Ringle, Christian M.Christian M.RingleSarstedt, MarkoMarkoSarstedtSchlittgen, RainerRainerSchlittgenTaylor, Charles R.Charles R.Taylor2019-12-122019-12-122013-09-01Journal of Business Research 66 (9): 1318-1324 (2013-09-01)http://hdl.handle.net/11420/4038Applications of the partial least squares (PLS) path modeling approach-which have gained increasing dissemination in business research-usually build on the assumption that the data stem from a single population. However, in empirical applications, this assumption of homogeneity is unrealistic. Analyses on the aggregate data level ignore the existence of groups with substantial differences and more often than not result in misleading interpretations and false conclusions. This study introduces a genetic algorithm segmentation method for PLS path modeling (PLS-GAS) that accounts for the critical issue of unobserved heterogeneity in the path model's estimates of relations. The results from computational experiments allow a primary assessment to substantiate that PLS-GAS effectively uncovers unobserved heterogeneity. Significantly distinctive segment-specific path model estimates further foster the development of differentiated results that render more effective recommendations.en0148-29632013913181324Genetic algorithmHeterogeneityPartial least squaresPath modelingSegmentationAllgemeines, WissenschaftJournal Article10.1016/j.jbusres.2012.02.031Other