Biomedical Engineering Reference
In-Depth Information
Similarly, Kahneman and Varey ( 1990 ) posit that high probability events are
perceived as more proximal than low probability events, and thus act as a direct
influence upon perceived risk.
10.11
De-biasing Top-down Strategies
Based on the above literature, base rates that are constructed using socially and
geographically proximate populations and framed in terms of being closer in time
are more likely to be attended to than those based on more socially distant, far popu-
lations. Beyond reconstructing base rates, one can also reframe base rates to make
them more likely to be attended to. In fact, when risks are low, and cannot be proba-
bilistically made higher, this may be the only managerially actionable route to
explore.
Reframing base rates : Halpern et al. ( 1989 ) assert that very few people understand
the properties of numbers, and, therefore, it is important to understand whether
consumers process this information accurately, whether its format and frame affect
their judgments, and whether it is assimilated into their risk estimates. Specifically,
Halpern et al. ( 1989 ) showed that respondents ignore the format in which numerical
information is provided and make judgments based on the absolute magnitudes of
the number provided. The frame “100 % greater” was perceived to mean “twice” as
large, and “200 % greater” was perceived to mean “twice” as large as well! Halpern
et al. ( 1989 ) also showed that the unit of measurement of magnitude information
was ignored: “4.15 times greater” was perceived to be the same as “415 % times
greater” rather than the appropriate “315 % times greater.” In an incomprehensible
twist attesting to people's difficulty in processing numbers, Halpern et al. ( 1989 )
went on to demonstrate that even though “4.15 times” and “415 %” are judged to
be equivalent, 415 % is perceived to be a greater risk of death than 4.15 times
because it has a higher magnitude! Consistent with these results, Raghubir and
Menon ( 1996 ) show that counter-biasing base rate information is more effective
when the frequency is worded as an actual number vs. a percent. Their results sug-
gest that respondents are more likely to infer “1 out of 5” as being lower in magni-
tude, and, therefore, more effective at reducing the overreport of socially desirable
behaviors than “20 %.” Even simply reframing a risk as 1/100 is more effective at
persuading people that they have a higher risk than framing the same risk as
10/1,000 (Raghubir 2008 ).
10.12
Overall Model of Risk
While there are multiple ways in which the two routes, bottom-up and top-down,
can be combined, one route is to incorporate the base rate as the constant ( α ) of the
bottom-up model. This is the proposed model set forth in Fig. 10.1 .
Search WWH ::




Custom Search