PLS-SEM in Big Data Analytics Quality

Introduction

• The study investigates the suitability of Partial Least Squares Structural Equation Modeling (PLS-SEM) for complex models, specifically in the field of big data analytics quality.
• The authors highlight the increasing significance of quantitative research in social sciences and business, emphasizing the need for causal modeling and validating theories.
• Structural Equation Modeling (SEM) is presented as a crucial tool for linking multiple latent constructs, useful in complex situations.
• The study uses PLS-SEM to estimate a complex, hierarchical model related to big data analytics quality (BDAQ).

Literature Review

• PLS-SEM was introduced as an alternative to Covariance-Based Structural Equation Modeling (CBSEM) to accommodate complex models.
• PLS-SEM is particularly useful when dealing with a larger number of latent variables and indicators, especially with smaller sample sizes.
• Its advantage lies in estimating latent variables without strong distributional assumptions.
• The study emphasizes that PLS-SEM is well-suited for complex models and is becoming increasingly prevalent in research areas like management information systems, strategic management, and marketing.

Research Model for Empirical Illustration

• Big data analytics (BDA) quality (BDAQ) is defined as the perceived excellence or superiority of a BDA platform.
• BDAQ is a hierarchical construct encompassing technology quality (system reliability, adaptability, integration, privacy) and information quality (completeness, accuracy, format, currency).
• The model posits that BDAQ has a positive impact on business value (BVAL) and that BVAL is a mediator in the relationship between BDAQ and user satisfaction (BDAS).

Methodology

• The study collected data from 302 big data analysts in the US and France.
• A 7-point Likert scale was used to measure most of the constructs.
• PLS-SEM was used to estimate the hierarchical reflective-formative model.
• The study employed non-parametric bootstrapping with 5,000 replications for estimating standard errors.

Findings

• The measurement model demonstrated adequate reliability, convergent validity, and discriminant validity.
• The structural model supported the hypothesized relationships, confirming that BDAQ positively affects business value and that business value mediates the relationship between BDAQ and user satisfaction.
• The results indicate that technology quality and information quality within BDAQ are key drivers in achieving business value and user satisfaction.

Conclusion

• PLS-SEM proves its suitability in modeling complex relationships, especially in analyzing the effects of big data analytics quality.
• The study highlights the importance of assessing both technology and information quality within BDAQ to drive business value and satisfaction.
• The authors suggest further research to explore the model's applicability and its potential for incorporating additional factors.