Because studies of human response to noise are based on samples it is virtually certain that a study's results never exactly describe the group or population from which the sample is drawn. To increase confidence in a study's accuracy various inferential statistics are calculated to indicate the likely precision of descriptive statistical estimators (mean, correlation, regression slope etc). Community noise study reports have not often included the relevant inferential statistics. This is partly because most inductive statistics assume a simple random sample of the population when in fact most noise surveys are based on samples which are clustered around noise measurement points. Such samples require different statistical techniques, ones illustrated in this paper. The paper is the second in a series which examine the probability sample design of the National Railway Noise Survey of Great Britain.

Proc Int Conf Noise Control Eng Inter-Noise '78, Des for Noise Control
Sprott School of Business

Fields, James M. (James M.), & Tomberlin, T.J. (1978). NOISE SURVEY DESIGN AND THE PRECISION OF STATISTICAL RESULTS: FURTHER EVALUATION OF THE DESIGN OF A NATIONAL RAILWAY NOISE SURVEY. Proceedings - International Conference on Noise Control Engineering, 597–600.