A structural equation model (SEM) is considered for determining rankings of institutions. In the proposed algorithm, we treat the rankings of institutions as missing values and then estimate the construct variables based on SEM using partial least squ...
A structural equation model (SEM) is considered for determining rankings of institutions. In the proposed algorithm, we treat the rankings of institutions as missing values and then estimate the construct variables based on SEM using partial least squares. The weights between manifest and construct variables as well as the coefficient between construct variables will be determined iteratively according to regression models. Differently from the usual weight and sum method, weights will be determined internally by the algorithm. Several simulations are performed to validate the proposed algorithm.