Power analysis, sample size, and assessment of statistical assumptions—improving the evidential value of lighting research

The reporting of accurate and appropriate conclusions is an essential aspect of scientific research, and failure in this endeavor can threaten the progress of cumulative knowledge. This is highlighted by the current reproducibility crisis, and this crisis disproportionately affects fields that use behavioral research methods, as in much lighting research. A sample of general and topic-specific lighting research papers was reviewed for information about sample sizes and statistical reporting. This highlighted that lighting research is generally underpowered and, given median sample sizes, is unlikely to be able to reveal small effects. Lighting research most commonly uses parametric statistical tests, but assessment of test assumptions is rarely carried out. This risks the inappropriate use of statistical tests, potentially leading to type I and type II errors. Lighting research papers also rarely report measures of effect size, and this can hamper cumulative science and power analyses required to determine appropriate sample sizes for future research studies. Addressing the issues raised in this article related to sample sizes, statistical test assumptions, and reporting of effect sizes can improve the evidential value of lighting research.