Identification of workplace harassment in higher education teachers based on responses to job satisfaction surveys

Abstract

In this article, the authors propose, using social exchange theory as a theoretical basis, to identify the presence of harassment or bullying in university teachers by means of support-vector machines and the application of an instrument that measures job satisfaction, instead of explicitly assessing the level of bullying. The sample consisted of 248 teachers who work in four public universities in Mexico. We obtained the following results: devaluation of the carried out work is the most frequent type of harassment, while personal mobbing hardly figures. The radial basis function (RBF) kernel is the best option for predicting workplace bullying in the dimensions of work overload, personal mobbing and devaluation of the carried out work, whereas the polynomial kernel is the best for organizational mobbing. The classification accuracy of the models is over 91%, and the F-score = 0.93, both in the worst case. According to the performance of the models, workplace harassment or job mobbing can be predicted accurately.