| This thesis examines, using the methods of clinical epidemiology, the hypothesis that medical journals modify significantly the impact of the information they transmit on the users of that information. Three primary studies were conducted. First, the impact of early release of journal articles via the Internet was compared with articles released through the traditional publishing process, using a matched cohort design to control for time, research area, and journal. Early-release articles had twice as many citations and Internet downloads per year as control articles. The second study was subsequently designed to overcome a limitation of the preceding one, namely that despite optimal methods to control for multiple relevant article characteristics, different research articles still differ in content. To address this, an instance of widespread duplicate publication was analyzed, revealing marked differences in citation among different journals, despite identical content. In the third study, the effect of an acronym study name on citation rate was analyzed, using systematic reviews from the Cochrane Collaboration to identify multiple studies addressing the same research question and employing hierarchical modeling to account for clustering. Studies with acronym names were cited about twice as often as non-acronym studies, an effect that persisted after adjusting for multiple other variables. These studies provide evidence that factors controlled at the journal level, which affect the context rather than the content of research presentation, can have significant effects on uptake of that information, illustrating both the influence of journals on science and the need for editorial decision-making to be subject to scientific guidance and public accountability. |
