Estimating utility functions in the presence of response error

Article Abstract:

Exploration is presented of the nature and degree of response error present when direct, multi-attribute utility assessment procedures are employed as a basis for modeling preferences for risky, multi-attribute options. Analysis is based on an experimental examination of preferences for optional air pollution control policies whose consequences were typified by three value attributes: level of pollution-associated illness, cost to consumers, and polluted-related mortality. Results suggest that direct assessments of outcome preferences were stable and reliable over a two-week period, as were parameter estimates for additive utility functions. Statistically fitted additive utility models resulted in strongly accurate predictions of directly assessed preferences two weeks earlier or later. Ranking outcomes before assigning utilities to them caused high degrees of serial correlation of errors.

Methods, Risk assessment, Error analysis (Mathematics), Response surfaces (Statistics)

---

Preferences for proxy attributes

Article Abstract:

A study of preferences for pollution control alternatives is used to test the hypothesis that assessing utility functions over proxy attributes demands complex inferences that may exceed the human ability to make consistent judgments. Subjects demonstrated a nearly universal tendency to overweight the proxy attribute relative to the prescriptions of anticipated utility theory. The large bias resulted in a major loss of expected utility in a simulated policy formation scenario. Three heuristic models were developed to account for this bias: best guess, worst case, and relative importance models. Results strongly support the relative importance model, in which decision makers assess scaling constants by depending on general opinions on the relative importance of various decision objectives rather than on well-articulated preferences for substitution rates between attribute pairs.

Management, Pollution, Control

---

Risk preferences for gains and losses in multiple objective decision making

Article Abstract:

The attitude toward risk concept acts in a central role for both behavioral and prescriptive decision theory. The article examines risk preferences for decisions with consequences that can be considered in at least two relevant attributes. The reference risk-value (RRV) model, which is basic in format though it has the ability to represent the types of multiattribute reference effects observed in research conducted by Payne, Laughhunn, and Crum, is examined in the article.

Management science, Distribution (Probability theory), Competing risks

Similar abstracts:

• Abstracts: Choice of market entry mode in China: the influence of firm-specific factors. Early steps to becoming a multinational
• Abstracts: Mini miners off the radar. Coal producers see boom beyond the gloom. The lost miners
• Abstracts: Quest for the best. The office savior. The price of practice
• Abstracts: Some Models for Estimating Technical and scale Inefficiencies in Data Envelopment Analysis. Response to 'Decision Problems under Risk and Change Constrained Programming; Dilemmas in the Transition'
• Abstracts: John Arme elected NAA president. Profile: NAA President Ernie Huband. Tom O'Toole, NAA president, 1989-90

This website is not affiliated with document authors or copyright owners. This page is provided for informational purposes only. Unintentional errors are possible.