A Measure of Originality

Perhaps because there is no precise definition of scientific originality, studies of peer review in scientific journals have shown that reviews can be arbitrary or ineffective. In this Note, a potential reference point in evaluating originality that may be useful in analyzing the problem is presented and tested - a typology of scientific originality based on a structural analysis of the scientific paper. As reported in a paper, each of three elements of scientific work (hypothesis, methods and results) either has been previously reported in the scientific literature, or is newly reported. Scientific originality can then be defined as a permutation of new and old information: eight types of originality, ranging from all three elements being previously reported, to all elements being new. To determine if the typology has face validity, highly experienced scientists were asked by mail survey to use the typology in two exercises: rating the eight originality types, and assigning an originality type to highly-cited articles the scientists had written. Of 301 scientists, 206 (68%) responded: between them, they had authored a total of 230 articles. The eight originality types were rated by 84% of the scientists, and an originality type was assigned to 209 of the 230 articles: the most frequent type was new hypothesis/previously-reported methods/new results. To see if scientific journals might vary in the type of originality they prefer, the articles were divided into two equal groups, by the age of the journals publishing them: younger (0-29 years, n = 106) or older (30-185 years, n = 103). The distribution of originality types was virtually identical for the two groups. The results of this study indicate that the typology merits further study as a means of investigating or evaluating scientific originality.