Increasing interest in the role of sensorimotor information in lexical-semantic processing has translated into a recent proliferation of psycholinguistic ratings (e.g. perceptual strength, body-object interaction - BOI). These norms are often used for stimulus selection and control in experiments to ensure methodological validity. Along with megastudies, their increasing availability has also paved the way for large-scale and high-powered regression analyses which considerably advance our understanding of how words are processed. Despite their importance, however, the ratings in themselves have been subject to surprisingly little methodological consideration. Using BOI norms collected by our team, we first illustrate the largely overlooked midscale disagreement problem in psycholinguistic ratings which results from computing averages on Likert-scale responses. Through examples from the published literature, we then derive three major implications which likely affect the reliability of a great number of studies. We conclude by offering some best-practices for the use and interpretation of the ratings, and open further avenues for research on the subject.