Sei sulla pagina 1di 12

Risk Analysis, Vol. 23, No.

5, 2003

Exploring the Dimensionality of Trust in Risk Regulation

Wouter Poortinga1 ∗ and Nick F. Pidgeon1

This article investigates possible differential levels of trust in government regulation across
five different risk contexts and the relationship between a number of concepts that might
be thought of as comprising distinctive “dimensions” of trust. It appeared that how people
perceive government and its policies toward risk regulation was surprisingly similar for each
of the five risk cases. A principal-component analysis showed that the various trust items could
best be described by two dimensions: a general trust dimension, which was concerned with a
wide range of trust-relevant aspects, such as competence, care, fairness, and openness, and a
scepticism component that reflects a sceptical view regarding how risk policies are brought
about and enacted. Again, the results were surprisingly similar across the five risk cases, as the
same solution was found in each of the different samples. It was also examined whether value
similarity has an additional value in predicting trust in risk regulation, compared to the more
conventional aspects of trust. Based on the two independent trust factors that were found in
this study, a typology of trust is proposed that ranges from full trust to a deep type of distrust.
It is argued that for a functioning society it could well be more suitable to have critical but
involved citizens in many situations.

KEY WORDS: Risk perception; risk regulation; trust

1. INTRODUCTION and acceptance of risks. This issue was first raised by


Wynne (1980), who argued that with technological
Trust has become a popular research subject in
risks some of the differences between “expert” and
the social sciences during the last two decades. Trust
“lay” perspectives could be traced to differing evalu-
is considered to lubricate social interactions on vari-
ations of the trustworthiness of risk managing institu-
ous levels so that these function smoothly and harmo-
tions. The relationship between trust and risk percep-
niously (e.g., Tyler & Degoey, 1996), it is thought to
tion has since gained widespread attention (e.g., Renn
reduce social uncertainty and complexity (Luhmann,
& Levine, 1991; Pidgeon et al., 1992; Slovic, 1993). Fur-
1979; Barber, 1983; Earle & Cvetkovich, 1995), and is
thermore, trust is seen as one prerequisite for effec-
seen to be an important element of social capital and
tive risk communication (e.g., Kasperson et al., 1992),
as a prerequisite for a healthy and flexible economy
while distrust may be associated with stigmatization
and democracy (e.g., Putnam, 1993; Fukuyama, 1995;
of technologies, such as nuclear power (Flynn et al.,
Dekker, 1999; Kasperson et al., 1999; Cook, 2000).
2001) as well as social amplification effects following
Also, in the field of risk research there is now
major failures of risk regulation (see Freudenburg,
general agreement that trust in risk management in-
2003; Kasperson et al., 2003).
stitutions may be an important factor in perception

1 Centre for Environmental Risk, University of East Anglia, 1.1. The Dimensionality of Trust
Norwich, NR4 7TJ, UK.
∗ Address correspondence to Wouter Poortinga, Centre for Envi- Although it is now widely recognized that trust in
ronmental Risk, University of East Anglia, Norwich, NR4 7TJ, institutions plays an important role in risk perceptions
UK; w.poortinga@uea.ac.uk. and responses to risk communication, there have been

961 0272-4332/03/1200-0961$22.00/1 
C 2003 Society for Risk Analysis
962 Poortinga and Pidgeon

many debates regarding what constitute and what of trust in risk regulatory or other institutions. For ex-
contribute to trust. In other words, what factors make ample, in their review of the literature on trust, Renn
people trust or distrust risk regulatory or other insti- and Levine (1991) identify five core components or
tutions. In very general terms, Rousseau et al. (1998) attributes, namely: perceived competence, which rep-
argue that trust, as conceptualized across a number of resents the degree of technical expertise of the source;
disciplines, can be defined as: “a psychological state objectivity, reflecting the absence of bias in informa-
comprising the intention to accept vulnerability based tion; fairness, or the degree to which the source takes
upon the behavior of positive expectations of the in- into account all relevant points of view; consistency, or
tentions of or behavior of another” (p. 395). However, the predictability of arguments and behavior based on
this definition does not, in itself, explain why people past experience and previous communication efforts;
might be willing to accept vulnerability. Classical work and faith, which reflects the perception of “good will”
on interpersonal trust suggests that it is mainly a two- of the source. Similarly, Kasperson et al. (1992) iden-
dimensional concept based on competence and care tify four key dimensions that play an important role
(or “trustworthiness”). Half a century ago, Hovland in the development and maintenance of trust: com-
et al. (1953) identified these two aspects in their semi- mitment, as social trust involves some degree of vul-
nal social psychological research program on commu- nerability, one wants to make sure that the trustee is
nication and persuasion. In a series of experiments, in fully committed to the mission, goal, or fiduciary obli-
which they varied specific characteristic of the com- gation; competence, since trust can only exist when
municator, they found that someone accepts informa- a person or institution is competent in the thing it is
tion more easily when the communicator is seen as an obliged to do, so that someone should not only be
expert (i.e., is a good source for valid assertions) and committed to his or her fiduciary responsibilities, but
when the communicator is seen as being trustworthy, should also fulfil it competently; caring, a perception
in the sense that the source is seen as willing to com- that an institution acts in a way that shows concern
municate the assertions he or she considers most valid for the people who put their trust in it; and finally
(i.e., has no motives to promote a particular view or predictability, in that people tend to trust people or
has lack of intent to persuade). However, more recent organizations that are consistent. Predictability of ar-
discussions suggest that there may be various “levels” guments and behavior means that people know what
of social trust (e.g., Renn & Levine, 1991; Kasperson to expect from a particular person or organization.
et al., 1992; Greenberg & Williams, 1999) such as “in- These and other scholars (e.g., Mishra, 1996) have
terpersonal” versus “institutional/social,” or “local” distinguished a wide range of seemingly different as-
versus “global.” Accordingly, more nuanced theoret- pects of trust. All of these aspects have some face
ical and practical understandings of the notion of trust validity, i.e., they all seem to reflect important com-
have begun to emerge, recognizing that it may often ponents of trust under some circumstances. However,
be considered to be complex and multidimensional, it is not always completely clear whether the public
as illustrated in the following quote: distinguishes between the different theoretical fea-
tures of trust. For example, the categories objectivity
“We trust you” may mean that we believe you can give and fairness of Renn and Levine (1991) may not be
us right answers and reliable information. It may mean
completely independent. A source that takes into ac-
that we believe that you are honest, and will tell us
all that you know. Or it may mean that we trust your count all relevant points of view may be perceived as
judgement, and rely on you for decisions which are wise, more objective. Likewise, some people may not dis-
impartial, ethical and in the public interest. We may criminate between the aspects of trust that Kasperson
trust you in one of these ways, without trusting you in et al. (1992) call commitment and caring. Perceptions
the others. In this case, if a pollster asks us whether we
of whether an institution acts in a caring way may in-
trust you, what are we to say? (House of Lords, 2000,
paragraph 2.29) fluence perceptions of whether that institution fulfils
its fiduciary obligations.
Within risk research, a wide range of theoretical Recently, Metlay (1999) has criticized researchers
(e.g., Renn & Levine, 1991; Kasperson et al., 1992; for making discussions about trust unnecessarily dif-
Johnson, 1999) and empirical studies (e.g., Frewer ficult. He argues that some researchers have the ten-
et al., 1996; Peters et al., 1997; Metlay, 1999) have dency to distinguish additional shades of meaning in
been conducted to identify the core elements of trust. the concept of trust, although it is not at all clear
That is, they have examined what kind of evaluative whether these are indeed (empirically) discernible as-
judgments contribute to the creation or destruction pects of trust. Metlay’s study of judgments of trust
Exploring the Dimensionality of Trust in Risk Regulation 963

in the U.S. Department of Energy (U.S. DoE) sug- terpret, but seems to encompass a vested interest or
gests that trust is not complex and multifaceted, but a accountability factor. Using a selection of the charac-
rather simple concept based on two distinctively dif- teristics of Frewer et al. (1996), French et al. (2002)
ferent components: (1) a tightly interconnected set identified similar dimensions of trust. A principal-
of affective beliefs about institutional behavior, which components analysis (PCA) showed that the eval-
Metlay calls “trustworthiness;” and (2) perceptions of uation of a (hypothetical) food-risk communication
how competent the institution is. There is other em- could be described by two dimensions, which related
pirical evidence to support this claim that, rather than to the perceived credibility of the information pro-
being based on a large number of components, trust is vided and the reputations of these organizations. Like
mainly a two-dimensional concept based on compe- Frewer et al. (1996), the first component was a general
tence and care (or “trustworthiness”). As mentioned trust dimension that encompassed competence as well
earlier, Hovland et al. (1953) identified these two as- as care. Although the second factor also included the
pects in their early work on trust. Also Jungermann aspect of “accountability,” it was more closely related
et al. (1996) found, in a study about communicating to what Barber (1983) calls fiduciary responsibility, as
the risks of hazardous chemical facilities, that trust it also covered “responsibility,” and “proven wrong in
in information sources could be described by a two- the past.”
factor solution, rather similar to that reported by Met- In conclusion, there would seem to be two slightly
lay, representing honesty and competence. differing empirical models of the dimensionality of
Evidence for trust as a two-dimensional construct trust. The first model, arising from the work of
has been mainly found using factor-analytical statis- Metlay (1999), consists of the dimensions of a general
tical techniques. A criticism of this approach is that trustworthiness factor alongside one of competence.
respondents are typically asked to use scales that The second model, from the work of Frewer et al.
are predetermined by the researcher, rather than (1996), overlaps partially with that of Metlay in that
characteristics that are selected by the respondents it also points to a general trustworthiness dimension
themselves. As a consequence, the resulting factor (including both competence and care) together with
structure may not correctly represent the public’s di- a separate vested interest dimension. A primary ob-
mensions of trust. Frewer et al. (1996) have tried to jective of the present article then is to test which of
avoid this criticism by personalizing their question- these two models best describes trust in government
naires on trust in relation to food-related hazards. In risk regulation on a range of issues.
a series of studies, they constructed a personalized
questionnaire by first asking respondents to indicate
1.2. Salient Value Similarity
their own reasons for trusting or distrusting a num-
ber of possible sources of food risk information. In Earle and Cvetkovich (1995) have argued that
this way, respondents could rate each of the infor- for most people it is far too demanding to base trust
mation sources on the dimensions they themselves on evidence of competence and fiduciary responsibil-
thought were important. Frewer et al. (1996) found a ities. According to Earle and Cvetkovich (1995), so-
two-factor structure that could best describe the rea- cial trust is particularly critical where complex socio-
sons for trusting or distrusting various information technical systems generate risks that are remote from
sources, later validated with a more representative everyday experience. However, most people will not
population sample. The first component comprised have the resources or interest to make a detailed as-
the characteristics: truthful, good track record, trust- sessment of whether or not it is worthwhile to trust
worthy, favor, accurate, factual, public welfare, re- a particular institution. Earle and Cvetkovich (1995)
sponsible, knowledgeable, and (negatively) the char- argue that it is more likely that under complex cir-
acteristics: distorted, proven wrong in the past, and cumstances trust is based on agreement and sympathy
biased. The second factor consisted of the character- rather than on carefully reasoned arguments or direct
istics: accountable, self-protection, and a vested inter- knowledge (see also Langford, 2002). In other words,
est versus sensationalism component. While Frewer people base their trust judgments on whether they feel
et al. (1996) did not themselves label the two compo- that the other person or organization shares the same
nents, their first factor seems once again to reflect a values, i.e., is seen as having the same understand-
general trust evaluation of an information source, in- ing of a specific situation. Although conventional
terestingly encompassing both competence and care trust concepts (such as competence) are often used
components. The second factor is more difficult to in- in discourses on trust, Earle and Cvetkovich (1995)
964 Poortinga and Pidgeon

reason that this is without regard to whether these 2. To empirically investigate the relationship be-
really are the critical factors underlying people’s trust tween a range of concepts that might be
judgments. Rather, people may use these concepts to thought of as comprising distinctive “dimen-
provide an explanation for their general agreement, sions” of trust. In particular, whether people’s
sympathy, or trust in an institution because espoused perception of government’s risk policies can
value similarity may not be regarded as a socially ac- be described by a limited number of under-
ceptable response. This salient value similarity (SVS) lying dimensions. The design provides a test
approach of Earle and Cvetkovich (1995) has been ap- of the two empirical models, noted above, of
plied successfully in a number of studies (e.g., Earle Metlay (1999) and Frewer et al. (1996). That is,
& Cvetkovich, 1995, 1997, 1999; Siegrist et al., 2000, whether the evaluation of government’s risk
2001). However, it has not been examined whether policies on the five risk cases could best be
the SVS approach has indeed an “additional” value described by the dimensions of general trust-
in explaining trust in the regulation of risk. In his own worthiness and competence (Metlay, 1999), or
study, Metlay (1999) reports an indirect test of the by the dimensions of general trust and vested
value-similarity thesis, for which he found little evi- interests (Frewer et al., 1996).
dence. In the current study we provide a more direct 3. To enhance Metlay’s original set of survey
test of this possibility. items in order to permit a more direct exami-
nation of whether the SVS approach of Earle
and Cvetkovich (1995) has an additional value
2. AIMS OF THE STUDY
in predicting trust in risk regulation over and
This study reports data from one part of a major above the more conventional conceptualiza-
quantitative survey conducted in Great Britain dur- tions noted above.
ing 2002, which had a main focus of making a detailed
comparative analysis of five risk cases on various risk- 3. THE SURVEY
relevant topics. This study examined people’s percep-
3.1. Procedure and Respondents
tions of the British government and its policy with
regard to several risk cases, namely, climate change, Data for this study were collected in the sum-
radiation from mobile phones, radioactive waste, ge- mer of 2002 (see Poortinga and Pidgeon, 2003).
netically modified food (GM food), and human ge- The quantitative survey was administered in Britain
netic testing on a wide range of trust-related aspects. (England, Scotland, and Wales) by the market re-
These cases were chosen because they are all promi- search company MORI. Fully trained and supervised
nent societal risk issues, and have complementary as market research interviewers carried out face-to-face
well as contrasting facets of risk, benefits, and un- interviews with people in their own homes, which
certainty. They all involve scientific knowledge, pub- lasted on average about 30 minutes. The survey was
lic trust in science, and risk regulation, allowing for run in 126 sample points. Interviewers approached se-
tests of governmental competence in this area. In ad- lected addresses within these sample points until they
dition they are all themes that are covered by vari- reached the quotas for gender, age, and work status.
ous surveys of public opinion, but where a richer set The quotas were reflective of the actual profile in each
of contextually referenced and comparative data is of the sampling points. The total sample of 1,547 peo-
lacking. ple aged 15 years and older comprised of five sep-
The current study had a number of discrete arate quota samples of about 300 respondents, each
objectives. covering one of the five risk cases (see Table I). So,
respondents were presented with a set of standard-
1. To investigate possible differential levels of ized questions on only one of the five risk issues listed
trust in government regulation across the dif- above.
ferent risk contexts. This will be done in re-
lation to several attributes that are generally 3.2. Measures
seen to constitute trust (see Section 3.2). Our
3.2.1. Evaluation of Government
expectation is that, given the rather different
nature and histories of the five risk issues cho- Respondents were asked to evaluate the British
sen, significant cross-issue differences in pat- government on how they thought it handles the
terns of trust ratings will emerge. five risk cases. People were presented with various
Exploring the Dimensionality of Trust in Risk Regulation 965

standardized statements on the government and its tion. Where the MANOVAs were statistically signifi-
policy on one of the five risk cases, depending on the cant (i.e., with respect to differences between the five
version of the questionnaire. The statements were se- risk cases) Tukey’s multiple pair-wise procedure for
lected from a review of previous work on trust (e.g., post hoc comparisons was used to establish which of
Renn & Levine, 1991; Frewer et al., 1996; Peters et al., the specific risk case means differed from the others.
1997; Johnson, 1999; Metlay, 1999). Eleven items were It appeared that the government and its policies
designed to measure the trust-related aspects of com- were evaluated differently across the five risk cases
petence, credibility, reliability, integrity (vested inter- (F(44, 4714) = 1.78, p < 0.001). However, what is
ests), care, fairness, and openness (see Table I). The immediately striking about Table I is how few real
scores could vary from 1: “totally disagree” to 5: differences there are in people’s evaluation of gov-
“totally agree.” ernment with respect to the different risk cases. The
mean scores, while they differ between different ques-
tions, only rarely differ within any specific question.
3.2.2. Value Similarity This high level of across-issue agreement is surprising,
Two items were included that were aimed at mea- given that each risk issue was evaluated by a different
suring the extent to which the government was seen (albeit carefully matched) sample of respondents, and
as having the same values with regard to the five risk that each of the cases was chosen because of its par-
cases. These items were adapted from earlier work of ticular contrasting facets and histories. For example,
Earle and Cvetkovich (1995) and Siegrist et al. (2000). human genetic testing is a relatively “new” issue and
People were asked to indicate to what extent they might also be expected to hold higher perceived ben-
agreed with the statements “The government has the efits than would GM food. Indeed, a range of other
same opinion as me about [issue X],” and “The gov- psychometric items included in the survey, but not
ernment has the same ideas as me about [issue X],” reported here (see Poortinga & Pidgeon, 2003), did
where again the issue was the relevant risk case for the confirm that people held different beliefs about the
questionnaire version. Answers could be given on a risks, benefits, and other aspects of the individual five
five-point scale from 1: “totally disagree” to 5: “totally risk issues themselves.
agree.” Overall, the mean results showed a moderate lack
of endorsement of government across the items. Mean
responses were below the scale midpoint for most of
3.2.3. Trust in Risk Regulation the “positive” items such as competence (except the
Trust in risk regulation was measured using two item regarding availability of skilled people), care,
items. People were asked whether they agreed with and fairness, whereas the mean responses were above
the statement “I feel that current rules and regulations the scale midpoint for “negative” items such as influ-
are sufficient to control [issue X]” as well as the state- ence by industry, changes policies, and distorts facts.
ment “I feel confident that the British government Interestingly, the item with the lowest average agree-
adequately regulates [issue X],” where issue X (one ment was the one designed to measure the concept of
of the five risk cases) again depended on the version of openness. In Britain the need for greater openness in
the questionnaire that the respondent received. Re- government risk handling was a key recommendation
sponses to the two items were given on a scale ranging of the official inquiry into the BSE “mad cow” disease
from 1: “totally disagree” to 5: “totally agree.” crisis (Phillips et al., 2000).
The right-hand column of Table I shows the ques-
tionnaire items where the MANOVA reached sta-
4. RESULTS tistical significance, and different subscripts in the
Table indicate where specific means differed from
4.1. Comparing the Five Risk Cases
each other (Tukey’s test p < 0.05). Note though that
Multivariate analyses of variance (MANOVA) the statistical effect sizes for all of these differences
were conducted to examine whether people differed are small (of the order of 0.2 or 0.3).
in their mean responses across the five cases with There were consistent significant differences in
respect to the 11 items evaluating the government, the two perceived value similarity items across the five
whether government was seen as having the same val- risk cases (F(8, 2582) = 3.24, p < 0.001). More people
ues with regard to the five risk cases, and whether agreed with the statement that the government has the
there were overall differences in trust in risk regula- same opinion about radiation from mobile telephones
966 Poortinga and Pidgeon

Table I. Evaluation of Government

Climate Mobile Radioactive GM Genetic


Risk Statement Change Phones Waste Food Testing p-value

Competence
The government is doing a good job with 2.42a 2.52ab 2.66b 2.45ab 2.68b <0.01
regard to . . . (0.97) (0.88) (0.84) (1.00) (0.99)
The government is competent enough to 2.60 2.61 2.59 2.51 2.71 n.s.
deal with . . . (1.08) (1.06) (1.03) (1.16) (1.17)
The government has the necessary skilled 2.99 2.93 3.08 2.85 3.09 n.s.
people to carry out its job with regard to . . . (1.09) (1.01) (1.02) (1.07) (1.15)
Credibility
The government distorts facts in its 3.71 3.49 3.68 3.71 3.50 <0.01
favour regarding . . . (0.93) (0.95) (0.94) (0.90) (1.12)
Reliability
The government changes policies regarding 3.53 3.41 3.41 3.60 3.45 n.s
. . . without good reasons (0.89) (0.83) (0.88) (0.94) (0.99)
Integrity (vested interests)
The government is too influenced by industry 3.73 3.57 3.58 3.65 3.54 n.s.
regarding . . . (0.86) (0.83) (0.87) (0.91) (0.94)
Care
The government is acting in the public 2.68 2.82 2.72 2.62 2.79 n.s.
interest with regard to . . . (1.01) (0.94) (0.95) (1.13) (1.11)
The government listens to concerns about 2.59 2.68 2.67 2.53 2.70 n.s.
. . . raised by the public (0.99) (0.97) (0.96) (1.05) (1.12)
The government listens to what ordinary 2.30 2.41 2.35 2.27 2.38 n.s.
people think about . . . (0.97) (0.96) (0.98) (1.05) (1.05)
Fairness
I feel that the way the government makes 2.53 2.61 2.63 2.45 2.55 n.s.
decisions about . . . is fair (0.91) (0.83) (0.87) (1.01) (1.02)
Openness
The government provides all relevant 2.15 2.19 2.07 2.08 2.05 n.s.
information about . . . to the public (0.92) (0.91) (0.88) (0.95) (1.04)
Value similarity
The government has the same opinion as me 2.46ab 2.63b 2.44ab 2.29a 2.47ab <0.01
about . . . (0.97) (0.90) (0.87) (1.06) (1.00)
The government has the same ideas about . . . 2.33a 2.57b 2.45ab 2.29a 2.43ab <0.01
as me (0.93) (0.92) (0.89) (0.98) (0.97)
Trust in risk regulation
I feel that current rules and regulations in the 2.62a 2.86ab 2.67a 2.65a 2.92b <0.001
U.K. are sufficient to control . . . (1.08) (0.93) (0.99) (1.10) (1.10)
I feel confident that the British government 2.59 2.67 2.74 2.61 2.75 n.s.
adequately regulates . . . (1.11) (0.97) (1.00) (1.05) (1.20)
Sample size (N) 321 319 306 296 305

Note: The scale ranged from 1: “strongly disagree” to 5: “strongly agree;” standard deviations are given in parentheses. Means with different
subscripts are significantly different from each other.

than about GM food, and more people agreed that the scale could be constructed from the two items
government has the same ideas about mobile phones (Cronbach’s α between 0.77 and 0.85 for the five
than about climate change and GM food. However, cases).
the average responses to both items were once more Also, differences in trust in risk regulation were
below the middle of the scale, meaning that people found across the five risk cases (F(8, 2724) = 3.54,
did not see the government as having the same values p < 0.001). Table I shows that overall trust in risk reg-
as themselves with regard to any of the five risk cases. ulation was not high for any of the five risk issues.
A reliability analysis showed that one value-similarity The mean responses to the two items designed to
Exploring the Dimensionality of Trust in Risk Regulation 967

measure this were below the scale midpoint for all allows comparisons between the five risk cases on the
cases. Small differences were found in one of these resulting factors.
items. It appeared that more people felt that cur- Table II shows that the PCAs in each of the five
rent rules and regulations in the United Kingdom are risk cases produced very similar factor solutions. The
sufficient to control genetic testing than to control initial 11 items were described by two main compo-
radioactive waste, climate change, and GM food. The nents, which accounted for between 58.0% and 68.2%
reliability of both items was sufficient in each of the of the original variance. In each of the five risk cases
five risk cases (Cronbach’s α between 0.65 and 0.81), the first component, which accounted for about 40%
so a trust in risk regulation variable was constructed of the variance, comprised the items designed to mea-
combining the two. sure competence, care, fairness, and openness. This
can be interpreted as a general trust component. That
is, it represents a general evaluation of government
4.2. The Dimensionality of Trust
policy on the five risk issues. The second compo-
To examine whether the 11 statements evaluat- nent, which accounted for about 20% of the variance
ing government could be described by a number of of the original items, was concerned with credibility,
underlying dimensions, separate PCAs with varimax reliability, and integrity (vested interest). The items
rotation were conducted for each of the five risk cases. “the government distorts facts in its favor regard-
As the results were similar for all risk cases, one sin- ing . . . ,” “the government changes policies regard-
gle PCA was conducted across all risk cases. This also ing . . . without good reasons,” and “the government

Table II. Factor Loadings After Varimax Rotation

Climate Mobile Radioactive Genetic


Change Phones Waste GM Food Testing Overall

1 2 1 2 1 2 1 2 1 2 1 2

The government is doing a good 0.80 −0.11 0.75 −0.27 0.73 −0.18 0.77 −0.39 0.79 −0.30 0.77 −0.25
job
The government is competent 0.72 −0.26 0.71 −0.27 0.76 −0.16 0.74 −0.42 0.81 −0.29 0.76 −0.27
enough
The government has the necessary 0.57 −0.20 0.47 −0.35 0.70 0.13 0.66 −0.31 0.71 −0.07 0.65 −0.12
skilled people
The government distorts facts in −0.23 0.80 −0.23 0.86 −0.22 0.73 −0.28 0.82 −0.19 0.83 −0.23 0.82
its favor
The government changes policies −0.23 0.81 −0.21 0.88 −0.09 0.86 −0.22 0.86 −0.27 0.83 −0.22 0.84
without good reasons
The government is too influenced −0.11 0.73 −0.36 0.60 −0.24 0.76 −0.35 0.73 −0.19 0.74 −0.24 0.73
by industry
The government is acting in the 0.73 −0.18 0.65 −0.30 0.70 −0.28 0.73 −0.30 0.75 −0.25 0.72 −0.26
public interest
The government listens to 0.72 −0.06 0.72 −0.21 0.71 −0.24 0.77 −0.15 0.79 −0.09 0.75 −0.13
concerns raised by the public
The government listens to what 0.68 −0.16 0.75 −0.25 0.70 −0.28 0.82 −0.26 0.76 −0.26 0.75 −0.24
ordinary people think
I feel that the way the government 0.79 −0.29 0.75 −0.20 0.75 −0.29 0.77 −0.37 0.79 −0.28 0.77 −0.28
makes decisions is fair
The government provides all 0.71 −0.23 0.76 −0.20 0.56 −0.26 0.76 −0.19 0.70 −0.30 0.69 −0.25
relevant information to the
public
Eigenvalue 4.25 2.13 4.14 2.42 4.08 2.29 4.77 2.73 4.83 2.40 4.45 2.33
Explained variance 38.6 19.4 37.6 22.0 37.1 20.8 43.4 24.8 43.9 21.9 40.5 21.2
Average agreement 2.53 3.66 2.60 3.49 2.61 3.55 2.48 3.64 2.65 3.51 2.58 3.57
Cronbach’s α 0.87 0.74 0.88 0.78 0.87 0.75 0.92 0.82 0.92 0.78 0.90 0.77

Note: The scale ranged from 1: “totally disagree” to 5: “totally agree.” Factor loadings higher than 0.40 are in bold. Factor interpretations:
(1) general trust; (2) scepticism.
968 Poortinga and Pidgeon

is too influenced by industry regarding . . .” loaded (top half of Table III). The explained variance of
high on this second component in all five risk cases. trust in risk regulation varied between about 20 and
The latter component found here reflects a scepti- 50%. The two predictors best explained trust in the
cal view on how risk policies are brought about, and regulation of genetic testing, GM food, and radioac-
was labeled as such (scepticism). Once more, this high tive waste, but were less successful in the explana-
level of stability in the factor solution is surprising, tion of trust in the regulation of climate change and
given that each risk issue was evaluated by different the regulation of radiation from mobile phones. In
respondents. Model 2 (bottom half of Table III), where the com-
bined value similarity scale was added to the initial
set of predictors, the explained variance of trust in
4.3. The Explanatory Power of Value Similarity risk regulation only marginally increased for most
In order to examine whether the SVS approach risk cases. Only genetic testing saw a reasonable in-
has an additional value in explaining trust, two sets of crease in explained variance of about 6%. In Model
multiple regression analyses were conducted for the 2, general trust remained the strongest predictor for
five risk cases separately. The first regression model trust in risk regulation across the five cases. If less
(Model 1) included the factors resulting from the over- strong than general trust, scepticism was also a signif-
all factor analysis on the evaluation of government. icant predictor of trust in the regulation of all five
In the second regression model (Model 2), the com- cases. Value similarity appeared to be the weakest
bined value similarity scale was added to this initial predictor of trust in risk regulation, being a strong
set of predictors. The differences in explained vari- (additional) predictor of trust in the regulation of
ance between the two regression models were used as genetic testing, and a less strong but still significant
an indicator for the importance of value similarity to predictor in the context of mobile phones and GM
trust in risk regulation across the five risk cases. An- food. The addition of value similarity to the five re-
other indication that value similarity is an important gression models did not greatly affect the regression
determinant of trust in risk regulation is if its addition coefficients of general trust and scepticism, although
to the regression models leads to decreased regres- it did somewhat decrease the regression coefficients
sion coefficients for the other independent variables. of general trust and scepticism in the contexts of
In either case, value similarity would be a significant radiation from mobile phones, GM food, and
predictor for trust in risk regulation. genetic testing.
Table III shows the results of the analyses for
each of the two models. When trust in the regula-
5. DISCUSSION
tion of the five risk cases were regressed onto the two
components resulting from the overall factor analysis This study has dealt with how people perceive
(Model 1) both the general trust and scepticism com- government and its policies toward risk regulation in
ponents were highly significant in all five risk cases different risk contexts. The first aim was to compare

Climate Mobile Radioactive Genetic


Change Phones Waste GM Food Testing

Model 1
General Trust 0.42∗ 0.38∗ 0.57∗ 0.59∗ 0.59∗
Scepticism −0.16∗∗ −0.24∗ −0.33∗ −0.24∗ −0.39∗
R2 19.0 22.1 38.4 44.2 47.6
Adj. R2 18.4 21.4 37.9 43.6 47.1 Table III. Standardized Regression
Model 2 Coefficients
General Trust 0.43∗ 0.24∗∗ 0.47∗ 0.46∗ 0.43∗
Scepticism −0.18∗∗ −0.18∗∗ −0.31∗ −0.17∗∗ −0.33∗
Value Similarity −0.01 0.22∗∗∗ 0.12 0.20∗∗∗ 0.26∗
R2 19.4 24.0 39.5 45.3 53.6
Adj. R2 18.4 22.9 38.8 44.5 53.0

Note: ∗ p < 0.001; ∗∗ p < 0.01; ∗∗∗ p < 0.05; Dependent variable is trust in risk regulation.
Exploring the Dimensionality of Trust in Risk Regulation 969

evaluations across the five rather different risk cases, to a smaller number of dimensions. Moreover, there
i.e., climate change, radiation from mobile phones, ra- were close similarities between the items used in our
dioactive waste, GM food, and genetic testing. Here, and Metlay’s studies to measure the different aspects
evaluation of government on the various trust-related of trust. Even one of the five risk cases in the cur-
items was surprisingly similar for each of the five rent study (radioactive waste) was similar to the one
risk cases, with only very few specific differences. Al- investigated by Metlay.
though there were some significant differences in per- There are at least two plausible explanations
ceived value similarity between the risk cases, they for the differences in results. The first explanation
were once more small. The average responses of most might reside in cross-cultural differences. It may be
items were somewhat below the midpoint value for that there are key differences between British and
all of the five cases. Likewise, there were only mi- American samples in relation to beliefs about trust.
nor differences in overall trust in risk regulation, with If true, this would also account for the similarities
people having the highest trust in the regulation of of the current findings to the models of Frewer et
genetic testing and the lowest trust in the regulation al. (1996) and French et al. (2002). A second, more
of climate change, radioactive waste, and GM food. plausible and theoretically interesting explanation lo-
Again, the average response was below the middle of cates the differences in the survey populations and
the scale for all risk cases, indicating that, in general, in the evaluated institution(s). Whereas in the cur-
there is a tendency toward distrust in government. rent study the survey was administered to represen-
The second aim of this study was to explore the tative samples of the British public, Metlay (1999)
dimensionality of trust. Our initial expectation had used a selective sample of people interested in the
been that trust in government with regard to the five management of radioactive waste, and who had par-
risk cases would be in accordance with the findings ticipated in a long-running exercise involving at least
of Metlay (1999). That is, the evaluation of govern- one hour per week over the course of one year. More-
ment could best be described by a component that over, Metlay looked at trust in a specific government
represents technical competence, and one that rep- department: i.e., the U.S. DoE. As a result of these
resents a general trustworthiness dimension, encom- two factors Metlay’s respondents may have had more
passing care for the public interest (among other specific and stable attitudes about the management
things). Although a clearly interpretable two-factor of radioactive waste (and in particular the compe-
solution was found for each of the five risk cases, the tence of U.S. DoE in its handling of this issue). By
results were in the event closer to those obtained by contrast, the current study asked people to evalu-
Frewer et al. (1996). The first component in the cur- ate government in general. The minor differences in
rent results, reproduced across five different issues mean scores and factor structure between the five
and samples of respondents, was concerned with a risk cases suggest that people evaluated government
wide range of trust-relevant aspects, such as compe- policy as a whole, rather than specific governmental
tence, care, fairness, and openness. As such, this factor policy or decisions with respect to each of the five
can be interpreted as a general trust dimension, albeit risk cases. So, rather than the evaluation of a specific
one that conflates the competence and care dimen- institution, people may have assessed the wider polit-
sions found by Metlay. The second factor obtained ical and administrative system of risk governance. As
here resembled the vested interest factor of Frewer noted at the start of this article, it is often argued that
et al. (1996). However, in the present study the sec- there may be various “levels” of social trust. It may
ond factor also included the aspects of credibility and well be that the level evoked in any particular study
reliability. This factor seems to reflect a sceptical view will vary as a function of familiarity with the institu-
regarding how risk policies are brought about and en- tion being judged. In particular, the more specific the
acted, and was therefore labeled here as a scepticism evaluated subject, the more likely it is that someone
dimension. Again, the results were surprisingly sim- will have more differentiated views. This could also
ilar across the five risk cases, as the same two-factor help to explain why some recent studies, which take
solution was found in each of the different samples. a more locally grounded approach to issues of pub-
It is an interesting question as to what has caused lic understanding of risk and trust, elicit much more
the differences in results between this study and the complex referents for these concepts, and their rela-
findings of Metlay (1999). In both studies the same sta- tionship with such things as economic dependence,
tistical technique of PCA with varimax rotation was place, identity, and stigma (e.g., Irwin et al., 1996;
used to examine whether the original items reduced Greenberg & Williams, 1999; Williams et al., 1999;
970 Poortinga and Pidgeon

Bush et al., 2002; Bickerstaff & Walker, 2003; aspects of trust, it could also be argued that the vari-
Horlick-Jones et al., 2003). Clearly, there are fur- ables used here to measure general trust and scepti-
ther fruitful research avenues—both quantitative cism (such as the item “the government is acting in the
and qualitative—to be conducted on the relation- public interest”) also have value implications. Siegrist
ship between issues of institutional trust and locally et al. (2000) argue that the SVS approach should not
grounded risk perceptions. It also is worth noting here be seen as a substitute for other models of trust. That
that the operation of affect is seen as an increasingly is, SVS does not exclude the conventional aspects
important part of the way in which lay perceptions of trust if those are considered salient for the per-
of risk issues are constructed (Alakhami & Slovic, taining risk case. From that perspective, the current
1994; Finucane et al., 2000; Langford, 2002; Slovic, study seems to suggest that for some cases (especially
in press). In this sense our current results are rather genetic testing) other value aspects than the ones mea-
more in line with Metlay’s findings than might first sured by the factors general trust and scepticism are
seem the case, in that people’s judgments can be de- important as well. This line of reasoning, however,
scribed with relation to a dominant, predominantly raises questions about the nature and measurement
affective, dimension of trustworthiness, with compe- of value similarity as a separate concept or heuristic
tence loading alongside a range of more clearly af- to come to trust judgments, as well as about the pre-
fective variables. Note that the second factor identi- cise relationship between value similarity and other
fied here, that of scepticism, intuitively has affective trust-relevant aspects. Future theoretical and empiri-
properties too. The current results therefore suggest cal research should address whether value similarity
that there is much further interesting research to be is indeed a (empirical and psychological) discernable
done concerning the way institutions (rather than aspect of trust and how this interacts with other eval-
just risks) are affectively perceived, how this relates uations of (risk regulatory) institutions.2
to an institution’s historical record of competence, Reflecting upon the overall pattern of results,
whether this varies with different levels of familiar- Trettin and Musham (2000) have recently questioned
ity or “closeness” to the institution, and how this then the importance of enhancing trust in institutions. That
feeds through into evaluations of (different levels of) is, the public does not necessarily expect or see trust as
trust. an achievable goal in their relation with institutions.
The third aim of this study was to examine Recent evidence from more qualitative work suggests
whether the value-similarity approach of Earle and that the public does not necessarily consider “trust”
Cvetkovich (1995) has an additional value in explain- to be the most appropriate word to describe their re-
ing trust in risk regulation, compared to more con- lationship with risk management institutions (Wynne
ventional aspects of trust. The current results suggest et al., 1993; Walls et al., submitted). Two decades ago
that value similarity may not contribute universally Barber (1983) had already observed that:
to people’s conceptualizations of risk regulation.
The public is less passively deferential in its relations
Having said that, while value similarity is the least with experts and others in authority and is more likely
important factor in explaining trust in risk regulation, to take an active part in monitoring the fulfillment
it still was a significant predictor in the contexts of of professionals’ claims to absolute trustworthiness.
radiation from mobile phones, GM food, and genetic (Barber, 1983, p. 134)
testing. These three issues are just the risk cases that
According to Barber (1983) (see also O’Neill,
are either highly polarized (GM food) or that are (rel-
2002), the importance of full trust tends to be exag-
atively) new issues that have not fully developed yet
gerated. The public has become more competent and
(radiation from mobile phones and genetic testing).
knowledgeable enough to have “effective” distrust.
That is, it is still unclear whether radiation from mo-
Distrust in this sense is not destructive, but can be
bile phones has adverse health effects, and people
seen as an essential component of political account-
may not yet have a clear idea what to think about
ability in a participatory democracy.
genetic testing, as the consequences of this technol-
Pidgeon et al. (2003) posit that what is frequently
ogy are not yet fully understood. This seems to sug-
called trust or distrust exists along a continuum,
gest that in some contexts, especially in those with
low familiarity, value similarity is indeed of special
importance—a finding that is in line with the SVS ap-
2 Subsidiary analysis using structural equation modeling suggests
that the relationship between value similarity, the “conventional”
proach of Earle and Cvetkovich (1995). Although in aspects of trust, and trust in risk regulation may be more complex
the present study it was examined whether value sim- than our current discussion allows. The results of that analysis will
ilarity has an additional value to more conventional be reported in a subsequent article.
Exploring the Dimensionality of Trust in Risk Regulation 971

aggregate results display both moderate general dis-


trust, alongside moderate scepticism (although indi-
General Trust (Reliance)

vidual respondents might be expected to hold distinct


High Acceptance (Trust) Critical Trust
positions in the space shown in Fig. 1).
The proposed typology of trust raises some inter-
esting questions both for risk policy and the direction
of future trust research. In policy terms decision mak-
Low Distrust Rejection (Cynicism) ers may well be confusing “critical trust” with outright
“distrust” or rejection. Of course, these are not the
same thing. Nor do they necessarily demand similar
Low High policy responses. For example, instead of focusing on
how to increase trust in risk management organiza-
Scepticism
tions, it could be more fruitful to give attention to the
Fig. 1. A typology of trust in government. interaction between institutional structures, agency
behavior, and qualitative properties of perceptions
of trust. That is, what kind of relationship between
ranging from uncritical emotional acceptance to people and risk management institutions is achievable
(downright) rejection. Somewhere between these ex- and desirable? For a functioning society it could well
tremes a healthy type of distrust can be found that be more suitable to have critical but involved citizens
Pidgeon et al. call critical trust. Critical trust can be in many situations.
conceptualized as a practical form of reliance on a per-
son or institution combined with some healthy scep-
ACKNOWLEDGMENTS
ticism. The current study provides quantitative evi-
dence for the existence of this conceptualization of Work reported in this article was partly supported
trust, which, until now, has mainly been identified in by the Program on Understanding Risk, funded by a
qualitative work (see, e.g., Irwin et al., 1996; Poortinga grant of the Leverhulme Trust (RSK990021), and was
et al., in press; Walls et al., 2003). The two trust compo- partly supported by a grant from the ESRC Science
nents that were found in this study show that different in Society Program (L144250037). The authors would
degrees of general trust could coexist with different also like to thank the Leverhulme team, Ian Langford,
degrees of scepticism. Based on these two indepen- and others who have helped us in various ways. Special
dent components, a typology of trust is proposed that thanks goes to Daniel Metlay for a number of helpful
ranges from full trust to a deep type of distrust (see Fig. discussions and his comments on an earlier version of
1). The situation in which someone has high general this article.
trust in and is not sceptical about a certain institution
can be said to be one of trust. That is, one is likely to
accept decisions and communications from this partic- REFERENCES
ular organization. This would compare to what Walls Alakhami, A. S., & Slovic, P. (1994). A psychological study of the in-
et al. (2003) call “uncritical emotional acceptance.” verse relationship between perceived risks and perceived ben-
efit. Risk Analysis, 14, 1085–1096.
However, a high degree of general trust can also co- Barber, B. (1983). The Logic and Limits of Trust, New Brunswick:
exist with a relatively high level of scepticism. This Rutgers University Press.
situation could best be described as critical trust. One Bickerstaff, K., & Walker, G. (2003). The place(s) of matter: Matter
out of place—Public understandings of air pollution. Progress
may be willing to rely on information, but one is in Human Geography, 27, 45–67.
still somewhat sceptical, and thus may still (construc- Bush, J., Moffatt, S., & Dunn, C. E. (2002). Contextualisation of
tively) question the correctness of the received infor- local and global environmental issues in north-east England:
Implications for debates on globalisation and the “risk society.”
mation. The situation in which general trust is low, Local Environment, 7, 119–133.
combined with low scepticism, is labeled as distrust. Cook, K. S. (2000). Trust in Society, New York: Russell Sage.
Although distrust would not be a preferred situation, Dekker, P. (1999). Niveaus van vertrouwen in landen van de
Europese Unie [Levels of trust in countries of the Eu-
it could be contrasted to a more deep type of distrust: ropean Union]. In A. E. Bonner et al. (Eds.), Recente
cynicism, a situation in which one not only has no Ontwikkelingen in Het Marktonderzoek Jaarboek 1999.
trust in a specific institution, but one is also scepti- Haarlem: De Vriescheborg.
Earle, T. C., & Cvetkovich, G. T. (1995). Social Trust. Towards a
cal about its intentions. In the latter situation one is Cosmopolitan Society. London: Praeger.
likely to simply reject everything that comes from a Earle, T. C., & Cvetkovich, G. T. (1997). Culture, cosmopolitanism,
particular organization. As noted above, the current and risk management. Risk Analysis, 17(1), 55–65.
972 Poortinga and Pidgeon

Earle, T. C., & Cvetkovich, G. T. (1999). Social trust and culture in in Organizations: Frontiers of Theory and Research. London:
risk management. In G. T. Cvetkovich & R. E. Löfstedt (Eds.), Sage.
Social Trust and the Management of Risk. London: Earthscan. O’Neill, O. (2002). A Question of Trust. Cambridge: Cambridge
Finucane. M. L., Alhakami, A. S., Slovic, P., & Johnson, S. M. (2000). University Press.
The affect heuristic in the judgement of risks and benefits. Jour- Peters, R. G., Covello, V. T., & McCallum, D. B. (1997). The deter-
nal of Behavioral Decision Making, 13, 1–17. minants of trust and credibility in environmental risk commu-
Flynn, J., Slovic, P., & Kunreuther, H. (2001). Risk Media and nication: An empirical study. Risk Analysis, 17(1), 43–54.
Stigma: Understanding Public Challenges to New Technologies. Phillips, L., Bridgeman, J., & Ferguson-Smith, M. (2000). The
London: Earthscan. Report of the Inquiry into BSE and Variant CJD in the
French, S., Maule, J., Mythen, G., & Wales, C. (2002). Trust and UK. London: The Stationary Office. (Available at http://www.
Risk Communication (Technical Report). Leeds/Manchester: bseinquiry.gov.uk.)
Manchester Business School/Leeds University Business Pidgeon, N. F., Hood, C., Jones, D., Turner, B., & Gibson, R. (1992).
School. Risk perception. In Risk Analysis, Perception and Management:
Freudenburg, W. R. (2003). Institutional failure and the organiza- Report of a Royal Society Study Group (Ch. 5, pp. 89–134).
tional amplification of risks. In N. F. Pidgeon, R. K. Kasper- London: The Royal Society.
son, & P. Slovic (Eds.), The Social Amplification of Risk. Pidgeon, N. F., Walls, J., Weyman, A., & Horlick-Jones, T. (2003).
Cambridge: Cambridge University Press. Perceptions of and Trust in the Health and Safety Executive as
Frewer, L. J., Howard, C., Hedderley, D., & Shepherd, R. (1996). a Risk Regulator. Research Report 2003/100, Norwich: HSE
What determines trust in information about food-related risks? Books.
Underlying psychological constructs. Risk Analysis, 16(4), 473– Poortinga, W., Bickerstaff, K., Langford, I. H., Niewöhner, J., &
485. Pidgeon, N. F. (in press). The British 2001 foot and mouth cri-
Fukuyama, F. (1995). Trust: The Social Virtues and the Creation of sis: A comparative study of public risk perceptions, trust and
Prosperity. New York: Free Press. beliefs about government policy in two communities. Journal
Greenberg, M. R., & Williams, B. (1999). Geographical dimensions of Risk Research.
and correlates of trust. Risk Analysis, 19(2), 159–169. Poortinga, W., & Pidgeon, N. F. (2003). Public Perceptions of Risk,
House of Lords. (2000). Science and Society, 3rd Report. London: Science and Governance. Main Findings of a British Survey
HSMO. (Available at http://www.parliament.the-stationery- on Five Risk Cases (Technical Report). Norwich: Centre for
office.co.uk/pa/ld199900/ldselect/ldsctech/38/3801.htm.) Environmental Risk.
Hovland, C. I., Janis, I. L., & Kelley H. H. (1953). Communication Putnam, R. D. (1993). Making Democracy Work. Civic Traditions
and Persuasion. Psychological Issues of Opinion Change. New in Modern Italy. Princeton: Princeton University Press.
Haven: Yale University Press. Renn, O., & Levine, D. (1991). Credibility and trust in risk commu-
Horlick-Jones, T., Sime, J., & Pidgeon, N. F. (2003). The social dy- nication. In R. E. Kasperson & P. J. M. Stallen (Eds.), Commu-
namics of environmental risk perception. In N. F. Pidgeon, R. nicating Risks to the Public. The Hague: Kluwer.
K. Kasperson, & P. Slovic (Eds.), The Social Amplification of Rousseau, D. M., Sitkin, S. B., Burt, R. S., & Camerer, C. (1998). Not
Risk. Cambridge: Cambridge University Press. so different after all: A cross-discipline view of trust. Academy
Irwin, A., Dale, A., & Smith, D. (1996). Science and Hell’s Kitchen: of Management Review, 23(3), 393–404.
The local understanding of hazard issues. In A. Irwin, & B. Siegrist, M., Cvetkovich, G. T., & Gutscher, H. (2001). Shared val-
Wynne (Eds.), Misunderstanding Science? The Public Recon- ues, social trust, and the perception of geographic cancer clus-
struction of Science and Technology. Cambridge: Cambridge ters. Risk Analysis, 21(6), 1047–1053.
University Press. Siegrist, M., Cvetkovich, G. T., & Roth, C. (2000). Salient value sim-
Johnson, B. B. (1999). Exploring dimensionality in the origins ilarity, social trust, and risk/benefit perception. Risk Analysis,
of hazard related trust. Journal of Risk Research, 2(4), 325– 20(3), 353–362.
354. Slovic, P. (1993). Perceived risk, trust and democracy. Risk Analysis,
Jungermann, H., Pfister, H. R., & Fischer, K. (1996). Credibility, 13(6), 675–682.
information preferences, and information interests. Risk Anal- Slovic, P. (in press). Risk as analysis and risk as feelings: Some
ysis, 16(2), 251–261. thoughts about affect, reason, risk, and rationality. Risk
Kasperson, R. E., Golding, D., & Kasperson, J. X. (1999). Risk, Analysis.
trust and democratic theory. In G. Cvetkovich & R. E. Löfstedt Trettin, L., & Musham, C. (2000). Is trust a realistic goal of envi-
(Eds.), Social Trust and the Management of Risk. London: ronmental risk communication? Environment and Behavior,
Earthscan. 32(3), 410–426.
Kasperson, R. E., Golding, D., & Tuler, S. (1992). Social distrust as Tyler, T. R., & Degoey, P. (1996). Trust in organizational authorities.
a factor in siting hazardous facilities and communicating risk. In R. M. Kramer & T. R. Tyler (Eds.), Trust in Organizations.
Journal of Social Issues, 48(4), 161–187. Frontiers of Theory and Research. London: Sage.
Kasperson, R. E., Kasperson, J. X., Pidgeon, N. F., & Slovic, P. Walls, J., Pidgeon, N. F., Weyman, A., & Horlick-Jones, T. (2003).
(2003). The social amplification of risk: Assessing fifteen years Critical Trust: Understanding Lay Perceptions of Health and
of research and theory. In N. F. Pidgeon, R. K. Kasperson, & Safety Risk Regulation. Journal of Risk Research (Report).
P. Slovic (Eds.), The Social Amplification of Risk. Cambridge: Norwich: Centre for Environmental Risk.
Cambridge University Press. Williams, B. L., Brown, S., & Greenberg, M. (1999). Determinants
Langford, I. H. (2002). An existential approach to risk perception. of trust perceptions among residents surrounding the Savan-
Risk Analysis, 22(1), 101–120. nah River nuclear weapons site. Environment and Behaviour,
Luhmann, N. (1979). Trust and Power. New York: Wiley. 31, 354–371.
Metlay, D. (1999). Institutional trust and confidence: A journey Wynne, B. (1980). Technology, risk and participation: On the social
into a conceptual quagmire. In G. T. Cvetkovich & R. E. treatment of uncertainty. In J. Conrad (Ed.), Society, Technol-
Löfstedt (Eds.), Social Trust and the Management of Risk. ogy and Risk Assessment. New York: Academic Press.
London: Earthscan. Wynne, B., Waterton, C., & Grove-White, R. (1993). Public Per-
Mishra, A. (1996). Organizational responses to crisis: The cen- ceptions and the Nuclear Industry in West Cumbria. Lancaster:
trality of trust. In R. M. Kramer & T. R. Tyler (Eds.), Trust Centre for the Study of Environmental Change.

Potrebbero piacerti anche