By Jillian O’Rourke Stuart
Imagine you have a severe skin rash and are faced with choosing one of two possible treatments:
Treatment A has a 75% success rate.
Treatment B has a success rate between 60%-90%.
Which treatment would you choose? You might recognize that both treatments have a 75% success rate on average, but the probabilities provided for Treatment B are what decision scientists call ambiguous. You also may have noticed that no explanation as to why this range of probabilities exists is provided.
Ambiguity in a Health Context
In a health decision context, there are a variety of situations in which important probability information can’t be known with precision, such as informing people about contagion risk, side effect risk, or diagnosis confidence. This is important because in general, people dislike ambiguity.
A well-established phenomenon in the judgment and decision making literature, ambiguity aversion, refers to the fact that people tend to prefer options with precisely stated probabilities as opposed to more ambiguous ranges despite equivalent underlying probabilities. Much of this research was established using hypothetical games of chance (e.g., gambles with ambiguous versus unambiguous odds), but how ambiguity affects health relevant decision making has also been explored more recently, with less clear-cut answers. Although people do display increased worry and risk perceptions in reaction to ambiguity in a medical context, few studies have explored how ambiguity affects treatment choice when faced with a scenario like the one laid out above. A few studies that have taken this approach (like this one and this one) have found that participants show more interest in treatments described with ambiguous probabilities.
One under-studied element of these decisions, particularly in a health decision context, is how explanations for the ambiguity alter willingness to adopt that treatment. In other words, why is the success rate for Treatment B not precisely known and how does that explanation affect interest in Treatment B? In a recent paper, my colleagues and I explored this question and found that the content of that explanation can make an important difference in how people respond to different treatment options.
Testing Different Explanations for Ambiguity
Using the skin rash scenario described in the opening, we conducted three studies. Specifically, participants were asked to choose between two skin creams to treat a severe skin rash. As depicted here, the underlying probabilities for the two treatments were always equivalent—even though ranges were used to describe probabilities for one of the treatments. Crucially, different explanations (or attributions) for these ranges were introduced to test how they affected willingness to adopt the ambiguous treatment.
We tested both impersonal and person-specific explanations for the presence of the ambiguous range – all based on real, plausible explanations for ambiguity in a health treatment context. Impersonal explanations are those that refer to elements that are external to the individual or otherwise outside of an individual’s control like different clinical trials producing different results. Person-specific explanations are those that involve something the individual has personal knowledge of or control over such as the strength of one’s immune system or medication adherence levels.
Impersonal Explanations
The first study tested two different impersonal explanations to see if providing any sort of explanation might increase interest in an ambiguous treatment option. In one condition, the ambiguity was attributed to different treatment-effectiveness studies producing different results. In the second condition, the range was described as varying due to truly random circumstances, like the original gambling research (i.e., the active ingredient in the cream becoming inert at different rates depending on the shipment). The majority of participants chose the unambiguous treatment, displaying classic ambiguity aversion, regardless of the explanation provided.
Impersonal versus Person-Specific Explanations
The second study pitted an impersonal explanation against a person-specific one. We retained the second impersonal condition described above (i.e., that the active ingredient became inert at different rates depending on the shipment) and added a person-specific condition, (i.e., that the range of success rates was due to the person’s overall health status). In this person-specific condition the range of success rates is attributed to the individual’s own characteristics rather than external forces. Not only are people more likely to have some control or knowledge over these characteristics but, given that people generally have a positive view of their health, they may be prone to be overoptimistic when these individual characteristics are involved. This time, the different explanations resulted in different levels of interest in the ambiguous treatment. Participants in the impersonal condition again displayed typical ambiguity aversion, but those in the person-specific condition were more willing to adopt the ambiguous treatment.
Different Person-Specific Explanations
Our third study explored three different person-specific explanations for ambiguity to learn more about what elements of an attribution about ambiguity tend to trigger interest in ambiguous options. Specifically, the reason for the range was described as varying depending on how regularly a person applies the cream, how strong the person’s immune system is, or a person’s specific genetic makeup. These conditions vary in how much control and knowledge a person might have regarding the outcome, with the most control over and knowledge about how regularly the cream is applied and the least control/knowledge over a person’s genetic makeup.
These person-specific explanations did affect treatment choices differently. Specifically, participants who thought the success varied based on how regularly the cream was applied or that it varied based on the person’s immune system were more likely to select the ambiguous treatment. However, participants in the genetic factors condition showed no preference for one treatment over the other. This indicates that attributing ambiguity to a self-relevant process is not necessarily enough to push people toward ambiguity seeking overall. Only some self-relevant explanations will have this type of effect.
Implications
Given that ambiguity is somewhat common in a health decision context, it is important to consider the explanations health care workers provide for that ambiguity (if any). Patients should be ablet to select the treatment that is best for them without systematic biases. As we demonstrated here, this understudied element of health-treatment decisions can have consequences when it comes to treatment choice.
