Conclusions

The amount of IT resource that any institution needs to expend on meeting the REF will be influenced by several factors:

• The final specification of the REF return and any relationship it has with other statutory reporting requirements such as Higher Education Statistics Agency (HESA)

• The size and complexity of the institution

• The amount of research that is carried out at the institution

• The degree to which the institution wishes to monitor its research outputs on an ongoing basis

• The frequency of REF exercises

All institutions used data from their HR systems to supply the staff record data. The areas of difficulty were common across all five of the case study institutions. It is very likely that the provision of staff details to the final REF specification will not be problematic, although additional data may be needed to be tracked and this could be achieved by relatively simple local systems or by HR system suppliers making minor amendments to their products.

In order to maintain an accurate up-to-date record of the research outputs being generated by an institution, some institutional department (possibly the institution's Research Office) will need to have in place either a publications database or an institutional repository and robust procedures for capturing all of the research outputs of the institution in an accurate and timely manner. Ideally the catalogue will be an IT system and will be continuously updated. Institutions that choose not to invest time and money into a publication database/repository are likely to find providing an accurate REF return difficult and very time consuming.

Institutions routinely use third party commercial sources such as WoS, CrossRef and SCOPUS as either primary sources of information on publications or to confirm and expand information sourced locally.

The difficulties of tracking research outputs without a central resource are likely to increase as the number of researchers in an institution increases. On the basis of this study it is not possible to be definitive about the point at which producing an accurate return would become impossible, but within the case studies even institutions with relatively small amounts of research activity have voiced the opinion that they will require a central database of publications.

Institutions constructing a link table manually found this process both extremely time consuming and error prone. The existence of a data warehouse bringing together staff and output data is highly desirable and institutions will need to make a cost benefit judgement on whether their research is of a volume and complexity that justifies the time and expenditure that putting such a system in place would entail.

Whatever the intentions of HEFCE, it seems inevitable that institutions will put in place systems to monitor their performance in terms of the REF and will use these indicators as part of their internal management, monitoring and assessment processes.