In response: do REF cycles really encourage 'poorer quality research'?

I think there was an effect of the 4 outputs per academic to be submitable was significant. As the census date came looming then the pressure to have those 4 outputs was extremely strong. This leads to an undoubted attempt to get more things published and thus submitted in the last year before census. As found in my research: https://kclpure.kcl.ac.uk/portal/en/publications/an-analysis-of-the-arts-and-humanities-submitted-research-outputs-to-the-ref2014-with-a-focus-on-academic-books(9cfc5250-07e0-4d82-9b0b-0853447024e6).html Page 26: "In the final year before the census date for the REF2014 an average of over 27% of books were published. The trend shown in Figure 10 is clear with more books published year on year from the relative low of 2008 through to the peak year of 2013. This effect is not confined just to books; with the average across all research output types being ~25% in the final year and increasing year on year from ~10% in the first year. Thus, there is a clear effect of the hard deadline of REF on when books are published. The back loading over the 5 year period with a last year rush will have significant effects upon capacity: for publishers, editors, peer review and academics alike." I believe that the REF2021 rules with an average of 2.5 outputs per academic will address the problem of rushing out research outputs that may not be as high quality or as polished as possible. The emphasis will be on quality not quantity and this will be a stronger indicator of the research strength of the Unit of Assessment.

Papers early in the cycle will have the luxury of being submitted initially to higher ranked journals even if they are not likely to be accepted; some will randomly be selected at a higher rank than warranted. This is the first thing that needs modeling - the choice of outlet. The second need is for a model of citation. If citations are naturally biased towards higher ranked journals, the more cautious behaviour of submission in the later part of the cycle will bias cites downward.

Regardless of the merits of the paper under discussion it would be very hard to deny that there has been ever-increasing pressure on people to publish a lot, regardless of whether they have anything to say. I can see no way that this pressure can increase quality. The REF (TEF etc etc) are just yet another statistically-illiterate way of ensuring that research is short term and corners are cut.

In the immortal words of Mandy Rice Davis ‘he would say that, wouldn’t he?’