Risk Assessment: The Consumer as an Intuitive Statistician - Innovation and Marketing in the Pharmaceutical Industry - page 298

”

PerceivedBaseRate = nN / where 0 < k < 1

When κ = 1, perceptions of base rates are in line with reality, and as κ → 0, there
is a systematic bias in underestimating the base rate. The larger the population
(denominator N ), the more difficult it is to identify with those at risk (numerator n ),
and the smaller the κ ( → 0). The larger the overall number of people who are at risk
(numerator n ), the easier it is to imagine oneself being a part of this group, and the
larger the κ ( → 1). The higher the actual incidence or as n → N , κ → 1 and percep-
tions of base rates are unbiased.
The question then is, what are the factors that determine the sources of bias in
base rates estimation and application (i.e., that can explain the deviation of κ from 1)?
These are explored next.

10.10 Sources of Bias in Base Rate Use

One of the most common alternatives to which base rate information has been
compared has been individuating information—that is, information about a single
person who performs a given action, or has a problem or disease. The social psy-
chology literature has documented that providing individuating information is more
effective than providing base rate information in a variety of judgment tasks (e.g.,
Bar- Hillel 1980 ). Objectively, base rate information is more reliable than individu-
ating information as it is based on a larger sample, but information about an indi-
vidual can be more persuasive as it is more vivid (Bar-Hillel 1980 ). Researchers
interested in improving the accuracy of judgments have studied ways of increasing
the utilization of base rate information. They have found that increasing the rele-
vance of base rate information increases the likelihood of its use in general (Tversky
and Kahneman 1982 ). In line with this, Ginossar and Trope ( 1980 ) argue that when
subjects see base rate information as relevant to the judgment they are required to
make, they do take it into account, with information used as a function of its per-
ceived diagnosticity for the task at hand. Ginossar and Trope ( 1980 ) reversed the
base rate effect by showing that subjects ignored individuating information in favor
of base rate information when the former was less task-relevant.
In this section we propose that overall population base rates are less meaningful
to a consumer than are base rates constructed from populations that they can relate
to, that is, are close to the consumer for a range of reasons, including their genetic
disposition, family history, and experience. This is because these populations are
easier to bring to mind, allowing the respondents to be able to imagine that they
personally will be harmed (Rothman and Kiviniemi 1999 ; Kahneman and Tversky
1982 ). Being able to imagine that the base rate does apply to them specifically
should cognitively lead to perception of higher risk and translate into higher inten-
tions to engage in preventive behaviors.
Kahneman and Varey ( 1990 ) originally demonstrated that proximity correlates
positively with people's perceptions of the probability of an event occurring, and
thus acts as a direct influence upon perceived risk. There are four different dimensions