A TYPOLOGY OF SCIENTIFIC ORIGINALITY:
INITIAL VALIDITY TESTING
Lynn Dirk
Institutional Review Board, University of Florida, Health Center, Box 100173, Gainesville, FL, 32610, USA
Objective: A typology of scientific originality was derived from the following elements of the scientific article: Introduction, Methods, and Results, which describe, respectively, the hypothesis, methods, and results of the scientific work being reported. For any article, the content of each section is either previously published in the scientific literature and thus "established" (E) or newly published (N). A permutation of the 3 elements as E or N yields 8 idealized types of originality, idealized because much, if not most, of science consists of modifications of previously published work. As a preliminary test of the validity of the typology, experienced scientists were asked to use the typology in 2 exercises.
Design: By mail survey, scientists who had authored highly cited, peer-reviewed biomedical articles (Citation Classics, Institute of Scientific Information) were asked to (1) rate the 8 idealized types of originality and (2) classify the hypothesis, methods, and results reported in their highly cited articles as E or N or "modified"; if modified, the scientists were asked to further characterize the work as slightly or substantially modified. For analysis, however, "modified" was transformed to E to highlight elements classified as N.
Results: Of 301 scientists, 205 responded to the survey and reported on 230 Classics. Eighty-one percent (n=167) of the scientists rated the 8 idealized patterns of originality, and the scientists classified 91% of the articles (n=209).
Conclusions: The results of this study indicate that the typology may have validity as a measure of scientific originality; more than 150 experienced scientists agreed to and were able to use the typology in 2 different exercises. The scientists' analyses of their articles' originality, however, may be biased; therefore, these data should not be considered an indication of the articles' originality. Because authors of scientific articles are required to clearly state in their articles what is new and what has been previously published (through citation), using statements with citations in each section of an article to analyze its originality may provide a consistent, objective measure of scientific originality. Therefore, as a test of reliability, the analyses of articles' originality produced by the authors could be compared with analyses of the same articles produced by nonauthors blinded to authors' analyses. If the typology is too complex to be useful for scientific journal peer review, this technique may at least prove to be a useful heuristic in studying the process of science.
Return to Session Information
