Factors Associating with the Future Citation Impact of Published Articles: A Statistical Modelling Approach

Abstract

This study investigates a range of metrics available when an article is published to see which metrics associate with its eventual citation count. The purposes are to contribute to developing a citation model and to inform policymakers about which predictor variables associate with citations in different fields of science. Despite the complex nature of reasons for citation, some attributes of a paper’s authors, journal, references, abstract, field, country and institutional affiliations, and funding source are known to associate with its citation impact. This thesis investigates some common factors previously assessed and some new factors: journal author internationality; journal citing author internationality; cited journal author internationality; cited journal citing author internationality; impact of the author(s), publishing journal, affiliated institution, and affiliated country; length of paper; abstract and title; number of references; size of the field; number of authors, institutions and countries; abstract readability; and research funding. A sample of articles and proceedings papers in the 22 Essential Science Indicators subject fields from the Web of Science constitute the research data set. Using negative binomial hurdle models, this study simultaneously assesses the above factors using large scale data. The study found very similar behaviours across subject categories and broad areas in terms of factors associating with more citations. Journal and reference factors are the most effective determinants of future citation counts in most subject domains. Individual and international teamwork give a citation advantage in majority of subject areas but inter-institutional teamwork seems not to contribute to citation impact.