Background: The use of a simpler procedure for the measurement of utilities could affect primarily the variance or both the mean and the variance of measurements. In the former case, simpler methods would be useful for population studies of preferences; however, in the latter, their use for such studies might be problematic.
Purpose: The purpose of this study was to compare the results of utility elicitation using single-item questions to computer elicitation using the Ping-Pong search procedure.
Methods: In a convenience sample of 149 primary care patients with symptoms of depression, the authors measured and compared standard gamble (SG) utilities elicited using a single-item “open question” to SG elicitations performed using a computerized interview procedure. Elicitations were performed 1 to 2 weeks apart to minimize memory effects.
Results: More than 90% of persons with utilities of 1.0 to the single-item standard gamble had utilities of less than 1.0 on the computer SG instrument. Consistent with this finding, the mean utilities were lower in computer interviews (0.80 vs. 0.90; P < 0.0001 for differences). Within subjects, utility measures had only a fair degree of correlation (r= 0.54).
Conclusions: Use of single-item questions to elicit utilities resulted in less precise estimates of utilities that were upwardly biased relative to those elicited using a more complex search procedure.
Reprinted with permission from Medical Decision Making, Vol. 21, No. 2, March-April 2001, pp. 97-104. Copyright © 2001 Society for Medical Decision Making.