DOES TEACHER'S EDUCATION AFFECT STUDENT PERFORMANCE?

Increasingly questions arise regarding the educational process and which factors are of key importance to increase the quality of education. According to Clare Kosnik, "Quality in education depends directly on the quality of teachers and the quality of the curriculum." However, to what extent is this true? It is known that, many agents intervene in the educational process as the quality of curriculum, students, teachers, teaching methods, governance, financing, evaluation and linkage with other apex institutions. In this sense, it is interesting to analyse how the quality of education should be evaluated, as a direct consequence of teacher training, the results of students, the conditions and strategies in which teaching is developed, or an evaluation in which all do these aspects interfere. This work aims to address one of the above questions, which causes major debates among educational analysts: to what extent do education level and research outcomes of teachers interfere with the skills and performance of higher education students? Specifically, we examined 165 higher education institutions (HEI), public and private, in Colombia in the year of 2016. The Colombian HEI system is composed of technical institutes focused on vocational education, university institutions focused on technological education, and universities focused on under-graduate degrees (e.g. bachelor or diploma) and post-graduate degrees (e.g. specialisation, master or doctorate). The data include twelve variables: six variables related with teacher education level and research outcomes (i.e. teachers with specialisation, teachers with masters, teachers with doctorate, teachers without postgraduate, number of citations and intellectual production score) and six variables related with student performance (i.e. quantitative reasoning, critical reading, written communication, employability and postgraduate ingresses). A canonical correlation analysis (CCA) was executed to explore relationship between academic qualification of teachers of higher education and performance of their students, aiming to identify the main canonical structures between them. Previously a methodology filtering were used to deal with outliers and Box-Cox transformations were applied to transform non-normal variables in near normal ones. Statistical tests based on Wilks Lambda, Hotelling-Lawley Trace, Pillai-Bartlett Trace and Roy's Largest Root was applied in order to test for statistical significance of canonical correlation coefficients. These approaches identify the degree and the most relevant relationships between variables of teacher training/research and variables of student performance in higher education.