There has been much debate about university research assessment exercises. In the UK, a major element of the 2014 Research Excellence Framework (REF 2014) has been the research environment. Here we analyse 98 REF 2014 ‘environment’ submissions in Business and Management Studies.We explore whether there are distinctive language-related differences between submissions of high and low ranked universities and conclude that submission writers have a strong incentive to exaggerate strengths and conceal problems. In addition, innate biases such as the ‘halo’ and ‘velcro’ effects may distract the attention of assessors from a submission’s strengths and weaknesses, since they are likely to influence their pre-existing impressions. We propose several changes to improve how environment is evaluated. We also argue that the research environment would be more likely to be enhanced if the number of outputs submitted in future was an average of two and a maximum of four per academic, rather than the maximum of five currently being considered.