Scaling reward value with demand curves versus preference tests

Between-good preference tests are a popular method for creating an ordinal scale of the value of different goods. However, such measures are influenced by how these goods interact with each other. As an alternative, I propose creating a value scale based on demand-curve elasticity. I tested both methods with monkeys by creating demand curves for three rewards, and then giving them pair-wise choices between the same rewards. In contrast to value assignments based on preference, a demand-curve approach showed little value difference between foods. The choice data in this report were compared to preference data with these subjects from another study, and showed between-study lability in preference. Given that demand analysis test-retest reliability is high while that of preference tests is not, I argue that demand analysis is the superior technique for scaling reinforcer value.