Reproducibility and repeatibility: Why acoustics consultants, manufacturers, and academics should care

Abstract

Trust in acoustic data from all sources is very low. The resulting crisis of confidence in end users, architects, contractors, and designers leads to overly conservative design standards or a sense that “nothing matters so why bother doing anything" that might improve the acoustic performance. While this is partly because buildings are complex systems with many interrelated variables affecting the results, part of it is due to normal measurement variation. Study of measurement uncertainty and experimental design began in the late nineteenth and early twentieth centuries, and the fruits of statistical analysis can be found in the repeatability and reproducibility measurements that populate the ASTM standard Precision and Bias statements. Repeatibility and reproducibility can be used for three purposes: 1. to evaluate the quality of the standard, 2. to adjudicate when two testers’ measurements disagree, and 3. to evaluate edge cases for compliance with specified values. Unfortunately, the reproducibility and repeatability data in the E33 Building and Environmental Acoustics standards are not typically statistically rigorous enough to provide useful information. By supporting and participating in round robins and interlaboratory studies, the uncertainty in our measurements can be quantified, and we as a community can be confident in our measurement data.