The effects of rating scales and individual characteristics on perceived annoyance in laboratory listening tests

Abstract

The issue of annoyance caused by noise has received considerable attention. There is evidence that annoyance assessment plays a crucial role in representing adverse sensory of noise. Although studies have provided some guidance for conducting annoyance assessment experiments, researchers have yet to systematically investigate the effect of rating scales, individual characteristics, and number of subjects on annoyance. This paper seeks to remedy such three problems by statistically analyzing annoyance from three listening tests (LT) in laboratory. Denote the scores of M subjects on K sounds as an M*K matrix, calculate the mean of each row to produce the mean subject score (MSS) of each subject, and similarly, calculate the mean of each matrix column to return the mean noise annoyance (MNA) of each noise sample. The first study found that the MNA and MSS calculated by 0 ∼ 100 linear scale (LT1) and 0 ∼ 10 category scale (LT2) both followed the Gaussian normal distribution, and the correlation between MNAs was 0.99. Statistical test results showed that there was no difference in MNAs and MSSs between the two listening tests, indicating that the rating scaling experiment results are independent of the type of the scale. Then, paired samples t-tests were used to assess whether individual characteristics (experimental environment/site, gender, occupation, age level, experimental experience, use/exposure to the noise source, and knowledge of acoustic) result in differences between MNAs. The test results revealed that the age and the familiarity of the target noise can lead to differences in the perceived annoyance of the subjects. Finally, in order to estimate the effect of the number of subjects on MNA, MNAs from various sample sizes were compared to those from a group of 88 subjects (linear scale) and a group of 104 subjects (category scale), respectively. The findings showed that the correlation coefficient between the MNA of 30 subjects and the MNA of 88 (or 104) individuals was greater than 0.98, and the maximum gap between MNAs is no more than 15 points on the 100-point linear scale (the gap was <1 point on the 9-level category scale). The conclusions presented here may facilitate application in laboratory sound quality assessment experiments.