Frequentist and Bayesian Approaches in the Study of Academic Plagiarism. An Innovative Proposal in Educational Research

The advance experienced by scientific research at the methodological level is scarce in the pedagogical field. With the aim of contributing to improve the situation, a methodological proposal is presented that allows overcoming the possible existing limitations in the interpretation of the significance tests of a study carried out on plagiarism with a sample of 10,800 university students. Its main objective is to show how an adequate combination of approaches (frequentist and bayesian) allows a better understanding of the problem of student plagiarism. The procedure followed consisted of submitting to linear regression analysis, both frequentist and Bayesian, the results obtained by studying the responses of the students to the 49 items of the questionnaire validated for this purpose. The result is the same in both approaches. Their analysis indicates a high significance, both for R2, and for 9 of the 11 factors that appear as causes of academic plagiarism. Of these, the total copy and the partial copy have the greatest weight. They have something minor: opportunity, work overload, ignorance and malpractice. All this allows us to design a Bayesian belief network as a tool for observing the influences of the predictive variables on the dependent variable.