Interpreting the ERA results
Some points to remember when interpreting the ERA 2012 results.
Sensitive to size and focus of the institution
Much of the ERA evaluation looks at averages across FoR groups. As such, a large field containing a hub of research excellence can therefore be diluted by a tail of average research activity.
Similarly a very strong sub-field (say, a 6-digit FoR group) can be diluted by weaker sub-fields (other 6-digit FoR groups) contributing to the one field (the 4-digit FoR group). The same situation can arise between 4-digit FoR groups contributing to one 2-digit FoR group.
The result of this is that a field or even an institution that is large and comprehensive risks having its hubs of excellence diluted to the point where they’re no longer apparent.
FoR codes do not equal faculties
Every single faculty contributed to multiple 2-digit FoR groups and every single 2-digit FoR group was contributed to by multiple faculties.
FoR codes do therefore not correspond to faculties or schools, even if there appears to be a clear correlation (e.g. Faculty of Pharmacy with FoR 1115 Pharmacology and pharmaceutical sciences).
It is therefore impossible to say that a result in an FOR code ‘belongs’ to a faculty or school.
There is no scope for interdisciplinarity to be captured or evaluated in ERA. If an item was assigned multiple FoR codes (e.g. a paper on a chemotherapy drug may be assigned an oncology FoR code and a medicinal chemistry FoR code) it was evaluated completely separately as half an item in each code.
ERA is backward looking
ERA 2012 evaluated research published as early as 2005, meaning that the research was actually carried out ten years or even longer ago, due to the inevitable delay in publishing the research.
Further, since the ERA census date was 31 March 20011, any appointments made after this time were completely excluded from ERA.
It should therefore be remembered that ERA gives us an indication of our past research performance, but not our current, and certainly not our future research performance.