Application of statistical analysis of sample size: How many occupant responses are required for an indoor environmental quality (IEQ) field study

Determining required sample size is one of the critical pathways to reproducible, reliable and robust results in human-related studies. This paper aims to answer a fundamental but often overlooked question: what sample size is required in surveys of occupant responses to indoor environmental quality (IEQ). The statistical models are introduced in order to promote determining required sample size for various types of data analysis methods commonly used in IEQ field studies. The Monte Carlo simulations are performed to verify the statistical methods and to illustrate the impact of sample size on the study accuracy and reliability. Several examples are presented to illustrate how to determine the value of the parameters in the statistical models based on previous similar research or existing databases. The required sample size including "worst " and "optimal " cases in each condition is obtained by this method and references. It is indicated that 385 is a "worst case " sample size to be adequate for a subgroup analysis, while if the researcher has an estimate of the study design and outcome, the "optimal case " sample size can potentially be reduced. When the required sample size is not achievable, the uncertainty in the result can properly interpret via a confidence interval. It is hoped that this paper would fill in the gap between statistical analysis of sample size and IEQ field research, and it can provide a useful reference for researchers when planning their own studies.