Table 1.
Characteristics of all Nature journals included in the study. Research and editorial experiences are computed for editors who had no homonym: thus, in some journals, only one editor satisfied this criterion, yielding a standard deviation (SD) equal to 0. ‘Editorial experience’ is limited to the experience at Nature journals. Research and postdoc were experienced prior to the editors being appointed by Nature journals. The number of articles as first or last author refers to original articles (see supplementary method for the criteria used to define original articles).
Fig 1.
Inclusion flow charts for journals (A) and editors of Nature journals (B).
Table 2.
General linear statistical models and mixed statistical models used in the analyses to plot the figures. The coefficients (beta) of the regressors in bold are used to plot the figures of the main text referenced in the first column. Regressors in italic are categorical variables. The random effect of all mixed models is set over the intercept.
Fig 2.
Average difference in the number of articles published by each author in the editor’s journal during the two years before and the two years after the editor’s appointment at Nature journals as a function of the status of the author (A); Modulation value of the orange and red slopes of Fig 1C (computed for articles of new authors) as a function of editor’s research experience and editorial experience at Nature journals (B); yearly percentage of original articles published by new authors only (orange and red) or by known intermediate authors (light and dark blue) in Nature journals as a function of year after journal launch (C, slope (standard error of the mean (SEM)): -1.34 (0.29), p<0.00001 (orange); -1.98 (0.15), p<0.00001 (red); 0.52 (0.16), p=0.001 (light blue); 0.76 (0.08), p<0.00001 (dark blue)).
In Fig 2B, the modulation captures the amount which must be added (if the modulation is positive), or subtracted, to the slope of Fig 2C in order to get the slope value for each editor. This amount is equal to the coefficient depicted in Fig 2C (e.g. the medium grey bar) multiplied by the number of years of academic research of the editor (for the medium grey bar). Comparisons of bars against zero: *: p<0.05; **: p<0.01; ***: p<0.001. Data were truncated to January 1st 1990 for Nature.
Fig 3.
Linear association (i.e., beta) between the variations in impact factor (i.e. IFyear n-IFyear n-1) and subsequent variations in articles published (i.e. Noriginal articles year n+1- Noriginal articles year n (respectively Noriginal articles year n+2- Noriginal articles year n+1 for the two bars on the right)) in Nature and Nature Communications (A); Average difference in the number of articles published by authors in the editor’s journal during the two years before and the two years after the editor’s appointment at Nature journals as a function of the status of authors and when the editor was appointed (i.e., before or after the establishment of the non-financial conflict of interest (COI) policy, B); Same as Fig 3B except that the number of articles published by editors’ former co-authors is replaced by the total number of articles by authors affiliated to research institutions where the editor worked before joining Nature journals (C).
Comparisons of bars against zero: *: p<0.05; **: p<0.01; ***: p<0.001.