A comparison of two ways of evaluating research units working in different scientific fields

摘要: This paper studies the evaluation of research units that publish their output in several scientific fields. A possible solution relies on prior normalization raw citations received by publications all In a second step, citation indicator is applied to units' field-normalized distributions. this paper, we also study an alternative begins applying size- and scale-independent impact distributions any unit calculated as average (weighted publication output) achieves each field. The two alternatives are confronted using 500 universities 2013 edition CWTS Leiden Ranking, whose evaluated according indicators with very different properties. We use large Web Science dataset consisting 3.6 million articles published 2005---2008 period, classification system distinguishing between 5119 clusters. main findings follows. Firstly, differences production practices 3332 clusters more than 250 account for 22.5 % overall inequality. After standard field-normalization procedure, where cluster mean used factors, quantity reduced 4.3 %. Secondly, university rankings solutions all-sciences aggregation problem small order magnitude both indicators.