The contingent valuation method (CVM) is widely used in evaluations of proposals for expenditures on public goods and in natural resource damage assessment. Two frequently observed phenomena, however, pose problems for the use of contingent valuation methods in welfare analysis: the differential between willingness to pay (WTP) and willingness to accept
(WTA) measures of value, and subadditivity in willingness to pay. Two competing explanations of consumer behavior are used to explain these phenomena, the demand--theoretic model and the psychological model. The key observation motivating the proposed study is that the psychological explanation implies subadditivity for both WTA and WTP, whereas the demand-theoretic explanation implies superadditivity for WTA. This permits a test of the competing hypotheses. Empirical tests of these hypotheses are proposed, the first part using the Knetsch and the second, the Kahneman-Knetsch approach.