Abstract
One of the more desirable avenues of improvement in spatial interaction modeling is to increase the precision in the specification and measurement of the attractiveness component. This article examines two promising improvements in the attractiveness component, the use of multiple attractiveness measures and the estmation of alternative-specific effects for attractiveness measures. An empirical example involving the purchase of a major consumer durable good is employed to assess the increases in predictive accuracy and behavioral detail attributable to the proposed model extensions.