Center for Research in Energy Systems Transformation (CREST)

Replicating scientific findings is a fundamental aspect of research. However, in studies of discomfort due to glare, it is difficult to make comparisons between the results of different experiments since the statistical tests usually reported do not allow independent findings to be directly compared to each other. Here we present an alternative Bayesian approach that can address this problem. To show how this approach works, we performed a laboratory test with 55 participants to validate the effect of order bias previously detected in a similar study evaluating discomfort due to glare but, this time, under a large luminous source. Using the luminanceadjustment procedure, the glare source was varied to meet four sensations of discomfort due to glare. Adjustments were performed under three different order sequences: ascending, descending, and randomised. Test participants provided glare settings using a newly proposed evaluation scale. The effect of order bias detected in the original study was compared to the data obtained with the same methodological procedure in the new experiment using Bayesian inferential tests. The results showed a close replication, highlighting that the order bias effect found in the original study was also present in the new experiment. The wide application of Bayesian methods in the design and analysis of experimental studies may improve the accuracy and validity of glare models.