Why You Shouldn't Use PLS: Four Reasons to Be Uneasy about Using PLS in Analyzing Path Models

When it was originally introduced, Partial Least Squares (PLS) was designed primarily for exploratory studies focusing on prediction (rather than hypothesis testing). Over time, however, PLS has become a very popular statistical analysis technique for testing hypothesized relationships (confirmatory studies) within MIS research. In this paper we note some challenges that have been raised relative to PLS, and then focus on four assertions concerning its use that we believe are problematic. We show that these frequently stated assertions of PLS strength actually suggest important weaknesses. In particular, we show evidence that PLS seems to capitalize on chance correlations among indicators at sample sizes typically used in MIS research. Our belief is that many PLS users are unaware of the trade-offs involved in PLS use, and we recommend a much reduced (or possibly non-existent) role for PLS in confirmatory work.

[1]  F. Bookstein,et al.  Two Structural Equation Models: LISREL and PLS Applied to Consumer Exit-Voice Theory , 1982 .

[2]  Jacob Cohen.  Statistical Power Analysis for the Behavioral Sciences , 1969, The SAGE Encyclopedia of Research Design.

[3]  Wynne W. Chin.  The partial least squares approach for structural equation modeling. , 1998 .

[4]  Mikko T. Siponen,et al.  Neutralization: New Insights into the Problem of Employee Systems Security Policy Violations , 2010, MIS Q..

[5]  Scott B. MacKenzie,et al.  Construct Measurement and Validation Procedures in MIS and Behavioral Research: Integrating New and Existing Techniques , 2011, MIS Q..

[6]  Jukka Ylitalo,et al.  International Conference on Information Systems ( ICIS ) 2010 CONSTRUCT VALIDITY IN PARTIAL LEAST SQUARES PATH MODELING , 2017 .

[7]  Martin Wetzels,et al.  On the Use of Formative Measurement Specifications in Structural Equation Modeling: A Monte Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares Model Estimation Methodologies , 2009 .

[8]  D. Straub,et al.  Editor's comments: a critical look at the use of PLS-SEM in MIS quarterly , 2012 .

[9]  Mary Tate,et al.  Testing Models or Fitting Models? Identifying Model Misspecification in PLS , 2010, ICIS.

[10]  William Lewis,et al.  Does PLS Have Advantages for Small Sample Size or Non-Normal Data? , 2012, MIS Q..

[11]  Jeffrey R. Edwards,et al.  The Fallacy of Formative Measurement , 2011 .

[12]  Wynne W. Chin,et al.  A Partial Least Squares Latent Variable Modeling Approach for Measuring Interaction Effects: Results from a Monte Carlo Simulation Study and an Electronic - Mail Emotion/Adoption Study , 2003, Inf. Syst. Res..

[13]  D. Rubinfeld,et al.  Econometric models and economic forecasts , 2002 .

[14]  Henry C. Lucas,et al.  The Value of IT-Enabled Retailer Learning: Personalized Product Recommendations and Customer Store Loyalty in Electronic Markets , 2011, MIS Q..

[15]  Michael H. Kutner.  Applied Linear Statistical Models , 1974 .

[16]  Herman Wold,et al.  Systems under indirect observation : causality, structure, prediction , 1982 .

[17]  Herman Wold,et al.  Soft modelling: The Basic Design and Some Extensions , 1982 .

[18]  Straub,et al.  Editor's Comments: An Update and Extension to SEM Guidelines for Administrative and Social Science Research , 2011 .

[19]  Wynne W. Chin,et al.  Extending the technology acceptance model: the influence of perceived user resources , 2001, DATB.