The multivariate statistics contribution for graduate assessment in Brazil: a study in Information Science

Abstract

This research aims to analyze the contribution of multivariate statistics as a method to support the understanding of the complex methodology of graduate assessment conducted by Coordination for the Improvement of Higher Education Personnel. More specifically, this research seeks to identify the cluster of graduate programs in Information Science generated by cluster analysis and the data generated by boxplot graphics, based on the set of indicators content in the Comparative Spreadsheets of Triennial Assessment of the area, and then comparing groups of graduate programs, according to the concept obtained by the Coordination for the Improvement of Higher Education Personnel assessment, with the result of the cluster obtained by cluster analysis, in order to verify the participation and adequacy of indicators in the allocation of concepts of graduate programs. For all 14 graduate programs of Information Science area, evaluated in the last triennial years (2010-2012), the following indicators were analyzed: total of theses and dissertations, total of papers published, according to Qualis stratification, total of papers published in conference proceedings, books and chapters, full text, collections, and other entries and artistic production published, weighted by faculty member. According to the results, it was observed that while the formed clusters present similar concepts, only the indicators present in the Comparative Spreadsheets of Triennial Assessment were not sufficient to discern them, the inclusion of quality indicators being necessary, since the production indicators are not sufficient to distinguish the programs according to their concepts.