developed for analyzing the inter-relationships among multiple variables in a model. Previous studies have shown that there seemed to be at least an implicit agreement about the factors that should drive the c...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Conference or Workshop Item |
| Language: | English |
| Published: |
2017
|
| Subjects: | |
| Online Access: | http://eprints.unisza.edu.my/1643/1/FH03-FESP-17-09162.jpg http://eprints.unisza.edu.my/1643/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Universiti Sultan Zainal Abidin |
| Language: | English |
| Summary: | Structural equation modeling (SEM) is the second generation statistical analysis technique developed for analyzing the inter-relationships among multiple variables in a model. Previous studies have shown that there seemed to be at least an implicit agreement about the factors that should drive the choice between covariance-based structural equation modeling (CB-SEM) and partial least square path modeling (PLS-PM). PLS-PM appears to be the preferred method by previous scholars because of its less stringent assumption and the need to avoid the perceived difficulties in CB-SEM. Along with this issue has been the increasing debate among researchers on the use of CB-SEM and PLS-PM in studies. The present study intends to assess the performance of CB-SEM and PLS-PM as a confirmatory study in which the findings will contribute to the body of knowledge of SEM. Maximum likelihood (ML) was chosen as the estimator for CB-SEM and was expected to be more powerful than PLS-PM. Based on the balanced experimental design, the multivariate normal data with specified population parameter and sample sizes were generated using Pro-Active Monte Carlo simulation, and the data were analyzed using AMOS for CB-SEM and SmartPLS for PLS-PM. Comparative Bias Index (CBI), construct relationship, average variance extracted (AVE), composite reliability (CR), and Fornell-Larcker criterion were used to study the consequence of each estimator. The findings conclude that CB-SEM performed notably better than PLS-PM in estimation for large sample size (100 and above), particularly in terms of estimations accuracy and consistency. |
|---|