Thurstone’s law of comparative judgment applied to 5-point verbal annoyance responses from community noise studies

Studies on community annoyance with aircraft noise often use a 5-point verbal scale in which the categories are assumed to be equally spaced (ISO/TS 15666). Responses are dichotomized into highly annoyed (4 and 5) or not highly annoyed (1, 2, and 3) responses, and generalized linear models are used to find relationships between the noise dose and annoyance response. A Thurstone approach avoids dichotomization of responses and interprets each response as a comparison between the selected verbal response and the boundaries of the 5-point verbal categories. From this, standard implementations of the Law of Comparative Judgment yield a latent annoyance scale. This approach gives a finite value to the percent highly annoyed for each response, even for values less than 4, which may be a modeling advantage for datasets with a low prevalence of annoyance. Thurstone analysis is also used to evaluate the assumption of equally spaced categories. Applying the Law of Comparative Judgment to 5-point verbal annoyance responses may be advantageous because it considers the psychological difference between perception (an internal, private judgment) and the external perceptual response.