Impact factor bias and proposed adjustments for its determination

Background: The impact factor (IF), a qualitative parameter used to evaluate scientific journals, has several flaws. The aim of the study was to evaluate two of its important constraints, journal self-citation and scientific field, and to investigate the potential for improvement. Methods: We studied the five or six highest impact journals from each of seven medical fields: anesthesiology, dermatology, genetics and heredity, immunology, general and internal medicine, ophthalmology and surgery. To correct for journal self-citation, we divided the number of 1998 citations of papers published in 1996 and 1997, minus the self-citations, by the number of papers published in the same period. For inter-field normalization we divided the IF by the mean of the IFs of the upper quartile for the same category of medical field (IF/f(cat) ). Results: For the 36 journals, there was a negative correlation between IF and self-cited and self-citing rates (r (s) = -0.765, P < 0.001 and r (s) = -0.479, P < 0.003, respectively). Self-cited rate is the ratio of a journal's self-citations to the number of times it is cited by all journals including itself. Self-citing rate relates a journal's self-citations to the total references it makes. The IF/f(cat) for the 36 journals are positively correlated with their conventional IF (r (s) = 0.91, P < 0.001). Conclusion: Correcting the IF of the 36 journals for self-citation did not significantly change journal rankings. The adjusted IF/f(cat) to normalize for the scientific field was positively correlated with the conventional IF.