INFLUENCE OF RANGE BIAS ON GLARE EVALUATION IN EXPERIMENT

Abstract

The performance of daylighting as the cause of glare is important for the residents' comfort. Today evaluation item on daylighting of interior environment were adopted by international certification system such as LEED or WELL. In order to disseminate these systems in Japan, it is required to use them with evaluation items in line with Japan's situation such as climate, density of city and residents' value. Therefore, it is necessary to continue to study the effect of daylighting on residents' subjective response. In such studies, experiments are often conducted in the laboratory. It is common in laboratory experiment to set several physical conditions and get subjects evaluate them. Previous studies suggests that the range of stimulus affects the result of evaluation. For example, the borderline between comfort and discomfort depend on the maximum luminance presented to the subjects. In this study, the effect of range width on psychological evaluation is named as range bias, and the tendency of the effect of range bias during glare evaluation experiment was examined.

The experiment was conducted in a laboratory room with a pseudo window. We prepared three sets of setting, each of which have four luminance of the pseudo window. The minimum luminance of each set is the same as 2000 cd/m², but the maximum luminance vary as 8000, 14000 and 20000 cd/m². Subjects evaluated each set at intervals of about one week. The order of the three sets and the four luminance levels within each set was completely randomized for each subject. 12 subjects evaluated the environment by the usual GSV scale. They also evaluated the degree of glaring and discomfort with stepless scale like GSV scale. Additionally subjects evaluated the impression of the room with SD scale and the acceptability as an office.

Data by the stepless scales and the SD scales was analyzed by two-way ANOVA with the presented 12 levels of luminance and subjects as factors. As a result, the main effect of luminance was observed on all scales but one. There was no significant difference among the three maximum luminance, and it was confirmed that range bias existed as the effect that subjects judge the maximum luminance in comparison to other luminance presented in one experiment, rather than the absolute judgement. On the other hand, there was a difference in the evaluation tendency between the case where the contents of the scale directly evaluated the physical conditions such as brightness and the case where the atmosphere of the room was evaluated. When directly evaluating physical conditions, the tendency of range bias may be small. Finally, the data of acceptability was analyzed by logistic regression analysis for each set. There was a big difference between the set with the highest maximum brightness and the other two sets. It is also suggested that there is a threshold at which range bias occurs.