Skip to main content icon/video/no-internet

Procedural Invariance and Its Violations

Procedural invariance states that preferences over prospects (i.e., gambles, or any other risky states that can be described as a probability p of getting outcome/payoff x) are independent of the method used to elicit them. In other words, procedural invariance, an important pillar of rational choice, demands that strategically equivalent methods of elicitation will give rise to the same preference order. Satisfaction of procedural invariance is implied by the orderability axiom of von Neumann-Morgenstern utility theory. In medical decision making, procedural invariance is the condition that a person's preference ranking of two health states should not depend on the elicitation procedure. For example, suppose one uses both the visual analog scale and the standard gamble to assign quality weights to two health states. Procedural invariance requires that the visual analog scale preference ranking of these two health states be the same as the standard gamble ranking. If the rankings are different, then it would be impossible to infer a single unique ranking, and, therefore, to assign a unique quality weight to each health state. Note that procedural invariance is different from description invariance, which states that preferences over prospects are purely a function of the probability distributions of consequences implied by the prospects and do not depend on how those given distributions are described. While procedural invariance is an assumption implicit in any conventional theory of choice, which seems natural to most economists and decision theorists and is rarely even discussed when stating formal theories of choice, in practice, this assumption fails.

Economic Violations of Procedural Invariance

One well-known phenomenon, often interpreted as a failure of procedure invariance, is the preference reversal. Reversals of preferences are observed when a so-called $-bet (offering a high money prize with low probability) is assigned a higher selling price than a P-bet (offering a lower money prize but with a higher probability) but is subsequently not chosen in a direct choice between the two. A violation of procedure invariance is currently the prevailing explanation of this pattern of behavior. For instance, when assessing the monetary values of gambles (or delayed payments), people base their actions on a particular value system. When choosing between gambles, however, they base their actions on another value system. Therefore, preferences from choosing are different from those yielded by monetary valuation; this fact constitutes a violation of procedure invariance. The explanations invoking a violation of procedure invariance are mostly based on weighted additive models. These models assume that, both for choosing and for valuation, a person evaluates a gamble-delayed payment ($x, p), where x is the expected monetary reward and p is the probability of obtaining this payoff, by using av(x) + bw(p), where v and w are values for the separate attributes and a > 0 and b > 0 sum to 1 and are importance weights. It is generally assumed that the value functions v and w are the same for choosing and valuation and that only the weights a and b vary. That is, for choosing, particular weights, such as ac and bc, are adopted, and for valuation, other weights, such as av and bv, are adopted. Note that because importance weights are positive, choosing and matching yield the same orderings over single attributes.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Sage Recommends

We found other relevant content for you on other Sage platforms.

Loading