Psychometric tests and questionnaires are self report tools. They often require that subjects express their psychological responses to semantic items o statements by numbers. The existence of this kind of tools is justified by the long tradition of research on the stimulus-response paradigm, which is typical of the psychophysics. Category rating, which is very often used in this kind of tools, is sometimes subjected to the so called response biases. They negatively interact with the possibility of conducting a valid and reliable measure of the trait we are interested in. The present paper illustrates the most frequent response biases encountered in psychometric research.