Data Structure and Process for the Pilot Return

The data required for the pilot return was structured around three tables; a staff table, a table of research outputs and a table to resolve the many-to-many relationships between staff and outputs (generally referred to as the 'link' table).

Full details of the structure of the pilot return can be found at the HEFCE website (REF: bibliometrics pilot briefing 9 July 2008) which also gives a detailed data dictionary for the return.

The pilot exercise was structured to provide a maximum of information and it is unlikely that the final definition of the REF will require all the information that was included in the pilot. However, as the final composition of the return is not yet decided, this report comments on all aspects of the data collection.

There are several different models of collecting data for bibliometric analysis including top down, bottom up and selective analysis. From the institutional perspective the structure of the pilot was on the 'top down' model where outputs from individual researchers within the institution are identified and attributed to their institution. Consequently this report concentrates on this aspect of analysis though it should be noted that the data collected could be used for other types of analysis.

From the institutional perspective the pilot exercise involved two major events:

• The construction of the initial return

• Verification that additional output records found by the analysis of Evidence Ltd were valid and correct

The case study visits were all carried out after the submission of the initial return but before the verification of the additional records. Therefore the experience of the verification stage is not included in this report.

In order to minimise the load on participating institutions for the bibliometrics pilot all the relevant definitions that had been used in the RAE2008 were carried forward. A full definition is available on the RAE website. In the final specification for the REF all definitions will be subject to change and once the specification has been published this web site will be updated to show new and amended definitions.