The national academy for the humanities and social sciences has called for a 10 per cent reduction in the research excellence framework's impact weighting, cautioning that the current figure is too high and could "damage" the pursuit of research.
In its submission to the REF consultation, due to close on 16 December, the British Academy calls for the impact weighting to be reduced from 25 to 15 per cent.
The other elements of the framework - "outputs" and "environment" - should be boosted by 5 per cent to 65 and 15 per cent respectively, it adds.
"We strongly advise that (25 per cent) is too high, given the untried nature of the impact-assessment process," it says. "Moreover, the costs of getting it wrong are substantial, as a flawed system will impede and damage research."
It also says that while it is appropriate to reduce the sub-panels for the humanities and social sciences, some of the proposed configurations are "forced and arbitrary".
"The brigading of disciplines purely for the sake of administrative convenience has resulted in some proposed sub-panel configurations that are unjustifiable," it says.
It argues that it would be more appropriate to reduce the 67 sub-panels in the research assessment exercise 2008 to 35 rather than the 30 currently proposed. The submission mirrors a flurry of activity among humanities and social sciences subject groups as they prepare their own submissions.
Jackie Eales, professor of history and American studies at Canterbury Christ Church University and a co-convenor of History (UK), which represents 80 history departments, said impact should count for 10-15 per cent, with outputs valued at 70 per cent.