1. EA: Early Advantage: Self-archiving preprints before publication increases citations (higher-quality articles benefit more) QA: Quality Advantage: Self-archiving postprints upon publication increases citations (higher-quality articles benefit more) UA: Usage Advantage: Self-archiving increases downloads (higher-quality articles benefit more) (CA: Competitive Advantage): OA/non-OA advantage (CA disappears at 100%OA) (QB: Quality Bias): Higher-quality articles are self- selectively self-archived more (QB disappears at 100%OA) OA Advantage OAA = EA + QA + UA + (CA) + (QB)

2. Self-Selected Self-Archiving vs. Mandated Self-Archiving I: Non-Normalised Grand Average Citation Ratios across fields for year 2004). S= articles self-archived at institutions with a self-archiving mandate (Sm, 237 articles) and without (Sn, 890). N = citation counts for non- archived articles at institutions with (Nm, 16485) and without (Nn, 89156) mandate (i.e., Nm = articles not yet compliant with mandate).There is no indication that Sn ratios are greater than Sm ratios: rather the contrary. (NB: Preliminary, unrefereed results. These averages are across fields and based on very different samples sizes. Following figure (II) compares like with like)

3. Self-Selected Self-Archiving vs. Mandated Self-Archiving II: Within-Journal Citation Ratios (for 2004, all fields). S = citation counts for articles self-archived at institutions with (Sm) and without (Sn) self-archiving mandate. N = citations for non-archived articles at institutions with (Nm) and without (Nn) mandate (i.e., Nm = articles not yet compliant with mandate). Grand average of (log) S/N ratios (106203 articles; 279 journals) is the OA advantage (18%); this is about the same as for Sn/Nn (27972 articles, 48 journals, 18%) and Sn/N (17%); ratio is higher for Sm/N (34%), higher still for Sm/Nm (57%, 541 articles, 20 journals); and Sm/Sn = 27%, so self-selected self- archiving does not yield more citations than mandated; rather the reverse. (All six within-pair differences significant: correlated sample t-tests.) (NB: preliminary, unrefereed results.)

4. 1. article age 2. journal impact factor 3. number of authors 4. open access Multiple Regression Analysis reveals 4 independent influences on citation counts (overall, and in all subsets): 1. article age 2. journal impact factor 3. number of authors 4. open access Raw citation countsLog citation counts

5. Stevan Harnad: Southampton and Montreal Metrics Citations (C) CiteRank Co-citations Downloads (D) C/D Correlations Hub/Authority Chronometrics: Latency/Longevity Endogamy/Exogamy Semiometrics

6. The G-factor International University Ranking measures the importance of universities as a function of the number of links to their websites from the websites of other leading international universities. Why is Southampton ranked 3rd highest in the UK and 25th in the world, above Columbia (27th) and Yale (51st)? Copyright Peter Hirst, 2006.

7. Reasons for U. Southampton's High Webmetric Rank: (1) U. Southampton's university-wide research performance (2) U. Southampton's Electronics and Computer Science (ECS) Department's involvement in many high-profile web projects and activities (among them the semantic web work of the web's inventor, ECS Prof. Tim Berners-Lee, the Advanced Knowledge Technologies (AKT) work of Prof. Nigel Shadbolt, and the pioneering web science contributions of Prof. Wendy Hall) (3) Since 2001 U. Southampton's ECS has had a mandate requiring that all of its research output be made Open Access on the web by depositing it in the ECS EPrints Repository, and that Southampton has a university-wide self-archiving policy (soon to become a mandate) too (4) Maximising access to research (by self-archiving it free for all on the web) maximises research usage and impact (and hence web impact) This all makes for an extremely strong Southampton web presence, as reflected in such metrics as the "G factor", which places Southampton 3rd in the UK and 25th among the world's top 300 universities or Webometrics,which places Southampton 6th in UK, 9th in Europe, and 80th among the top 3000 universities it indexes.G factor

8. Stevan Harnad: Southampton and Montreal “Online or Invisible?” (Lawrence 2001) “average of 336% more citations to online articles compared to offline articles published in the same venue” Lawrence, S. (2001) Free online availability substantially increases a paper's impact Nature 411 (6837): 521. http://www.neci.nec.com/~lawrence/papers/online-nature01/

9. Stevan Harnad: Southampton and Montreal Lawrence (2001) findings for computer science conference papers. More OA every year for all citation levels; higher with higher citation levels

10. Stevan Harnad: Southampton and Montreal Citation impact for articles in the same journal and year are consistently higher for articles that have been self-archived by their authors. (Below is a comparison for Astronomy articles that are and are not in ArXiv.)

11. Stevan Harnad: Southampton and Montreal DATA: Michael Kurtz

12. Stevan Harnad: Southampton and Montreal DATA: Michael Kurtz

13. Stevan Harnad: Southampton and Montreal DATA: Michael Kurtz

14. Stevan Harnad: Southampton and Montreal DATA: Michael Kurtz

15. Stevan Harnad: Southampton and Montreal Changing citation behaviour

16. Stevan Harnad: Southampton and Montreal Astrophysics General Physics HEP/Nuclear Physics Chemical Physics

17. Stevan Harnad: Southampton and Montreal Social Sciences Biological Sciences The citation impact advantage is found in all fields analyzed so far, including articles (self-archived in any kind of open-access website or archive) in social sciences (above right) biological sciences (below right) and all fields of Physics (self-archived in ArXiv, below). Note that the percentage of published articles that have been self-archived (green bars) varies from about 10-20%from field to field and that the size of the open-access citation impact advantage (red bars) varies from about 25% to over 300%, but it is always positive. http://opcit.eprints.org/oacitation-biblio.html Signal detection analysis of the hit/miss rate of the algorithm that searched for full-text OA papers on the web: d’ = 2.45 (sensitivity) b =.52 (bias)

18. gray By discipline: total articles (OA+NOA), gray curve; percentage OA: (OA/(OA+NOA)) articles, green bars; percentage OA citation advantage: ((OA-NOA)/NOA) citation, red bars, averaged across 1992-2003 and ranked by total articles. All disciplines show an OA citation advantage (Hajjem et al. IEEE DEB 2005)Hajjem et al. IEEE DEB 2005

19. By country: total articles (gray curve), percent OA articles (green bars), and percent OA citation advantage (red bars); averaged across all disciplines and years 1992-2003; ranked by total articles. (Hajjem et al. IEEE DEB 2005)Hajjem et al. IEEE DEB 2005

20. By year: total articles (gray curve), percent OA articles (green bars), and percent OA citation advantage (red bars): 1992-2003, averaged across all disciplines. No yearly trend is apparent in the size of the OA citation advantage, but %OA is growing from year to year. (Hajjem et al. IEEE DEB 2005)Hajjem et al. IEEE DEB 2005

21. Stevan Harnad: Southampton and Montreal Figure 3a: The yearly percentage (OAc) of the articles with c citations (c = 0, 1 2-3, 4-7, 8-15, 16+) that are OA (1992-2003). This graph should really be read backwards, as citations increase cumulatively as an article gets older (younger articles have fewer citations). Reading backwards, for articles with no citations (c=0), the percentage OAc decreases each year from 2003-1992, at first rapidly, then more slowly. For articles with one and more citations (c>0), OAc first increases rapidly from 2003 till about 1998, then decreases slowly 1998-1992. Notice that the rank order becomes inverted around midway (c. 1998), the percentages increasing from c=0 to c=16+ for the oldest articles (1992) and the reverse for the youngest articles (2003). The pattern is almost identical for NOA articles too (see NOAc inset), so this is the relationship between citation ranges and time for all articles, not a specific OA effect. The OA effect only becomes apparent when we look at OAc/NOAc (Figure 3b) (Hajjem et al. IEEE DEB 2005)Hajjem et al. IEEE DEB 2005 Figure 3b: The yearly ratio OA c /NOA c between the percentage of articles with c citations (c = 0, 1 2-3, 4-7, 8-15, 16 + ) that are OA and NOA (all disciplines). This ratio is increasing with time (as well as with higher citation counts, c), showing that the effect first reported for computer science conference papers by Lawrence (2001) occurs for all disciplines.

22. Stevan Harnad: Southampton and Montreal OA c /NOA c ratio (across all disciplines and years increases as citation count (c) increases (r =.98, N=6, p<.005). Percentage of articles is relatively higher among NOA articles with Citations = 0; it becomes higher among OA articles with citations = 1 or more. The more cited an article, the more likely that it is OA. (Hajjem et al. IEEE DEB 2005)Hajjem et al. IEEE DEB 2005

23. Stevan Harnad: Southampton and Montreal Measure usage and impact

24. Stevan Harnad: Southampton and Montreal Time-Course and cycle of Citations (red) and Usage (hits, green) Witten, Edward (1998) String Theory and Noncommutative Geometry Adv. Theor. Math. Phys. 2 : 253. 1. Preprint or Postprint appears. 2. It is downloaded (and sometimes read). 3. Next, citations may follow (for more important papers)…. 4. This generates more downloads… 5. More citations...

25. Stevan Harnad: Southampton and Montreal Usage Impact (downloads) is correlated with Citation Impact (Physics ArXiv: hep, astro, cond, quantum; math, comp) http://citebase.eprints.org/analysis/correlation.php downloads from first 6 months after publication predict citations 2 years after publicattion http://citebase.eprints.org/analysis/correlation.php (Quartiles Q1 (lo) - Q4 (hi)) All r=.27, n=219328 Q1 (lo) r=.26, n=54832 Q2 r=.18, n=54832 Q3 r=.28, n=54832 Q4 (hi) r=.34, n=54832 hep r=.33, n=74020 Q1 (lo) r=.23, n=18505 Q2 r=.23, n=18505 Q3 r=.30, n=18505 Q4 (hi) r=.50, n=18505 (correlation is highest for high- citation papers/authors) Most papers are not cited at all Average UK downloads per paper: 10 (UK site only: 18 mirror sites in all)

26. Stevan Harnad: Southampton and Montreal Research Assessment, Research Funding, and Citation Impact “Correlation between RAE ratings and mean departmental citations +0.91 (1996) +0.86 (2001) (Psychology)” “ RAE and citation counting measure broadly the same thing ” “Citation counting is both more cost-effective and more transparent” (Eysenck & Smith 2002) http://psyserver.pc.rhbnc.ac.uk/citations.pdf