Measuring Disagreement In Science

Dispute in science is central to the production of new knowledge. Such disputes leave traces in scholarly documents, generally through the form that is taken by citations. Based on the full text of scholarly papers from the Elsevier ScienceDirect database published between 1980 and 2016, this paper develops a methodology for investigating disagreement in science. Several signal phrases of disagreement are tested, and two are used ("contradict" and "conflict", with the filter phrases "studies" or "results") to assess the prevalence of disagreement across position within a paper and across disciplines. Results show that disagreement is relatively more common in the introduction and discussion sections of papers, as well as in fields of biomedical sciences, health sciences, social sciences, and humanities.