“…Specifically, USM derives starting values for the model's latent variables through principal component analysis, and then applies the Bayesian neural network approach to discover the optimal system of linear, nonlinear, and interactive pathways among the latent variables, with the aim of maximizing the variance explained (e.g., Henseler, Hubona, & Ray, 2016; McIntosh, Edwards, & Antonakis, 2014; Rigdon, Ringle, & Sarstedt, 2010). As a result, the USM approach typically yields higher explanatory power than PLS‐PM‐based linear modeling (Albashrawi, Kartal, Oztekin, & Motiwalla, 2017). Because USM has recently received increasing attention in business research (e.g., Al‐Ebbini, Oztekin, & Chen, 2016; Garbe & Richter, 2009; Oztekin, Kong, & Delen, 2011; Turkyilmaz, Oztekin, Zaim, & Demirel, 2013; Turkyilmaz, Temizer, & Oztekin, 2018), investigating the method's predictive accuracy would be particularly promising.…”