Sparse-rated data are common in operational performance-based language tests, as an inevitable result of assigning examinee responses to a fraction of available raters. The current study investigates the precision of two generalizability-theory methods (i.e., the rating method and the subdividing method) specifically designed to accommodate the…