Eliminating quality bias in explaining the OA impact advantage

Yassine Gargouri and Stevan Harnad are measuring how OA mandates affect the OA citation advantage.  They’ve posted two docs with preliminary versions of their findings (1, 2).  Here’s Stevan’s summary, by email:

  1. Green OA self-archiving is at about 15% unmandated by 2006 (with not much sign of spontaneous annual growth)
  2. Green OA self-archiving is at about 60% if mandated (by 2006)
  3. OA adds an independent positive increment to citation frequency, over and above other positive correlates of citation frequency such as article age, journal impact factor, number of co-authors, number of references, field, article length, country, institution
  4. The OA citation advantage is equal for mandated OA and for unmandated OA (hence the citation increase is not an artifact of self-selection for self-archiving the better articles, as some have tried to argue)
  5. The OA citation advantage is greater for the articles in higher-impact journals (hence the "better" articles benefit more from OA).


  • Yassine and Stevan are writing up their study for publication; this is just a preview. 
  • See especially point #4.  While all past studies have shown that OA articles are cited more often than non-OA articles, some have suggested that the effect is due to a tendency of authors to make their best work OA.  The current study has found beautiful leverage on that question:  OA mandates erase the effect of author self-selection.  If all articles from a certain funder or university must be deposited in an OA repository, then the resulting works will not show the effect of author self-selection.  Therefore any remaining citation advantage cannot be attributed to a quality bias in the sample, and is much more likely the result of OA itself.
  • On point #5, see Jonathan Wren’s study in BMJ for April 12, 2005, and my editorial commenting on it.