Risk Assessment: The Consumer as an Intuitive Statistician - Innovation and Marketing in the Pharmaceutical Industry - page 292

appetite changes, tiredness, feelings of guilt or worthlessness, and restlessness or
slowed activity) along all six dimensions. It is a behavior that is present or absent
vs. a feeling state that exists to some extent (increasing its threshold of detectability
and actual expected consistency, and reducing its measurement error), has a high
predictive ability for depression (increasing its causal clarity), with extreme conse-
quences and a low frequency of occurrence in the population, leading to its being
perceived to be more diagnostic of depression than the other symptoms (Raghubir
and Menon 2005 ).
The six characteristics of signals suggest four underlying characteristics of
symptoms that would affect their likelihood of being identified and integrated: their
ambiguity, their frequency of occurrence, their correlation with a disease, and their
extremity of consequence.
Given that the bottom-up framework implies that there are two aspects to the use
of an input: its identification (or accessibility) and its perceived diagnosticity, it is
possible that these four characteristics can affect both. Feldman and Lynch ( 1988 )
proposed the accessibility-diagnosticity model as a useful framework to think of
how people construct judgments in general. Its basic tenets are that people use an
input (in this case a specific behavior or symptom) to the extent it is accessible (in
this case identified or ψ ) and diagnostic ( β ) of the judgment at hand and as an inverse
function of the accessibility and diagnosticity of alternate inputs that can be used to
make the same judgment. We now examine how each characteristic of a symptom
or behavior can affect its likelihood of identification or perceived diagnosticity.
(a) Ambiguity : Symptoms that can be identified as occurring or not occurring (such
as specific behaviors) have a higher threshold of detectability than those that
can exist to some extent (such as states or feelings). Thus, in the domain of
behaviors, the threshold of detectability may be a nonissue, but, in the domain
of symptoms, identification errors may be due to symptom ambiguity.
Ambiguous symptoms need to be interpreted prior to being identified (e.g.,
“feeling tired” in the context of depression, Raghubir and Menon 2005 ). In one
of the earliest papers to examine how symptoms are used to constructs judg-
ments, Raghubir and Menon ( 2005 ) suggested that consumers differentially
interpret ambiguous symptoms in the context of depression as a function of
whether or not they believe they are at risk. Specifically, if respondents were
asked to estimate their level of risk prior to identifying whether or not they had
the nine symptoms associated with depression, they were less likely to identify
that they had any of the symptoms that were ambiguous. In fact, seven of the
nine symptoms (with the exception of the “thoughts of suicide and death” and
“feelings of guilt”) were less likely to be identified as symptoms that a person
possessed if that person was asked to identify the symptoms after completing
their estimate of risk vs. before completing their estimate of risk. The extent to
which a behavior or symptom can be reinterpreted increases its associated mea-
surement error. The more extreme and less ambiguous a behavior or state, the
less likely it can be reinterpreted as being something else. Unambiguous
symptoms and most behaviors (e.g., “thought of suicide or death” in the context