Personalizing papers using Altmetrics: comparing paper ‘Quality’ or ‘Impact’ to person ‘Intelligence’ or ‘Personality’ Brett Buttliere 1 • Ju ¨ rgen Buder 1 Received: 8 July 2016 / Published online: 3 February 2017 Ó Akade ´miai Kiado ´, Budapest, Hungary 2017 Abstract Despite their important position in the research environment, there is a growing theoretical uncertainty concerning what research metrics indicate (e.g., quality, impact, attention). Here we utilize the same tools used to study latent traits like Intelligence and Personality to get a quantitative understanding of what over 20 common research metrics indicate about the papers they represent. The sample is all of the 32,962 papers PLoS published in 2014, with results suggesting that there are at least two important underlying factors, which could generally be described as Scientific Attention/Discussion (citations), General Attention/Discussion (views, tweets), and potentially Media Attention/Discussion (media mentions). The General Attention metric is correlated about .50 with both the Academic and Media factors, though the Academic and Media attention are only correlated with each other below .05. The overall best indicator of the dataset was the total lifetime views on the paper, which is also probably the easiest to game. The results indicate the need for funding bodies to decide what they value and how to measure it (e.g., types of attention, quality). Keywords Bibliometrics Á Altmetrics Á Citation counts Á Psychometrics Á Factor analysis Á Intelligence Á Personality Á PLoS Introduction Determinations of which research to fund plays a crucial role in the process of science, ultimately deciding who has jobs and what work is done; it is this decision making power that makes identifying good science a necessary and worthwhile endeavor. As a response to the growing desire to identify and reward good science, there has been a sharp growth in & Brett Buttliere b.buttliere@iwm-tuebingen.de 1 Leibniz-Institut fur Wissensmedien, Tu ¨bingen, Germany 123 Scientometrics (2017) 111:219–239 DOI 10.1007/s11192-017-2246-9