- Both are free
- Fedora is a repository engine and while richer requires extensive configuration and the development of front ends, or buying into the VTLS pre-rolled solution
- Dspace can be built on a spare server in an afternoon (Yes, it can, I've done it) giving you an instant repository never mind about the management issues
So we can say that ROAR is under reporting usage, especially as the Fedora site breaks down the listings to 71 universities and 16 National or public libraries, both of whom could be presumed to be interested in Open Access.
So while we can say Fedora is less popular than Dspace we cannot say how much less popular. Especially given the availability of Fez as an open source alternative to VTLS's commercially licensed VITAL front end suite.
Given the ease of running up a Dspace instance as opposed to a Fedora instance one might be tempted to argue that institutions using Fedora have a greater commitment to their repository given the greater effort taken to deploy it (though Fez may make this less true).
What I can't come up with though is a sensible metric for this. Average number of items might be a suitable crude number as if one platform was little used once deployed, ie the hosting institution was not committed to populating their repository you would expect the numbers of items to be low in that repository. Unfortunately this could be completely wrong - if there were a large number of small institutions using one platform to the exclusion of the other - ie small institutions use dspace due to its lower resourcing demands - to create small but well populated repositories (after all 20 researchers, even if highly productive will produce less material than 200 researchers, or so you would hope).
Answers on a postcard ...