Perceptions of risk from nanotechnologies and trust in stakeholders: a cross sectional study of public, academic, government and business attitudes

Very little targeted research has been undertaken on differing stakeholder views of risk perception with regard to nanotechnology. Ho et al. [30] drew from two separate studies, one looking at public opinion, the other at the views of highly published scientists in the field of nanotechnology to compare their perceptions of risks and benefits. Although both surveys were not identical, the authors found the public perceived greater risks and less benefit for nanotechnology than the scientists, and were less likely to support funding for nanotechnology. In contrast Scheufele et al. [36] drew on the same two studies and showed that the scientists expressed more concerns than the public in the areas of pollution and new health problems with regard to nanotechnology. Siegrist et al. [23] undertook a cross sectional survey using quota sampling for the public and opportunistic sampling for experts. They found that perception of risk was significantly higher amongst the public than experts, however both had similar perceptions of benefits. The Siegrist et al. study was limited by the representativeness of both public and expert respondents with the public opinion undertaken via quota sampling and expert via attendance at a conference or working in a nano-laboratory. Our study has confirmed and enhanced this previous research in the area. We confirmed our first hypothesis that the public’s risk perception of nanotechnology is significantly greater than those working in the area of nanotechnology. This increased perception of risk is consistent regardless of whether the risk of manufactured nanomaterials is considered to oneself or if placed in one of five representative products.

Some work has been undertaken on comparative risk perceptions of specific nanotechnologies, with suggestions that food and health applications are likely to be more controversial [23]. The Australian public has shown itself less positive about nanotechnology applications that affect food and cosmetics than consumer hardware such as solar panels [25]. Further evidence has been collected to suggest that those nanotechnology applications that penetrate into the body are seen as riskier [28,37,38]. Our second hypothesis proposed that the public, academics, government and business respondents will perceive nanotechnologies that penetrate or have close and prolonged contact with the body as riskier. On this assumption, food would be seen as riskiest because of its internalisation in the body and the perception that manufactured nanomaterials are an unnatural additive to foodstuffs [5,28]. Previous research has suggested that cosmetics/sunscreens and medicines are likely to be perceived as the next most risky applications [25]. We hypothesised that their risk rating is high due to their internalisation or close prolonged contact with the body, but they are seen as less risky than food because of their beneficial to health nature. Pesticides are likely to rank after cosmetics/sunscreens and medicines because although they may be seen as hazardous, exposure will be seen as infrequent. Finally computers/tennis racquets will be seen as the least risky because of their level of contact with the body, beneficial use and lack of mental association with a hazardous product.

Our study provides evidence that both the public and academics discriminate between the risks of nanomaterials when used in different applications (Table 2). This confirms part of our second hypothesis. The survey results show that the public can differentiate between the applications of nanotechnology. The mental strategies required to undertake this differentiation were explored in our second hypothesis, which suggested that all stakeholders would rank the relative risks of nanotechnology higher or lower depending on exposure to the product, its perceived “naturalness” and its association with previous human health “disasters”. Although the concept of exposure appeared strong, with all stakeholders ranking the relative risk of manufactured nanomaterials in food or cosmetics/sunscreens as high and computers/tennis racquets as low, only the public ranked the risks as we hypothesised (food the highest risk followed by cosmetics and medicines, pesticides and finally computers). The main difference lay in the ranking of pesticides of which academic, business and government responders placed greater relative importance upon than the public (Table 2). It is difficult to interpret why this is the case. One scenario is that the relative ‘real’ risk of manufactured nanomaterials in pesticides is greater than what the public think. Another possibility is that, those working in the nanotechnology area might be more sensitive to negative perceptions of pesticides compared to the other products, or work in an area where pesticides are their interest (such as pesticide testing and regulation, especially for business and government responses where respondent numbers were low).

Examining the prevalence of risk concern, more than 60% of the Australian public agree that manufactured nanomaterials in products surveyed were a risk, excepting only consumer products such as tennis racquets and computers. Therefore in terms of regulatory importance, those sectors that are sensitive to public concern include food, cosmetics/sunscreens, medicines and pesticides. In Australia, the regulators of these sectors include Food Standards Australia New Zealand, the National Industrial Chemicals Notification and Assessment Scheme, the Therapeutic Goods Administration and the Australian Pesticides and Veterinary Medicines Authority. Overlay this with the relative importance other stakeholders place on the issues, and food and cosmetics/sunscreens are areas of greatest concern, and therefore could be seen as areas most likely to create public outrage.

One limitation of these findings comes from changing public awareness of the nanotechnology debate. The survey was administered during March 2013 just as a media debate was started around the potential risks of nanomaterials in sunscreens [39]. The ‘pro’ argument was championed by the Public Health Association of Australia, Australian Cancer Council and Australian Commonwealth Scientific and Industrial Research Organisation who based their support on Australia’s very high rates of melanoma. In this view, the demonstrable protective effects of sunscreen easily overcame the (unknown) potential risks that manufactured nanomaterials in sunscreens may present [40]. The ‘anti’ argument was championed by the environmental activist group Friends of the Earth Australia, who demanded a broader public ‘right to know’ about the use of nanomaterials. They relied on a strong version of the precautionary principle to argue that regulatory approval should be withheld until more was known about the possible detrimental effects of nanomaterials issues [41].

A review of the existing quantitative research on nanotechnology perception found inconsistent evidence on how levels of knowledge and familiarity with nanotechnology relates to perceptions [24]. The relationships between familiarity and perceptions of nanotechnology have been explored in a number of ways including surveys of support for nanotechnology in food and food packaging [29], attitudes towards nanotechnology [42], the support of use and moral acceptability of nanotechnology [43] and support for funding of nanotechnology [44]. Studies examining the interaction between nanotechnology risk perception and knowledge/familiarity have found increased knowledge/familiarity is significantly associated with a reduction in risk perception [31,45]. These results are based on views of nanotechnology as a whole. The value of this very general assessment of risk has been questioned [20], in the face of evidence that different nanotechnology applications evoke different perceptions [28,46]. These critics have advocated the assessment of nanomaterials in a more product based manner [47,48]. Our third hypothesis aimed to address these criticisms by examining the relationship between public familiarity and risk perception in a more product based manner. We found that increased public familiarity with nanotechnology was significantly associated with a reduced perception of risk. This confirms our third hypothesis. When public familiarity was compared to academic, government and business opinions using comparison of confidence intervals we found that familiarity alone could not explain the entire difference between public and academic, government and business perceptions in most cases (Tables 1 and 4). These results show that even when informed, the public is unlikely to achieve the same level of reduced risk perception in nanotechnology as those who have a professional interest in nanotechnology, a disparity important to consider in policy making. Further, it would be premature to assume from these results that increasing public familiarity with nanotechnology for example, through education, is likely to reduce perceived risks and the gap between public and academic, business and government perceptions. While familiarity was associated with a reduction in risk perception, we cannot rule out the influence cultural factors may have on these results. It should be noted that some believe values [20,45] are more important factors to consider. Another important factor to consider is message framing with qualitative studies showing, post discussion, an increase [49] or decrease [38,50] in perceived risk depending on how the message was framed. However, it can also not be ruled out that this disparity is in part induced by measurement bias, that being the measure of moderate familiarity is classified at a lower knowledge base than those who have a professional interest in nanotechnology.

Trust has long been promoted as an important factor associated with risk perception [51]. Trust reduces the complexity of decision making (or opinion formation) by accepting the risk of relying on the judgement of another party [52]. Models of trust include the causal chain mode of trust, which advocates that the direction of the association between trust and risk perception is from trust to risk perception, i.e. levels of trust shape perceptions of risk [52,53]. Within the broad trust concept, it has been argued that pre-existing knowledge, type of risk and type of trust may all influence the strength of the trust – risk perception relationship [51,54].

Trust has been shown as a significant factor when it comes to nanoparticle risk perception [24] and is an important part of the framing of public policy towards the use of nanomaterials [5,10,20]. The trust the public have in government agencies regarding nanotechnology was found to be lower than that of people working in nanotechnology based industries and nanotechnology researchers [23]. Our fourth hypothesis compared perceptions of trust between the public and academic, business and government respondents. It suggested that the trust the public vests in scientists, the health department, journalists and politicians will be less than that held by business, academic, and government officials. We partially confirmed this hypothesis by finding some significant differences between public, and academic and government levels of trust. The public do have less trust in scientists and the health department to keep them safe from any potential health effects of manufactured nanomaterials than those working in nanotechnology in academia or government.

Studies have shown that the Australian public is more likely to trust scientists and scientific institutions, followed by government agencies with industry and mass media receiving the least amount of trust [25,26]. We confirmed this in our fifth hypothesis where we found the public’s overall trust in scientists and the health department is significantly greater than the trust the public place in journalists or politicians.

Public trust in business leaders [27], science and consumer protection agencies [28] and government agencies [29] have all been associated with decreased nano risk perception. We partially confirmed this view in our sixth hypothesis where we found that those in the public who have higher levels of trust in scientists, the health department and politicians to keep them safe from any potential health effects of manufactured nanomaterials were significantly more likely to perceive lower levels of risk of manufactured nanomaterials to their health and for most products. Surprisingly, the levels of trust the public have in journalists to keep them safe from any possible health effects of nanomaterials was not associated with differing perceptions of risk. This result is unexpected, given the concern expressed over the relatively large influence the media is likely to have on public opinion in this area [5,30,55]. It is possible that this misalignment is caused through measurement bias, that is the public see the role of the journalist as an alarm bell to alert them (and the government) to the problem, rather than as a quasi-regulatory function of ‘keeping them safe’.

The results on levels of trust, hypothesis 4 to hypothesis 6, should alert policy makers to the disparity between the views of the public and ‘expert’ stakeholders. If the causal chain mode of trust, where trust affects risk perception, is applied to these results [52,53], Australian scientists and health department officials are best placed to communicate to the public about the risks and benefits of nanomaterials. However policy makers should be aware that the ‘faith’ the public bestows on scientists and health department officials may not be as large as the experts believe (or hope).

Our study contained a number of strengths. Unlike other peer reviewed publications in this area it sought opinions from a number of ‘expert’ stakeholders who are most likely to shape policy in this field. We provided a neutral definition of nanotechnology and nanomaterials. Given the potential influence framing may have on the perception of this issue, we ensured neither definition suggested benefits or risks. The concept of risk was presented in a holistic context. Respondents were asked to consider both the risks and benefits when providing a level of risk, with the Likert response scale ranging from minus two (agree strongly) to plus two (disagree strongly) to try to capture this balance. This decision was made after cognitive testing revealed that if the risk question was reversed, using the word ‘benefit’ rather than ‘risk’, the respondents perceived this ‘benefit’ question as the exact opposite of the ‘risk’ question. Risk and benefit has been justified as an inverse relationship [56], and asking about risk in this manner has been undertaken elsewhere [45,46]. Finally questions within topics were randomised and all questions that were compared across stakeholders were repeated verbatim to ensure comparability.

Our study has a number of limitations. Measurement bias may have been introduced if those who responded to our study were statistically different from those who declined to participate. In an anonymous CATI survey it is not possible to contact those who refused to participate to determine this bias. Information on academic, government and business contacts in the area of nanotechnology are difficult to acquire and all reasonable attempts were undertaken to ensure these were complete. However it is possible that our sample frame for academic, business and government did not include all those involved in the academic, business and government nanotechnology area in Australia. Also, the small number of respondents for business and government provided little power to determine associations. The response rate was similar or higher than for other groups, the sample size was an artefact of the small number of people employed in these categories. Finally, given the number of comparisons undertaken in our analysis a significance value of 0.05 meant a high probability of a type 1 error (obtaining significance by chance). Where the sample size allowed (public and academic) we also considered a significance level of 0.01. While the risk of a type 1 error was likely, many of our comparisons (and conclusions) were made across a range stakeholders or applications, with statistical significance consistent across these multiple analyses, giving us greater confidence that the effect we were detecting was real.