Guest editorial: Sports management research using partial least squares structural equation modeling (PLS-SEM)

The development of PLS-SEM into a standard method in management research Wold (1982) developed themethod of partial least squares structural equationmodeling (PLSSEM) (also see J€oreskog and Wold, 1982b, c). Lohm€oller (1989) extended the initial method and prepared the first software (LVPLS; see Lohm€oller, 1984, 1987). The dissemination of PLS-SEM started with the publication of tutorials by Barclay et al. (1995), Chin (1998) and Tenenhaus et al. (2005), as well as the availability of PLSGraph (Chin, 2003), the first software package that included a graphical user interface for Lohm€oller’s LVPLS program. In the last decade, PLS-SEM has become an integral part of the methodological toolbox of economists, sociologists and several other social science disciplines. This is not only evidenced by the exponentially increased use of PLS-SEM in research articles (Hair et al., 2022, Preface) but also by dedicated chapters on PLS-SEM in main methods textbooks (e.g. Hair et al., 2019a) and handbooks (e.g. Sarstedt et al., 2021b). Several review articles on the use of PLS-SEM (Table 1), special issues (Table 2) in different disciplines and dissemination of research networks (Khan et al., 2019) echo this finding. The publication of textbooks (e.g. Garson, 2016; Hair et al., 2018, 2022; Ramayah et al., 2018; Wong, 2013) and “how to” articles (e.g. CepedaCarri on et al., 2019; Hair et al., 2011a, 2019b; Rold an and S anchez-Franco, 2012) on PLS-SEM further contributed to the method’s spread, as well as the availability of R software packages, like cSEM (Rademaker and Schuberth, 2020) and SEMinR (Hair et al., 2021b; Ray et al., 2021) and GUI-based standalone software programs, like PLSGraph (Chin, 2003) and SmartPLS (Ringle et al., 2005, 2015). Of these applications, SmartPLS is particularly popular among users due to its ease of use and functionality (Memon et al., 2021; Sarstedt and Cheah, 2019). In the early phase of PLS-SEM, much of the discussion focused on comparisons with covariance-based SEM (CB-SEM). However, CB-SEM and PLS-SEM are different statistical methods, and the algorithms are designed to achieve different research objectives (J€oreskog and Wold, 1982a; Lohm€oller, 1989). Their results also differ (Rigdon, 2012; Sarstedt et al., 2016), and each method is known for its unique modeling capabilities (Rigdon et al., 2014, 2017). The two approaches should therefore be viewed as complementary rather than competitive. CB-SEM is particularly useful when the research objective is to test the theoretically established model structure as a whole (i.e. how well the model-implied covariance matrix fits the covariance matrix of the data sample). In contrast, PLS-SEM aims to minimize the residuals of the measurement models as well as the structural model. PLSSEM therefore excels when the focus is on prediction-oriented research and is particularly useful when the goal is to identify the sources of competitive advantage and in success factor research (Albers, 2010; Sarstedt et al., 2021b). The choice between CB-SEM and PLS-SEM should also consider the type of model to be estimated. This is the position of Rigdon et al. (2017) when they state: “Our third recommendation is that researchers use a technique that is consistent with the type of model that they intend to estimate – in other words, that they correctly estimate their chosen model. There has been a tendency in the literature to treat CB-SEM and PLS-SEM as if they were Guest editorial