Modeling community noise tolerance with non-Gaussian distributions

Previously, a hierarchical (multi-level) model was shown to predict exposure-annoyance curves in excellent qualitative agreement with noise survey data aggregated from multiple communities D. K. Wilson et al. [J. Acoust. Soc. Am. 142(5), 2905–2918 (2017)]. The hierarchical model includes variations in exposure and annoyance occurring at the individual and community levels. The community-level annoyance variations in the hierarchical model are conceptually similar to the community tolerance level (CTL) concept proposed by Fidell et al. [J. Acoust. Soc. Am. 130(2), 791–806 (2011)]. Wilson et al. modeled the exposure and tolerance variations in the hierarchical model using Gaussian distributions. However, recent research indicates that noise exposure can have significantly non-Gaussian behavior; in particular, noise-level distributions tend to be “heavy-tailed,” indicating that loud events occur more often than would be predicted by a Gaussian distribution. Similarly, human tolerance to noise could have significant non-Gaussian behavior. This presentation examines the impact on the exposure-annoyance curves of extending the previous hierarchical model to non-Gaussian distributions, specifically to the family of Lévy-alpha stable distributions.