This work analyses the correlation between decisions regarding the CNPq research scholarships (taken at the end of 2009) and some bibliometric measures. For all levels, and for all scientific subareas we calculated the correlation of the decision to raise, keep or lower the researcher's evaluation level and many bibliometric measures such as: total production (journal and conference papers, book an book chapters), production in the last 5 years, total production and last 5 years production that is indexed in the Web of Science, citations per article received, citations per article published in the last 5 years, H index and so on. We used both Google Scholar and Web of Science to obtain the citation data. The correlations are aggregated for all subareas of each of the CNPq large subject areas (Agriculture Sciences, Biology Sciences, Exact Sciences, Human Sciences, Health Sciences, Social Sciences, Engineering and Arts). We show which bibliometric measure has the higher correlation to the scholarship decisions, for each level and for each large area. We discuss some of the cases where there seems to be a coherent theme regarding the most highly correlated measure across all levels of the scholarship.
Bibliometrics; Peer evaluation; CNPq; Scholarship
