An empirical comparison of the efficacy of covariance-based and variance-based SEM

Abstract

Variance-based SEM, also known under the term partial least squares (PLS) analysis, is an approach that has gained increasing interest among marketing researchers in recent years. During the last 25 years, more than 30 articles have been published in leading marketing journals that have applied this approach instead of the more traditional alternative of covariance-based SEM (CBSEM). However, although an analysis of these previous publications shows that there seems to be at least an implicit agreement about the factors that should drive the choice between PLS analysis and CBSEM, no research has until now empirically compared the performance of these approaches given a set of different conditions. Our study addresses this open question by conducting a large-scale Monte-Carlo simulation. We show that justifying the choice of PLS due to a lack of assumptions regarding indicator distribution and measurement scale is often inappropriate, as CBSEM proves extremely robust with respect to violations of its underlying distributional assumptions. Additionally, CBSEM clearly outperforms PLS in terms of parameter consistency and is preferable in terms of parameter accuracy as long as the sample size exceeds a certain threshold (250 observations). Nevertheless, PLS analysis should be preferred when the emphasis is on prediction and theory development, as the statistical power of PLS is always larger than or equal to that of CBSEM; already, 100 observations can be sufficient to achieve acceptable levels of statistical power given a certain quality of the measurement model.

Perceived autonomy support and autonomous motivation toward mathematics activities in educational and out-of-school contexts is related to mathematics homework behavior and attainment

“…However, unlike methods used in covariance-based SEM, the partial least-squares algorithm is based on ranked data and is, therefore, distribution-free (i.e., the estimation is less affected by the complexity of the model, small sample size, or non-normality of the data). This makes it ideal for use with the current data set given the complexity of the model and the greater statistical power offered by the VB-SEM method (Reinartz, Haenlein, & Henseler, 2009). In the proposed model, each transcontextual model construct was represented as a latent variable indicated by the set of items proposed to measure that construct.…”

CSR and technology companies: A study on its implementation, integration and effects on the competitiveness of companies

“…• PLS can be estimated models with small samples, in fact, the PLS modeling algorithms tend to get results with high levels of statistical power (Reinartz, Haenlein & Henseler, 2009), even when the sample size is very modest (Rigdon, 2014 SmartPLS is an estimation model and SEM analysis, uses the estimation process in two steps, evaluating the model measurement and structural model (Hair Jr. et al, 2014). First, the measurement model where the relationship between the indicators and the construct is determined (Roldán & Sánchez-Franco, 2012 (Hair Jr. et al, 2014).…”

“…• PLS puede estimar modelos con muestras pequeñas, de hecho, los algoritmos de modelado de PLS tienden a obtener resultados con altos niveles de potencia estadística (Reinartz, Haenlein & Henseler, 2009), incluso cuando el tamaño de la muestra es muy modesto (Rigdon, 2014).…”

Impacts of the CSR strategies of technology companies on performance and competitiveness

“…Several previous studies have indicated that the required sample size is between 30 and 100 cases if the conceptual model has at least 3 or 4 indicators per construct (Roldán & Sánchez-Franco, 2012). Moreover, a Monte Carlo simulation by Chin and Newsted (1999) showed that PLS can collect significant information from samples as small as 20 (Reinartz, Haenlein, & Henseler, 2009). Therefore, our sample was considered sufficiently large enough to carry out a statistical analysis using structural equationmodelling (SEM) based on PLS.…”