Method: We surveyed a random sample of 2,137 Norwegians aged 40-67. Respondents allocated a given amount of resources between two patient groups (rare versus common disease). Diseases descriptions were identical in terms of severity and expected treatment benefits, but the rare disease was more costly to treat. The allocation decision was accomplished by moving a pointer along a horizontal bar marked with different combinations of numbers of rare and common disease patients. Respondents were randomized to 7 survey versions, in which either the opportunity cost of treating the rare disease, the mechanism for making the allocation decision, or the amount or type of other information provided were varied, allowing us to test for price effects and three different framing effects. Outcome measures in each case were the mean share of funds devoted to treating rare disease patients and the percent of respondents recommending an equal division of resources between rare and common disease patients.
Result: Framing had significant effects on outcomes. Particularly important was the finding of a visual central tendency bias associated with using a horizontal bar for the allocation decision; 40% of respondents divided resources equally using the horizontal bar compared to 13% using the alternate selection mechanism (p = 0.05). Outcome variables were also sensitive to “labeling” and to an extra task of selecting the allocation principle respondents felt health authorities should apply in setting priorities. Evidence of responsiveness to opportunity costs was less clear. The share of resources allocated to the rare disease declined as the opportunity cost of the rare disease increased, as economic theory predicts, but the extent of the change was minimal, possibly indicating “scope insensitivity.”
Conclusion: Preferences elicited using simple trade-off techniques may be sensitive to a variety of framing effects. The small response to varying opportunity cost may be indication of scope insensitivity, calling into question the degree to which results were responsive to theoretically relevant economic variables.