A serious threat to the validity of psychological assessments is differential item functioning (DIF; Bauer, 2020), that is, a situation, in which test items function differently across different groups. The moderated nonlinear factor analysis (MNLFA) approach allows for simultaneous testing of DIF in multiple categorical and continuous covariates if those covariates are specified a priori and if their DIF is linear. SEM trees and forests (Brandmaier et al., 2013; Brandmaier et al., 2016) could provide a viable alternative to exploring DIF in latent factor models. SEM trees allow for non-linear DIF effects and provide a way to systematically search for important covariates (and their interactions). In this project, we design a simulation study to investigate the conditions under which each of the two methods returns favorable results.