Despite nearly a century of research, we are still unable to accurately predict discomfort due to glare.
The vast majority of discomfort glare research uses subjective evaluations – primarily category rating or luminance adjustment. This article discusses methodological issues which contribute to uncertainty in evaluations.
Evaluations of discomfort using category rating were compared against involuntary physiological responses.
Subjective evaluation of discomfort was highly correlated with eye movement and pupil constriction: Severe glare discomfort increased the speed of eye movement and caused larger pupil constriction.
A series of experiments were conducted to explore the influence of variations in experimental design, the aim being to explain the variance between different studies.
In this first experiment we explored the influence of anchor bias in luminance adjustments.
When Hopkinson used his multiple criterion scale, test participants used adjustment to set the luminance representing a defined degree of discomfort. One limitation was that the four degrees of discomfort were evaluated only in ascending order.
In this experiment we repeated Hopkinson’s method but also using descending and randomised orders of discomfort. The different orders led to significant differences in luminance for a given degree of discomfort.
When evaluating discomfort the test participant must look somewhere, and for evaluating discomfort from peripheral sources a fixation task is used. This experiment compared the use of a simple fixation mark and a task requiring a degree of cognitive attention. The results confirmed a lower degree of discomfort is expressed when engaged in a task demanding a higher degree of cognitive attention.
Stimulus range bias is probably the greatest problem within subjective evaluations of magnitude. Here we demonstrate its influence on evaluations of discomfort from glare using adjustment. The conclusion, that range bias is significant, is not unexpected but direct demonstration helps to support the message.
When you ask test participants to evaluate a series of items which vary in magnitude, it has been proposed that first the range is demonstrated. In other words, before any evaluations, show participants the smallest and largest magnitudes they are about to evaluate. This helps anchor the response range to the magnitudes. This experiment was conducted to determine whether such demonstration would affect the evaluations.
Kent M, Fotios S. The effect of a pre-trial range demonstration on subjective evaluations using category rating of discomfort due to glare. Leukos. Online first, 23/07/2019. doi.org/10.1080/15502724.2019.1631177
Jan Wienold (EPFL) and several colleagues recently published a paper in which they tested several possible metrics for discomfort from daylight glare using a several independent sets of data. In terms of a robust analysis, this was outstanding work. In this item of correspondence I raised a few of questions. One question asked about fairness, given that the proposers of the two best-performing models are also authors of the analysis. A further question asked about stimulus range bias: the authors had suggested it was not a problem but that is unlikely to be the case.
Further articles have been submitted which investigate the influence on discomfort evaluations of showing a pre-trial demonstration of the stimulus range, and of the bias induced by stimulus range itself.