Pooling and Comparing Noise Annoyance Scores and “High Annoyance” (HA) Responses on the 5-Point and 11-Point Scales: Principles and Practical Advice

Abstract

The use of different noise annoyance scales across studies and socio-acoustic surveys, in particular the popular 5-point verbal and 11-point numerical scales, has made the evaluation, comparison, and pooling of noise annoyance responses among studies a taxing issue. This is particularly the case when “high annoyance” (HA) responses need to be compared and when the original studies used different scales; thus, there are different so-called cutoff points that define the part of the scale that indicates the HA status. This paper provides practical guidance on pooling and comparing the respective annoyance data in both the linear and logistic regression context in a statistically adequate manner. It caters to researchers who want to carry out pooled analyses on annoyance data that have been collected on different scales or need to compare exposure–HA relationships between the 5-point and 11-point scales. The necessary simulation of a cutoff point non-native to an original scale can be achieved with a random assignment approach, which is exemplified in the paper using original response data from a range of recent noise annoyance surveys. A code example in the R language is provided for easy implementation of the pertinent procedures with one’s own survey data. Lastly, the not insignificant limitations of combining and/or comparing responses from different noise annoyance scales are discussed.