The effect of attribute ranges on weights in multiattribute utility measurements

Multiattribute utility theory requires a specific relation between the range of outcomes of each attribute and the weight for that attribute. The greater the range, the greater the weight has to be. Experimental results show that subjects do not adjust their judgements properly if the range is varied. For the two methods tested the adjustment is smaller than required by theory. The bias was smaller for a regression procedure than for the direct ratio method. Weights based on an intuitive range were not found to be superior to those elicited over different ranges.