Geoscience Reference
In-Depth Information
expected result ('confi rmation bias'); and the tendency to over- or underestimate
some uncertainties (over-/under-confi dence: see Kahneman et al., 1982 for an
overview).
Much research in the 1970s and 1980s focused on the psychological aspects of
uncertainty and risk. This research aimed to: (i) improve methods of eliciting opin-
ions about uncertainty and risk; (ii) provide a basis for understanding and anticipat-
ing public responses to various hazards; and (iii) improve the communication of
risk information between lay people, technical experts and policymakers (Slovic
et al., 1981). The so-called 'psychometric' approach, built on advances in cognitive
and social psychology, became popular in the 1970s (Slovic et al., 1974). Data were
typically gathered by questionnaire, and statistical relationships were developed
between various aspects of personality and risk perception. Perhaps the best-known
outcome of this work is the scatter diagram created by Paul Slovic and colleagues
whereby lay attitudes towards a variety of hazards are plotted on two axes, labelled
'unknown' risk, and 'dread' risk. The resulting scatter often indicates a preference
for stricter controls of less familiar, more frightening risks. Public perceptions were
thus conceived as exaggerating the unknown when accident statistics reveal that
many familiar risks are incrementally more costly (Adams, 1995).
The psychometric approach clarifi ed some important psychological controls on
risk perception, such as voluntariness, familiarity, the nature of the hazard, and the
types of people exposed to or benefi ting from risk (e.g., children versus adults).
However, it was also premised on an assumption that 'gaps' between expert and
public understandings of uncertainty and risk are caused by erroneous public per-
ceptions of 'true' risks. More recent social and political analyses have questioned
this long-standing dichotomy, suggesting instead that all perceptions of risk, whether
expert or lay, represent partial and selective views of events (Wynne, 1996).
Attitudes towards uncertainty and risk are socially as well as psychologically
constructed. Language is crucial in translating general concepts, such as 'eutrophica-
tion' (nutrient enrichment of water bodies) into particular entities, such as 'algal
blooms', and then into measurable quantities, such as 'chlorophyll-a' (Richards
et al., 1997). Theories may perform badly against observations if this translation is
ambiguous
or
if the theories or observations are inadequate
or
if the criteria used
to compare them (the 'demarcation criteria') are inadequate (Brown, 2004). Trust
in the sources of information and the processes through which knowledge is gained
also affect confi dence in research outcomes. In terms of the former, MacKenzie
(1990) describes a 'certainty trough' whereby those nearest to and most alienated
from scientifi c research will harbour the greatest uncertainties about its outcomes.
MacKenzie argues that scientifi c output will appear most certain at an 'intermediate
distance', where it is suffi ciently close to be valued by its users but suffi ciently remote
to avoid detailed criticism of its methods. Building on this concept, Shackley and
Wynne (1995) suggest that climate impacts modellers are frequently overconfi dent
in integrated assessment models because they are unfamiliar with some of the dis-
ciplinary assumptions made. This may be reinforced by the use of naïve coupling
procedures, whereby disciplinary models are connected through their numerical
inputs and outputs (cf. climate-forcing scenarios) without considering the possibility
of structural feedbacks, such as human responses to climate change (Shackley and
Wynne, 1995).
Several authors have sought to emphasise the cultural aspects of uncertainty and
risk. 'Cultural Theory' was developed by Mary Douglas and colleagues in the 1970s
