Three methods for calibrating noise annoyance ratings were tested in a
group of 206 persons from different work places. One was Master scali
ng, which involves calibration for differences in mean ratings as well
as the slope of the psychophysical function for seven levels of a pin
k noise reference sound. The other two methods calibrated only for dif
ferences in mean ratings of reference sounds. The reference sounds wer
e either the pink noises or verbal descriptions of three everyday soun
ds. Calibration for differences in mean ratings leads to a sixfold inc
rease of the variance explained by sound level. Calibrating for differ
ences in slope did not lead to any further improvement. The described
sounds worked only slightly worse than the recorded sounds as referenc
e sounds. (C) 1997 Academic Press Limited.