In health services research, the investigator often needs to obtain quantitative evaluations of different states of health from raters who are either experiencing or imagining these states. Such evaluations are used in health status indices, screening and monitoring care, clinical decision analysis, clinical trials, cost-utility analysis, or program evaluation (Berzon & Schumaker, 1994). First, descriptions of the relevant health states are prepared (Llewellyn-Thomas, 1996). A health state description is a paragraph consisting of different "attributes" (e.g., the ability to carry out self-care, the ability to carry out usual work activities, the intensity of a symptom like pain). Because each attribute can be represented at different levels, all the possible attribute and level combinations yield a set of different health state descriptions. In the full set, one description will approximate a particular individual's (or rater's) experienced health state, whereas the remaining descriptions represent imaginary health states.Then, for each health state, the investigator elicits an evaluative score representing the rater's judgment of its overall desirability, or relative position on the good health-death continuum. The most common elicitation techniques are the standard gamble (von Neumann & Morgenstern, 1953), the time trade-off (Torrance, 19761, and the linear analogue scale (Hoepfl & Huber, 1978; Llewellyn-Thomas, Sutherland, Ciampi et al., 1984). The theoretical rationale underlying these techniques, their procedural details, and the issues regarding their use are treated in depth elsewhere (Froberg & b e , 1989; Torrance, 1976). The focus of this chapter is on investigators who are interested primarily in the response shift