THE UNDERLYING DISTRIBUTION OF LIGHTNESS MATCHES
The normality assumption for lightness matches was tested. A large number of observers were run on the well-known simultaneous lightness contrast illusion and the distribution of the obtained data was analysed. The lightness matches were gathered using the Munsell scale. The distribution testing was done both for the reflectance data and data transformed into log reflectance. The tests showed supporting evidence in the case of reflectance data. They could be treated as normally and double-exponentially (Laplace) distributed. However no such symmetrical distribution can be assumed for the log transformed reflectance values. Results strongly suggest that the use of standard statistical methods should be limited only to reflectance values. Some alternative methods are suitable in case of Laplace distribution and can be used as well.