Attitudes and attitudinal ambivalence change towards nanotechnology applied to food production

Abstract

The strategic development of novel nanotechnologies will be determined by their public acceptance, which in turn may be influenced by public perceptions of the risks and benefits associated with the specific applications. At the present time, public opinions towards nanotechnologies remain largely inchoate, although this is likely to change with increasing public exposure to relevant information. In two experiments, a total of 618 participants, from the UK population, were provided with different risk–benefit information on nanotechnology application in food. The results show that the provision of both risk and benefit information does not influence average attitude, but results in some individuals becoming more positive and less ambivalent and others more negative and less ambivalent towards nanotechnologies. A third group maintained a neutral attitude and became more ambivalent. It is concluded that to understand public opinion formation about nanotechnology keeping track of polarization and ambivalence is important.

Keywords

attitude formation attitudinal ambivalence nanotechnology new technology public debate

1. Introduction

Nanotechnologies are frequently identified as the next scientific breakthrough that will revolutionize society (Balbus et al., 2006). To date, it appears that public knowledge about nanotechnology remains very limited. Limited information is publicly and actively communicated and the debate about the technology’s potential risks and benefits has been scarce (Kyle and Dodds, 2009). The development of attitudes towards nanotechnology will be potentially based on media reporting (Kjærgaard, 2010; Petersen et al., 2009) and public perceptions of risks and benefits involved with nanotechnology (Binder et al., 2012; Gupta, Fischer and Frewer, 2012).

The development and application of emerging technologies has been shown to be, to some extent, contingent upon societal responses to those technologies and their applications (see, inter alia, Cameron, 2006; Frewer et al., 2004). Societal and consumer response may significantly influence the development of these technologies (see e.g. Schenk et al., 2011; Tenbült et al., 2008; Van Kleef et al., 2010). Citizen and consumer protest has, for example, acted as a barrier to the successful application and commercialization of genetically modified organisms in many parts of the world (e.g. Messick and Brewer, 1983), particularly in food applications (see e.g. Devos et al., 2006; Frewer and Shepherd, 1995; Gaskell et al., 1999, 2004). Understanding the formation of public response to emerging technologies is therefore an integral part of developing a successful research and governance strategy with regard to such technologies (Ward and Barnes, 2001).

There are indications that both risk and benefit information will play an important role in the initial attitudes the public will form about nanotechnologies (Cobb and Macoubrie, 2004; Wilkinson et al., 2007). It is therefore important to take due account of how benefit and risk information contribute to public response to new technologies (Schenk et al., 2008; van Dijk et al., 2012).

Research on public attitudes about technologies and associated products has emphasized the role of risk perception (see e.g. Slovic, 1987, 1993; Slovic, Fischhoff and Lichtenstein, 1982) and has paid less attention to benefit perceptions (Johnson, 2003). In addition, much of the communication about nanotechnology applied to food production has focused on potential risks (Ronteltap, Fischer and Tobi, 2011). Nevertheless it has been found that benefit and risk perceptions cannot be studied separately from each other, and tend to be negatively correlated (Alhakami and Slovic, 1994; Siegrist et al., 2007). In addition, benefits are an important determinant of consumer decision-making, as perceived benefits may be “traded-off” against perceived risks (Fischer and De Vries, 2008; Frewer et al., 2004; Schenk et al., 2008; van Dijk et al., 2012).

The proponents of nanotechnologies frequently promote their potential benefits across a diverse number of application areas, including the controversial domain of food production (see e.g. Astier, 2009; El-Gazzar and Marth, 1992; Kleter et al., 2009). Specific characteristics of nanotechnologies are expected to facilitate the development of new products and production processes, which may have features that cannot be realized using conventional technologies.

Inevitably humans, animals and the environment will be intentionally and unintentionally exposed to nanoparticles during the production, usage and disposal of these products. The same qualities of nanotechnologies that are responsible for specific benefits, which are unachievable by conventional methods, may also cause unforeseen risks (see for example, Aitken et al., 2006; Maynard, 2006; Oberdörster, Stone and Donaldson, 2007). In this context, both risks and benefits are potentially associated with the use of engineered nanoparticles in food products, while there is still substantial uncertainty about potential health and environmental side effects (Renn and Roco, 2010; Royal Society and the Royal Academy of Engineering, 2004).

In order to understand public response towards nanotechnology, it does not suffice to assume the public will use and interpret provided information in the same way as experts do (Hansen et al., 2003). It is important to include a range of different arguments, even those deemed of less relevance to the expert. The inclusion of arguments related to the desirability of technological progress may, for example, be important to the public, while this may not be the case for nanotechnology experts (Brossard et al., 2009; Vandermoere et al., 2010, 2011). This makes the disclosure of both risk and benefit information important as non-transparent communication violates the moral right of the public to decide voluntarily (Jacobs, van de Poel and Osseweijer, 2010). The provision of benefit information, without communication about established or potential risks, may undermine public trust in those responsible for communication, as the public may infer that the communicator has a vested interest in promoting a particular issue (McKee and Coker, 2009). Increased institutional activity focused on transparent communication and fully disclosing information about science and its applications is becoming the standard for ethically sound practice (Jensen and Sandoe, 2002).

Given that many individuals do not have extensive experience with nanotechnologies (Waldron, Spencer and Batt, 2006), the information that is available may influence public opinion formation (Fischer and Frewer, 2009). Initial expressions of attitude towards nanotechnologies may change considerably as more information becomes available (Scheufele and Lewenstein, 2005), in particular since attitudes that are not strongly established do not appear to be stable predictors of long-term opinions or behaviors (Petty and Wegener, 1999). At this stage in the development of nanotechnology products, research shows that the public holds a limited and uncertain (perceived) knowledge base towards nanotechnology (Vandermoere et al., 2011; Waldron et al., 2006). The less certain individuals are about their existing attitude, the more amenable that attitude is to change following information provision (Krosnick et al., 1993).

In the current situation, public attitudes towards nanotechnology are likely to be weak or based on a few possibly contradictory statements and hence either neutral or somewhat ambivalent, partly as a consequence of a lack of information and media attention. It is important to understand how ambivalent attitudes towards nanotechnology are affected by risk–benefit information, both in relation to attitudes becoming more positive or negative, and in relation to whether attitudes become more, or less, certain (Poortinga and Pidgeon, 2006a), as attitude certainty is a predictor of stability and predictive validity of an attitude. It is important to note that attitude certainty is not necessarily linked to the extremity of an attitude (i.e. being very positive or negative). Repeated exposure to information that confirms an existing attitude may cause the recipient of the information to become more certain of their attitude, even if the attitude does not become more positive or negative (Tormala and Petty, 2002). Alternatively, if a person has received information that opposes the direction of their existing attitude, the attitude itself might not change, but the certainty with which the individual holds that attitude might decrease (Tormala, Clarkson and Petty, 2006). This shows that the formation of attitude based on risk and benefit perceptions remains a complicated topic that requires investigation regardless of the determinants of the risk and benefit perceptions themselves. As such the current research complements other research that provides more in-depth analysis of determinants of risk and benefit perceptions (e.g. Brossard et al., 2009).

Attitudinal ambivalence is one way of considering attitude certainty. Attitudinal ambivalence refers to the extent to which an individual considers an attitude object to consist of both negative and positive elements simultaneously (Jonas, Broemer and Diehl, 2000). If a person perceives both positive and negative elements to be associated with an issue, this person can be described as holding a highly ambivalent attitude. If a person only perceives positive or negative elements to be associated with an object or issue, their attitude is low in ambivalence. Similarly, if a person associates neither positive, nor negative elements with an object or issue, their (neutral) attitude can also be low in ambivalence. It is important to note that neutral attitudes can also be attitudes that evoke both very positive and very negative evaluations, in which case neutral attitudes are highly ambivalent (Hartnett, Paoli and Schaffner, 2009). The level of, and specific content of that knowledge about nanotechnology, will determine how positive, negative, neutral or ambivalent an attitude is. In the case of nanotechnology, low knowledge in the population is likely to result in a large group of neutral attitudes (such as for example observed by Cobb and Macoubrie, 2004), as well as comparable groups that tend towards a more positive or a more negative attitude (see e.g. Cobb and Macoubrie, 2004; Scheufele and Lewenstein, 2005). The differences in attitudes between people with and people without awareness are generally small (Scheufele and Lewenstein, 2005), however, the distinction between ambivalent and neutral attitudes cannot be made based on these results.

High levels of attitudinal ambivalence indicate low attitude certainty and increase the likelihood that information results in attitude change (Petty et al., 2006). It has been argued that the increased likelihood of attitude change for ambivalent attitudes is related to the negative experience of cognitive dissonance induced by holding conflicting positive and negative arguments in mind (Festinger and Carlsmith, 1959; Priester and Petty, 1996; Stone and Cooper, 2001). Information that is inherently ambivalent (such as that containing comparably extreme risk and benefit information) may therefore actually reduce ambivalence as this inherently conflicting information is likely to trigger the resolution of cognitive dissonance. This would result in individuals evaluating either the positive or the negative information as being most relevant or important (Nordgren, van Harreveld and van der Pligt, 2006). Cultural, or demographic differences are likely to influence whether the benefit or the risk information is more dominant (Kahan et al., 2009), resulting in some part of the population adopting the risk information, and shifting towards more negative attitudes, while others shift to more positive attitudes (Kahan et al., 2009).

The primary aim of the research presented here is to investigate how contradictory information about nanotechnology applications (operationalized as risk–benefit information) influences attitudes and attitudinal ambivalence associated with nanotechnologies. In the absence of prior attitudes, we expect that one-sided information (i.e. only risk or benefit information) results in attitude changes congruent with this information. The provision of both risk and benefit information (that is comparable in extremity) is not expected to necessarily lead to a change in average attitude for the whole population. However, providing arguments about both risks and benefits is likely to result in changes in attitudinal ambivalence. More specifically, following cognitive dissonance reduction theories (Festinger and Carlsmith, 1959; Priester and Petty, 1996; Stone and Cooper, 2001), it is expected that some individuals will make up their mind and adopt a less ambivalent and more outspoken, either favorable or unfavorable, position towards the technology. In other words, we expect some individuals to become more positive and others to become more negative. Such polarization of attitudes does not necessarily show in a change in average attitude in the whole population. Instead, evidence for polarization in attitudes will be indicated by increased variance in attitudes after receiving information.

2. Experiment 1

Participants

Data were collected through a market research agency (GfK-NOP; see: www.gfknop.com), with a standing panel of individuals, of whom socio-demographic information is known. The panel consists of 12,000 participants, who are repeatedly invited to participate in studies. The panel is maintained through a range of sampling techniques taking care that the panel remains representative for the population. Agreement to join the panel is between 10% and 35%; about 20% of panel members are replaced each year. The response rate for individual studies for panel members is about 60% to 70%. The use of this existing panel allowed sampling of a national sample representative on gender, age and education level of 307 individuals in the United Kingdom. Data were collected in November and December 2007, through an internet questionnaire, which participants could complete at home. Of the participants, 51% were woman, 36% had a low educational level, 42% had an intermediate educational level, and 21% had a high educational level. The mean age of participants was 44 years (SD = 14.7). Additional data were gathered on other production technologies in the same data collection wave (van Dijk et al., 2012).

Design

The experiment had four information conditions: benefit information, risk information, balanced information, and no information. Participants were randomly assigned to experimental conditions.

Attitude was measured before and after information provision on a 4-item 7-point semantic differential scale anchored by “extremely bad – extremely good,” “extremely unfavorable – extremely favorable,” “extremely undesirable – extremely desirable,” and “extremely inappropriate – extremely appropriate” (used to measure pre-information attitude α = 0.96) and “extremely dislikeable – extremely likeable,” “extremely disagreeable – extremely agreeable,” “extremely dissatisfactory – extremely satisfactory,” “extremely negative – extremely positive” (used for post-information attitude α = 0.98). Different pre- and post-information attitude scales were used to avoid repetition of responses to the same items pre- and post-information. In a pretest, the scales were shown to measure the same attitude construct (van Dijk et al., 2012).

Manipulation

Depending on the condition, participants received information about the health related risks and/or benefits associated with the use of nanotechnology in food production (see Appendix at http://pus.sagepub.com). Food production was chosen as this is an application that is likely to spark controversy (Macoubrie, 2006; Siegrist et al., 2008). In a pretest (n = 30), the risk and benefit statements were shown to be realistic (M = 4.45 and M = 4.76 on a 7-point scale) and understandable (M = 4.62 and M = 4.67 on a 7-point scale). In the same pretest, it was shown that participants perceived the risks (M = 4.55) and benefits (M = 4.52) as equally large (F(1,56) = .01; p = .92), allowing the creation of information with comparably strong perceived risk and benefits.

Procedure

Participants were invited to take part in the study by email. Participants could start the survey at their convenience during the period when data collection was active. On entering the study, participants were randomly assigned to a condition. They then received a short description of nanotechnology applied to the agri-food sector – “Applying nanotechnology to the production of food products involves the use of the special properties of very small (nano-scale) components” – and were asked to rate their pre-information attitude towards nanotechnology applied to food production. Participants were subsequently exposed to information on nanotechnology according to the condition to which they were assigned. In the conditions where participants received both risk and benefit information, the information was provided to participants on sequential computer presentations. The order of providing risk and benefit information was balanced to rule out order effects. After reading the information, the participants were asked to rate their perceptions of risks and benefits associated with nanotechnology applied to food production, followed by the post-information attitude scale. At the end of the experiment, participants provided some demographic background information. Participants received a small reward in the form of credit points from the research agency, which could be saved for a gift coupon.

Results

Change in attitudes differed across conditions (F(3,303) = 8.95; p < .001), with pairwise comparisons showing participants receiving risk information becoming less positive (M_{change risk} = -0.63; p = .03), participants receiving benefit information becoming more positive (M_{change benefit} = 0.43; p = .01) and participants receiving both risk and benefit information remaining similar in attitude (M_{change both} = -0.40; p = .19) compared to participants who did not receive information (M_{change no information} = -0.15). The apparent change in attitude towards a more negative attitude in the conditions without information was not significant (t(59) = -1.3; p = .20).

The extent to which attitudes changed after the provision of combined risk–benefit information across participants was investigated. More specifically, whether some participants became more positive and others more negative depending on information condition was analyzed by considering the change in variance, i.e. how dispersed the opinions of the participants were after receiving information. For this purpose the statistical variance, i.e. the squared standard deviation of the group mean was adopted, as this measure allows the calculation of an F value and associated levels of significance through analysis of variance. The variances in attitudes of participants who received no information did not change (pre-information attitude variance = 1.32; post-information attitude variance = 1.73; F(59,59) = 1.31; p = .15), neither did those of participants who received benefit information (pre-information attitude variance = 1.88; post-information attitude variance = 2.21; F(61,61) = 1.18; p = .26) or those who received risk information (pre-information attitude variance = 1.72; post-information attitude variance = 2.23; F(59,59) = 1.28; p = .17), showing that across the participants the opinions did not diverge after being provided with these types of information. The variance in attitudes for participants who received both risk and benefit information increased after receiving the information (pre-information attitude variance = 1.30; post-information attitude variance = 1.94; F(124,124) = 1.49; p = .01), showing that the range of attitudes across the sample became larger after the balanced information was provided.

Discussion

Provision of risk information resulted in more negative attitudes towards nanotechnology, benefit information resulted in more positive attitudes, and both risk and benefit information did not result in average changes in attitude compared to those participants who did not receive information. In addition, when both risk and benefit information was provided, attitudes showed significantly larger variance after the information than before, an effect not observed in any of the other conditions. This indicates that when both risk and benefit information was provided, some participants became more positive and some participants became more negative in their attitudes towards nanotechnology. Although such effects are likely related to reductions in ambivalence, this was not assessed directly, therefore a second experiment focused more explicitly on the effects of balanced risk–benefit information on attitudinal ambivalence.

3. Experiment 2

Participants

Data collection and sampling were identical to experiment 1 and data were collected in late 2007. Three hundred and eleven participants, who were not invited to experiment 1, took part in experiment 2. Fifty-three percent were women, 38% had low levels of education, 40% had intermediate levels of education, and 22% had high levels of education. The mean participant age was 43 years (SD = 13.6).

Design and procedure

Only combined risk–benefit information was provided in addition to a control condition (no information provision). Environmental risk and benefit statements were developed and combined with the health risk and benefit statements from experiment 1 to increase the ecological validity of the study. The provision of environmental risk or benefit, and the order of risk or benefit information, was counterbalanced, creating four variants of the risk–benefit information condition.¹ The risk and benefit statements were shown to be realistic and understandable in a pretest (no measure significantly below scale midpoint). Risks and benefits were matched in perceived extremity (confirmed in a pretest; n = 17, repeated measures F(3,13) = 2.90; p = .07; with the tendency that the health statements were perceived somewhat higher in both risk and benefit: M_{health risk} = 4.29; M_{health benefit} = 4.57; M_{environmental risk} = 3.64; M_{environmental benefit} = 3.64; see Appendix for the statements). The remainder of the procedure was identical to that in experiment 1.

Measures

The bipolar pre- and post-information attitude scales were transformed into complementary unipolar attitude statements; for example the item anchored at extremely bad – extremely good, from experiment 1 was transformed into two items on 7-point scales anchored at “not at all good” to “extremely good” and “not at all bad” to “extremely bad.” Use of complementary unipolar scales enables the assessment of implicit attitudinal ambivalence (see Jonas et al., 2000).

The overall attitude scale used to determine attitude (with the negative items recoded) was reliable (Cronbach α = .91 pre-information; Cronbach α = .97 post-information). A measure for attitudinal ambivalence was based on a positive attitude component (a subscale consisting of the items “How positive/agreeable/likable/satisfactory is the use of nanotechnology in the production of potatoes to you personally?”), and a negative attitude component (consisting of the items “How negative/disagreeable/dislikable/dissatisfactory is the use of nanotechnology in the production of potatoes to you personally?”). Both the negative and the positive component exhibited good internal reliability (Cronbach α = .97 for each component). The ambivalence measure was based on the difference in extremity of positive and negative attitude components (Jonas et al., 2000), as ambivalence does not consider valence (positive or negative) in itself, but the difference in extremity on the positive and negative attitude components (Brown and Farber, 1951 cited in Jonas et al., 2000; see equation 1).

Attitudinal Ambivalence = \frac{{Least extreme Attitude Component}^{2}}{Most extreme Attitude Component}

eq. 1

Results

There was no difference in post-information attitude between the control condition (where participants did not receive information) and the conditions in which participants received risk–benefit information (F(1,309) = 2.47; p = .12). Attitudes were not influenced by difference between health risk and environmental benefit or environmental risk and health benefit information (F(1,244) = 0.02; p = .89), the order of risk and benefit information (F(1,244) = 0.02; p = .90), or the interaction between order and type of risk or benefit (F(1,244) < 0.01; p = .99).

An ANOVA was applied to investigate the effects of information provision on attitudinal ambivalence. The results confirm our expectation that, across all participants who received risk and benefit information, attitudinal ambivalence is lower compared to that of participants who received no information (F(1,306) = 4.07; p = .04).

As there is no overall change in attitude, the lower ambivalence should be due to some participants becoming more positive and other participants becoming more negative. This is confirmed by the significant increase in variance of attitudes of participants who received risk–benefit information (pre-information attitude variance = 1.57; post-information attitude variance = 2.27; F(247,247) = 1.45; p < .01). The variance of attitudes in the control condition did not change (pre-attitude variance = 1.19; post-attitude variance = 1.44; F(62,62) = 1.21; p = .23). Additionally, the post-attitude variance of the risk–benefit information condition (N = 248; Variance = 2.27) is larger than that of the non-information condition (N = 63; Variance = 1.44; F(247,62) = 1.58; p = .02).

While the increase in variance demonstrated that some participants had become more positive and others more negative after information provision, this does not allow us to analyze whether there were systematic differences between groups of individuals becoming more positive or negative. To distinguish between such groups of participants a cluster analysis was applied to post-information attitude and post-information attitudinal ambivalence (using Ward’s method: a hierarchical cluster analysis to determine the number of clusters, followed by a K-means cluster to determine cluster membership). Three clusters were identified which significantly differed on post-information attitude (F(2,245) = 789.31; p < .01) and post-information attitudinal ambivalence (F(2,245) = 122.74; p < .01). There were sufficient participants in each post-information cluster: cluster 1: N = 103 (42%); cluster 2: N = 115 (46%); cluster 3: N = 30 (12%) to allow interpretation of the clusters (see Table 1).

Table 1.

Relation between attitude and attitudinal ambivalence (post–information clusters, N = 248).

	Total sample	Cluster 1	Cluster 2	Cluster 3
		Negative attitude, non–ambivalent	Neutral attitude, highly ambivalent	Positive attitude, non–ambivalent
N	248	103	115	30

Average of clustering variables (SD)^†					F	p

(post–information) Attitude	3.15 (1.51)	1.59^a (0.67)	3.96^b (0.26)	5.45^c (0.82)	(2,245) = 789.31	<.01
(post–information) Attitudinal ambivalence	1.37 (1.31)	0.19^a (0.34)	2.64^b (0.67)	0.44 ^a(0.45)	(2,245) = 122.74	<.01

Interpretation variables^‡					χ²	p ^§

Gender (% female)	52%	58%^a	53%	23%^b	11.52 (df = 2, N = 248)	<.01
Age (18–29)	23%	16%^b	25%	40%^a	13.26 (df = 6, N = 248)	.04
Age (30–39)	18%	16%	21%	13%^b
Age (40–49)	17%	22%^a	11%^b	17%
Age (50+)	43%	47%	43%	30%^b
Low education	39%	43%	40%	20%^b	11.80 (df = 4, N = 248)	.02
Medium education	40%	42%	39%	37%
High education	21%	13%^b	16%^b	53%^a
Below average income	52%	54%	52%	47%^b	0.66 (df = 4, N = 248)	.96^ns
About average income	24%	20%	23%	23%
Above average income	26%	25%	25%	30%

Average of attitude and attitudinal ambivalence change (SD)^†					F	p

Attitude change	−0.38 (1.10)	−1.10^a (1.14)	0.03^b (0.55)	0.52^c (1.13)	(2,245) = 58.29	<.01
Attitudinal ambivalence change	−0.36 (1.30)	−0.94^a (1.22)	0.27^b (1.14)	−0.79^a (1.04)	(2,245) = 31.65	<.01

†

Different scores between clusters are identified by different superscript characters (Tukey HSD).

‡

Superscript ^a indicates a value at least 5% higher than the expected value, superscript ^b indicates a value at least 5% lower than the expected value.

Significant at α = .05; superscript ^ns indicates non–significant difference between clusters.

Together clusters 1 and 2 covered 88% of the total sample. Compared to participants who were members of cluster 2, members of cluster 1 were more negative, and held less ambivalent attitudes. Members of cluster 2 were more neutral and ambivalent, indicating that their attitudes towards nanotechnology in food consisted of both positive and negative aspects. The demographic differences between these clusters were limited. Cluster 1 was slightly dominated by women (58%) although it is important to note that this does not reflect a huge difference in population terms. Members of cluster 1 also tended to fall into the older age group, and have a low-to-moderate level of education. The demographic make-up of cluster 2 contained slightly fewer female participants, tended to be slightly younger, and contained more highly educated participants compared to cluster 1. Members of cluster 3 were more positive and less ambivalent following the provision of information about nanotechnology. This cluster contained only 12% of the total sample. Participants in this cluster tended to be male (77%), and tended to fall into either the youngest or oldest age groups (40% and 30% of cluster membership). Members of cluster 3 also tended to be the most highly educated in the sample, although part of this may be due to the combination of age and gender, where especially older women tended to have lower education.

While the cluster analysis resulted in three groups with different levels of attitude and attitudinal ambivalence, this analysis does not necessarily confirm that these differences are the result of changes in attitude and ambivalence due to information. There may be an alternative explanation, i.e. that the post-information values for attitude and ambivalence are determined by pre-information attitude and ambivalence. To rule out this alternative explanation, change in attitude and attitudinal ambivalence between pre-information and post-information measures was investigated.

A MANOVA was conducted on the changes between pre- and post-information attitude and attitudinal ambivalence, confirming that attitude and attitudinal ambivalence changed differently between the clusters (F(2,245) = 59.66; p < .01). ANOVAs showed that both attitude (F(2,245) = 58.29; p < .01) and attitudinal ambivalence (F(2,245) = 31.65; p < .01) changed differently for participants in the three different post-information clusters, confirming that the information did indeed have different effects across the clusters on both attitude and attitudinal ambivalence. More specifically, participants in cluster 1 (negative attitude, low ambivalence) had changed significantly more towards the negative (M_change = -1.10) compared to both other clusters; participants in cluster 3 (positive attitude, low ambivalence) had changed significantly more toward the positive (M_change = 0.52) compared to both other clusters (post hoc comparison, Tukey HSD). Participants in clusters 1 and 3 (negative attitude low ambivalence and positive attitude low ambivalence, respectively) had significantly changed towards a less ambivalent attitude (M_change1 = -0.94; M_change3 = -0.79) compared to cluster 2 (M_change = 0.27). There was no significant difference in ambivalence change between clusters 1 and 3 (post hoc comparison, Tukey HSD). Thus it was confirmed that the provision of information resulted in two clusters whose members became more extremely positive or negative in attitude to show a reduction in ambivalence compared to the neutral cluster. A one-sample t-test was subsequently used to investigate whether there were significant changes between pre- and post-information attitude and attitudinal ambivalence within clusters. The change in attitude was negative for cluster 1 (M_change = -1.10; t(102) = -9.83; p < .01) and positive for cluster 3 (M_change = 0.52; t(29) = 2.54; p = .02). No change in attitude was found for cluster 2 (M_change = 0.3; t(114) = 0.51; p = .61). Changes in attitudinal ambivalence indicated a significant ambivalence reduction for clusters 1 and 3 (cluster 1: M_change = -0.94; t(102) = -7.81; p < .01; cluster 3: M_change = -0.79; t(29) = -4.19; p < .01). Change in attitudinal ambivalence was significantly positive for cluster 2 (M_change = 0.27; t(114) = 2.53; p = .01). This indicates that the clusters that became more extreme also became less ambivalent, while the cluster that remained neutral became significantly more ambivalent.

Discussion of experiment 2

Experiment 2 confirmed the findings of experiment 1 that the provision of risk–benefit information leads to increased diversity in post-information attitudes of participants, as indicated by the larger variance in attitude scores. It was confirmed that this increased variance was due to changes in ambivalence, with some participants making up their mind, becoming either positive or negative and less ambivalent, while, unexpectedly, a large group of participants did not yet make up their mind and remained neutral but became more ambivalent. Socio-demographic variables related to participants becoming more positive or negative and less ambivalent.

4. General discussion and conclusion

Public response towards nanotechnology is likely to be influenced by information about both risks and benefits associated with the technology. The current research showed that provision of risk and benefit information did not change average attitudes, either when risks and benefits were balanced within the health domain, or when they were balanced across health and environmental information. Average attitudinal ambivalence of participants exposed to information reduced after the provision of information having contradictory, positive and negative elements. Although on average attitudes did not change, a number of participants became more focused on either positive or negative arguments, which resulted in reduced ambivalence and more extreme attitudes that were either more positive or more negative. It is of interest to note that only a small proportion of individuals became more positive – and this subsample of participants tended to be male and more highly educated. Many of the individuals who received risk–benefit information, however, did not form a more positive or negative attitude, but remained neutral and showed an increased ambivalence following information provision instead.

The finding that some individuals form more extreme positive or negative attitudes when exposed to balanced information suggests that these individuals do not like to have conflicting ideas on a topic and prefer to reconcile the different arguments presented, such that only one argument is adopted, and the other is ignored (Sengupta and Johar, 2002). This implies these individuals either adopt the information about benefit and form a more positive and less ambivalent attitude, or use the information about risk and form a more negative and less ambivalent attitude. This is in line with cognitive dissonance resolution (Johnson et al., 2008; Jonas et al., 2001), which predicts that individuals are motivated to resolve the problem of holding opinions based on multiple, conflicting, arguments by actively discarding part of the information, resulting in an individual adopting only one side of the argument; either becoming more positive and less ambivalent, or more negative and less ambivalent (Nordgren et al., 2006).

The large group of individuals whose attitudes did not change but did become more ambivalent are of particular interest, as they seem not to have solved cognitive dissonance and accept high levels of uncertainty in their attitude. The further development of attitudes in this rather large and highly ambivalent, neutral segment in the population may determine societal acceptance of nanotechnology. It is unlikely that the lack of change in attitude is due to these individuals being uninvolved in the topic and having ignored the information, as their attitudinal ambivalence increased post-information provision. These individuals may instead be postponing the decision about whether nanotechnology is positive or negative until trusted “opinion leaders” express a positive or negative view. In this context it becomes important to note that the agenda regarding technology adoption will likely be set by actions of its proponents and opponents, not those who are ambivalent. In other words the opinion leaders are likely not to be neutral. Different population segments may trust different opinion leaders based on the fit between their own values and those values they perceive an opinion leader holds (Poortinga and Pidgeon, 2006b). Differentiation in attitude may also be influenced by a match between personal values, and the specific information provided. For example, some people may put more emphasis on health arguments, while others weigh environmental effects more heavily (Schenk et al., 2011). Additionally, it is possible that some statements compared to others, are perceived at different levels of abstraction, or otherwise as psychologically more removed from the self, causing differential interpretation across population groups (cf. Kahan et al., 2009; Liberman, Trope and Wakslak, 2007). As a consequence, different segments of the public may form different attitudes. For example, the cluster that became more positive showed an overrepresentation of men and highly educated participants. Although some care is needed to interpret these outcomes at “face value” as the size of the group under consideration was small, this aligns with previous research which has shown that these individuals tend to be favorable towards new technologies regardless of associated risks (cf. Flynn, Slovic and Mertz, 1994; Lindberg, Grimes and Giles, 2005; Marshall et al., 2006). For bioengineering, similar results showed that where men tended to become more positive towards technology with increasing knowledge, the opinion of women did not change or even became somewhat more critical with increasing knowledge (Simon, 2010). This effect may be partially determined by culture, for example because the cultural background of women may still reflect long-standing practice of steering women away from science (e.g. Baker and Leary, 1995; Simon, 2010). Thus future research is needed to state with confidence what demographics, valuation and interpretation of statements across different levels of abstraction, and different value domains in relation to associated cultural opinions, are the best predictors for public response and public ambivalence towards new technologies (cf. Kahan et al., 2009). A word of caution in interpreting these results is needed. Effort was undertaken to ensure representativeness on key demographics in the sample. Nevertheless, the use of an existing panel through the internet may have introduced some bias. First of all, some self-selection biases in agreeing to participate in a panel may be present, which do not show on the demographic variables used to compare the sample to the overall population; secondly there may be some bias in internet access which covered 61% of UK households at the time of the reported survey.² Therefore, the actual quantified sizes of the identified segments may be somewhat different from the population as a whole. There is, however, no reason to doubt the higher level observation that some people become more negative, others more positive, while others remain neutral and become more ambivalent, and that this is related to specific socio-economic determinants,

The current study shows that some individuals tended to form an opinion about nanotechnology quickly. In contrast a second group was identified who are less quick in forming opinions. The crystallization of the opinions of this large group of undecided individuals in society will be pivotal in the debate on nanotechnology. The development of attitudes and attitude certainty of this group may decide to what extent the public debate becomes polarized and entrenched. To understand the dynamics in the public debate about nanotechnologies it is essential to comprehensively monitor the development of attitudes, attitude polarization and attitudinal ambivalence.

Footnotes

Funding

The research reported is funded by the European Commission and results from the fourth work package (WP4) of the integrated project SAFE FOODS, Promoting food safety through a new integrated risk analysis approach for foods (FP6-506446).

Notes

Author biographies

Arnout R.H. Fischer is assistant professor in the Marketing Consumer Behaviour group at Wageningen University.

Heleen van Dijk is postdoctoral researcher in the Marketing Consumer Behaviour group at Wageningen University.

Janneke de Jonge is assistant professor in the Marketing Consumer Behaviour group at Wageningen University.

Gene Rowe is a visiting scientist in the Marketing Consumer Behaviour group at Wageningen University, and is interested in “science and society” issues.

Lynn J. Frewer is Professor of Food and Society at the Centre for Rural Economy, Newcastle University.

References

Aitken

Chaudhry

Boxall

ABA

Hull

(2006) Manufacture and use of nanomaterials: Current status in the UK and global trends. Occupational Medicine 56(5): 300–306.

Alhakami

Slovic

(1994) A psychological study of the inverse relationship between perceived risk and perceived benefit. Risk Analysis 14(6): 1085–1096.

Astier

(2009) Milk falsified by melamine: Another scandal in public health [Lait falsifié par la mélamine : un autre scandale de santé publique]. Annales Pharmaceutiques Francaises 67(1): 1–2.

Baker

Leary

(1995) Letting girls speak out about science. Journal of Research in Science Teaching 32(3): 3–27.

Balbus

Florini

Denison

Walsh

(2006) Getting it right the first time: Developing nanotechnology while protecting workers, public health, and the environment Annals of the New York Academy of Sciences 1076: 331–342.

Binder

Cacciatore

Scheufele

Shaw

Corley

(2012) Measuring risk/benefit perceptions of emerging technologies and their potential impact on communication of public opinion toward science. Public Understanding of Science 21(7): 830–847.

Brossard

Scheufele

Kim

Lewenstein

(2009) Religiosity as a perceptual filter: Examining processes of opinion formation about nanotechnology. Public Understanding of Science 18(5): 546–558.

Cameron

(2006) Nanotechnology and the human future: Policy, ethics, and risk. Annals of the New York Academy of Sciences 1093: 280–300.

Cobb

Macoubrie

(2004) Public perceptions about nanotechnology: Risks, benefits and trust. Journal of Nanoparticle Research 6(4): 395–405.

10.

Devos

Reheul

De Waele

Van Speybroeck

(2006) The interplay between societal concerns and the regulatory frame on GM crops in the European Union. Environmental Biosafety Research 5(3): 127–149.

11.

El-Gazzar

Marth

(1992) Foodborne disease: Investigative procedures and economic assessment. Journal of Environmental Health 55(2): 24–27.

12.

Festinger

Carlsmith

(1959) Cognitive consequences of forced compliance. Journal of Abnormal and Social Psychology 58(2): 203–210.

13.

Fischer

ARH

De Vries

(2008) Everyday behaviour and everyday risk: An approach to study people’s responses to frequently encountered food related health risks. Health Risk and Society 10(4): 385–397.

14.

Fischer

ARH

Frewer

(2009) Consumer familiarity with foods and the perception of risks and benefits. Food Quality and Preference 20(8): 576–585.

15.

Flynn

Slovic

Mertz

(1994) Gender, race, and perception of environmental health risks. Risk Analysis 14(6): 1101–1108.

16.

Frewer

Shepherd

(1995) Ethical concerns and risk perceptions associated with different applications of genetic engineering: Interrelationships with the perceived need for regulation of the technology. Agriculture and Human Values 12(1): 48–57.

17.

Frewer

Lassen

Kettlitz

Scholderer

Beekman

Berdal

(2004) Societal aspects of genetically modified foods. Food and Chemical Toxicology 42(7): 1181–1193.

18.

Gaskell

Allum

Wagner

Kronberger

Torgersen

Hampel

Bardes

(2004) GM foods and the misperception of risk perception. Risk Analysis 24(1): 185–194.

19.

Gaskell

Bauer

Durant

Allum

(1999) Worlds apart? The reception of genetically modified foods in Europe and the U.S. Science 285(5426): 384–387.

20.

Gupta

Fischer

ARH

Frewer

(2012) Socio-psychological determinants of public acceptance of technologies: A review. Public Understanding of Science 21(7): 782–795.

21.

Hansen

Holm

Frewer

Robinson

Sandoe

(2003) Beyond the knowledge deficit: Recent research into lay and expert attitudes to food risks. Appetite 41(2): 111–121.

22.

Hartnett

Paoli

Schaffner

(2009) Modeling the public health system response to a terrorist event in the food supply. Risk Analysis 29(11): 1506–1520.

23.

Jacobs

van

Poel

Osseweijer

(2010) Sunscreens with titanium dioxide (TiO₂) nano-particles: A societal experiment. NanoEthics 4(2): 103–113.

24.

Jensen

Sandoe

(2002) Food safety and ethics: The interplay between science and values. Journal of Agricultural and Environmental Ethics 15: 245–253.

25.

Johnson

(2003) Further notes on public response to uncertainty in risks and science. Risk Analysis 23(4): 781–789.

26.

Johnson

Cooke

Croker

Wardle

(2008) Changing perceptions of weight in Great Britain: Comparison of two population surveys. BMJ 337(7664): 270–272.

27.

Jonas

Schulz-Hardt

Frey

Thelen

(2001) Confirmation bias in sequential information search after preliminary decisions: An expansion of dissonance theoretical research on selective exposure to information. Journal of Personality and Social Psychology 80(4): 557–571.

28.

Jonas

Broemer

Diehl

(2000) Attitudinal ambivalence. European Review of Social Psychology 11: 35–74.

29.

Kahan

Braman

Slovic

Gastil

Cohen

(2009) Cultural cognition of the risks and benefits of nanotechnology. Nature Nanotechnology 4(2): 87–90.

30.

Kjærgaard

(2010) Making a small country count: Nanotechnology in Danish newspapers from 1996 to 2006. Public Understanding of Science 19(1): 80–97.

31.

Kleter

Prandini

Filippi

Marvin

HJP

(2009) Identification of potentially emerging food safety issues by analysis of reports published by the European Community’s Rapid Alert System for Food and Feed (RASFF) during a four-year period. Food and Chemical Toxicology 47(5): 932–950.

32.

Krosnick

Boninger

Chuang

Berent

Carnot

(1993) Attitude strength: One construct or many related constructs? Journal of Personality and Social Psychology 95(9): 1132–1151.

33.

Kyle

Dodds

(2009) Avoiding empty rhetoric: Engaging publics in debates about nanotechnologies. Science and Engineering Ethics 15(1): 81–96.

34.

Liberman

Trope

Wakslak

(2007) Construal level theory and consumer behavior. Journal of Consumer Psychology 17(2): 113–117.

35.

Lindberg

Grimes

Giles

(2005) Farm-to-table: A situation awareness model for food safety assurance for porous borders. Comprehensive Reviews in Food Science and Food Safety 4(2): 31–33.

36.

Macoubrie

(2006) Nanotechnology: Public concerns, reasoning and trust in government. Public Understanding of Science 15(2): 221–241.

37.

Marshall

Picou

Formichella

Nicholls

(2006) Environmental risk perceptions and the white male effect: Pollution concerns among deep-South coastal residents. Journal of Applied Social Science 23(2): 31–49.

38.

Maynard

(2006) Nanotechnology: Assessing the risks. Nano Today 1(2): 22–33.

39.

McKee

Coker

(2009) Trust, terrorism and public health. Journal of Public Health 31(4): 462–465.

40.

Messick

Brewer

(1983) Solving social dilemmas: A review. Review of Personality and Social Psychology 4: 11–44.

41.

Nordgren

van Harreveld

van der Pligt

(2006) Ambivalence, discomfort, and motivated information processing. Journal of Experimental Social Psychology 42(2): 252–258.

42.

Oberdörster

Stone

Donaldson

(2007) Toxicology of nanoparticles: A historical perspective. Nanotoxicology 1(1): 2–25.

43.

Petersen

Anderson

Allan

Wilkinson

(2009) Opening the black box: Scientists’ views on the role of the news media in the nanotechnology debate. Public Understanding of Science 18(5): 512–530.

44.

Petty

Wegener

(1999) The elaboration likelihood model: Current status and controversies. In: Chaiken

Trope

(eds) Dual Process Theories in Social Psychology. New York, NY: Guilford Press, 37–72.

45.

Petty

Tormala

Brinol

Jarvis

WBG

(2006) Implicit ambivalence from attitude change: An exploration of the PAST model. Journal of Personality and Social Psychology 90(1): 21–41.

46.

Poortinga

Pidgeon

(2006a) Exploring the structure of attitudes toward genetically modified food. Risk Analysis 26(6): 1707–1719.

47.

Poortinga

Pidgeon

(2006b) Prior attitudes, salient value similarity, and dimensionality: Toward an integrative model of trust in risk regulation. Journal of Applied Social Psychology 36(7): 1674–1700.

48.

Priester

Petty

(1996) The gradual threshold model of ambivalence: Relating the positive and negative bases of attitudes to subjective ambivalence. Journal of Personality and Social Psychology 71(3): 431–449.

49.

Renn

Roco

(2010) Nanotechnology Risk Governance. Geneva: International Risk Governance Council, 108.

50.

Ronteltap

Fischer

ARH

Tobi

(2011) Societal response to nanotechnology: Converging technologies–converging societal response research? Journal of Nanoparticle Research 13(10): 4399–4410.

51.

Royal Society and the Royal Academy of Engineering (2004) Nanoscience and Nanotechnologies: Opportunities and Uncertainties. London: The Royal Society and the Royal Academy of Engineering, 111.

52.

Schenk

Fischer

ARH

Frewer

Gilissen

LJWJ

Jacobsen

Smulders

MJM

(2008) The influence of perceived benefits on acceptance of GM applications for allergy prevention. Health, Risk and Society 10(3): 263–282.

53.

Schenk

van der Maas

Smulders

MJM

Gilissen

LJWJ

Fischer

ARH

Van der Lans

Frewer

(2011) Hypoallergenic food products as a novel approach to alleviate mild food allergy. Food Quality and Preference 22(1): 83–91.

54.

Scheufele

Lewenstein

(2005) The public and nanotechnology: How citizens make sense of emerging technologies. Journal of Nanoparticle Research 7(6): 659–667.

55.

Sengupta

Johar

(2002) Effects of inconsistent attribute information on the predictive value of product attitudes: Toward a resolution of opposing perspectives. Journal of Consumer Research 29(1): 39–56.

56.

Siegrist

Cousin

Kastenholz

Wiek

(2007) Public acceptance of nanotechnology foods and food packaging: The influence of affect and trust. Appetite 49(2): 459–466.

57.

Siegrist

Stampfli

Kastenholz

Keller

(2008) Perceived risks and perceived benefits of different nanotechnology foods and nanotechnology food packaging. Appetite 51(2): 283–290.

58.

Simon

(2010) Gender differences in knowledge and attitude towards biotechnology. Public Understanding of Science 19(6): 642–653.

59.

Slovic

(1987) Perception of risk. Science 236(4799): 280–285.

60.

Slovic

(1993) Perceived risk, trust, and democracy. Risk Analysis 13(6): 675–682.

61.

Slovic

Fischhoff

Lichtenstein

(1982) Why study risk perception? Risk Analysis 2(2): 83–93.

62.

Stone

Cooper

(2001) A self-standards model of cognitive dissonance. Journal of Experimental Social Psychology 37(3): 228–243.

63.

Tenbült

De Vries

Dreezens

Martijn

(2008) Intuitive and explicit reactions towards “new” food technologies: Attitude strength and familiarity. British Food Journal 110(6): 622–635.

64.

Tormala

Petty

(2002) What doesn’t kill me makes me stronger: The effects of resisting persuasion on attitude certainty. Journal of Personality and Social Psychology 83(6): 1298–1313.

65.

Tormala

Clarkson

Petty

(2006) Resisting persuasion by the skin of one’s teeth: The hidden success of resisted persuasive messages. Journal of Personality and Social Psychology 91(3): 423–435.

66.

van Dijk

Fischer

ARH

De Jonge

Rowe

Frewer

(2012) The impact of balanced risk-benefit information and prior attitudes on post-information attitudes. Journal of Applied Social Psychology 42(8): 1958–1983.

67.

Van Kleef

Fischer

ARH

Khan

Frewer

(2010) Risk and benefit perceptions of mobile phone and base station technology in Bangladesh. Risk Analysis 30(6): 1002–1015.

68.

Vandermoere

Blanchemanche

Bieberstein

Marette

Roosen

(2010) The morality of attitudes toward nanotechnology: About god, techno-scientific progress, and interfering with nature. Journal of Nanoparticle Research 12(2): 373–381.

69.

Vandermoere

Blanchemanche

Bieberstein

Marette

Roosen

(2011) The public understanding of nanotechnology in the food domain: The hidden role of views on science, technology, and nature. Public Understanding of Science 20(2): 195–206.

70.

Waldron

Spencer

Batt

(2006) The current state of public understanding of nanotechnology. Journal of Nanoparticle Research 8(5): 569–575.

71.

Ward

Barnes

(2001) Control and affect: The influence of feeling in control of the retail environment on affect, involvement, attitude, and behavior. Journal of Business Research 54(2): 139–144.

72.

Wilkinson

Allan

Anderson

Petersen

(2007) From uncertainty to risk? Scientific and news media portrayals of nanoparticle safety. Health, Risk and Society 9(2): 145–157.