On 15 Sep 2006, at 13:08, Theo Andrew wrote:

> Congratulations Simon on passing the 1000 item milestone. Hopefully  
> many
> more repositories in the UK will pass this figure soon!

Further to my previous posting on viable repositories (of which  
Cranfield is an exemplar), I thought I should try to differentiate  
between repositories according to the size of the institutions that  
they serve. My previous estimate of 1000 researchers per institution  
is a not unreasonable "back of the envelope figure", but closer  
approximations can be obtained on a per-institution basis by looking  
at the number of "research-active" individuals that were returned in  
the last RAE. These figures range from about 3000 each for Oxford and  
Cambridge, through the Russell-group institutions in the 1000-2000  
region, to a long tail that returned around 100 or less.

This disparity (a factor of 50 between existing repositories at the  
two extremes) makes it hard to compare the effectiveness of  
repositories just by the size of their holdings, or the speed of  
deposits. Instead, we can consider the repository's activity "per  
researcher", as the number of deposits per researcher per year. We  
might imagine that a repository that was working flat out (in the  
steady state) would be ingesting some small number of items from each  
researcher every year, as they completed new papers or finalised new  
data sets or made new collections of scholarly materials.

The exact number will doubtless vary from discipline to discipline  
(physics vs economics) and even from institution to institution  
(depending on the deposit policy). Even so, I would suggest that  
there is an intuitive baseline achievement of 1.0 deposits per  
researcher per year (d/r/y) that represents a repository having  
achieved a widespread acceptance throughout the institution on an  
ongoing basis.

RESULTS
Recomputing the ROAR data on this basis shows that about 1/3 of the  
UK's repositories are already performing well (between 0.5 - 17.0 d/r/ 
y), about 1/3 are gaining acceptance (between 0.01 - 0.25 d/r/y) and  
about 1/3 are still being established (showing 0.00 d/r/y). Given the  
difficulty of institutional change, it is encouraging that a  
substantial fraction of the UK's existing repositories are performing  
so well.
--
Les Carr

CAVEAT EMPTOR
The above figures are attempts to distill a helpful 'state of play'  
from a number of estimates. Here are some of the difficulties in  
interpretation of the data:
(a) the yearly deposit rates were extrapolated from data taken from  
July-August 2006 only. This may not be a representative time of year  
to measure repository activity.
(b) the deposits were 'measured' by the changes in OAI records as  
recorded by the Celestial OAI proxy service. Not every repository had  
a full set of entries in this service.
(c) a new OAI record may not correspond to a genuinely new deposit:  
it may be a new version of an existing deposit or a new set of  
metadata for an existing deposit.
(d) a repository deposit may be an Open Access Self Deposited  
research article, or it may be a photograph of someone's children. No  
account of information purpose is taken by these crude figures.
(e) the RAE figures for returned researchers are 5 years out of date.
(f) The RAE figures were originally compiled for a financial  
calculation; it is hoped that they provide a reasonably accurate  
figure for our purposes.
(g) Not all the repositories listed were full institutional -some  
were departmental. In those cases an estimate of the size was taken,  
based on the size of the institution.
(h) Batch deposits from legacy systems may account for very large  
deposit figures. However, it is believed that no bulk deposits were  
seen in figures for the July-August period.