Determinants of reactions to gene technology: a generic approach

Abstract

This paper examines the reactions to gene technology (the intention to buy gene-tech food, worry about abuse, and the public's desire that different actors be able to influence decisions) in a sample of the Dutch population (n = 1010) and studies the relationship between these reactions and perception, trust, experience, knowledge and personal interest. The survey reveals that large parts of the public are concerned about the abuse of gene technology, are not willing to buy gene-tech products, and want actors to have an influence on legislation and enforcement. Path analysis shows that these reactions can be well explained using a generic model. Trust in authorities, personal interest in gene technology, and perception of gene technology are important predictors of people's reactions, whereas experience and knowledge are less important.

Keywords:

• perception
• trust
• gene technology
• cognitive and affective factors

Introduction

Gene technology is currently the subject of both scientific and public debate. While technological development continues and the technology's potential grows, public support has not followed suit. When science approaches value-laden issues like therapeutic cloning or stem cell research, the public is increasingly claiming a voice in regulation (Leshner 2005). Perceived risks for health and the environment, the perception that producers benefit more than consumers, and ethical questions have led the public to hesitate about accepting gene technology in medical applications and especially in food and agricultural applications (Gaskell et al. 2004, 2006). With applications like gene-tech food, storing DNA information in databases, or mapping DNA in the search for cures for inherited diseases, gene technology applications have reached the transaction level (Horst et al. 2007). This means that consumers actually encounter gene technology products and services, and are not exclusively dependent on media information in order to form an opinion (Gutteling 2005, Bauer and Gutteling 2006).

Individuals play different roles – as consumers, citizens or patients – and this means that at the transaction level, the processes by which individuals reach decisions about an application's risks, benefits and the associated ethical considerations are more complex than those at the previous, information level. This also means that the consequences these decision processes have on the individual's behavior may become clearer.

An extensive review of the peer-reviewed literature on public reactions toward gene technology (Pin and Gutteling 2006) showed that only a few studies have addressed transaction-level issues regarding gene technology (e.g. Townsend and Campbell 2004). Many studies were characterized by single-issue designs, in which public reactions were analyzed or modeled for one particular application of gene technology, such as gene-tech food (e.g. Moon and Balasubramanian 2004, Mucci et al. 2004) and prenatal genetic testing and prenatal genetic engineering (e.g. Urban 1996), and others focused on one particular dependent variable, such as the acceptance of gene technology (e.g. Tanaka 2004). This research strategy assumes that each application or dependent variable is associated with a separate psychological decision process. The question that we want to address in this paper is whether this is a valid assumption.

Using data from a representative sample of the Dutch public, we analyzed the determinants of the psychological decision process for various applications of gene technology at the transaction level. Path analysis produced a first attempt to deliver a generic perspective on public reactions toward gene technology applications. Our literature review did not reveal any study that has attempted this before, although studies focusing on particular domains of this topic have used statistical modeling techniques (e.g. Siegrist 1999, Grunert et al. 2003, Tanaka 2004).

Theoretical framework

This study focuses on three different reactions, namely the intention to buy gene-tech food, the concern about the abuse of gene technology, and the public's desire that different (societal) actors be able to influence gene technology development. The three reactions are easily measurable and can be used to operationalize the hopes and concerns of the public. These different reactions reflect the transaction level of gene technology development. Unlike earlier phases, when the public had only indirect experiences to rely on, now the public can respond to a combination of direct experiences with the technology and indirect experiences are based on media information, narratives, and so on.

Determinants of reactions to gene technology

These three reactions have a broad range of socio-psychological determinants. Since the early 1980s, various models have been developed to explain and predict the public perception of technological developments (Dowling and Staelin 1994). Positive or negative perception has proven to be an important determinant of reactions to new technologies. If we perceive a benefit from a behavior or choice, the risk associated with it seems smaller. If there is no perceived benefit, the risk seems larger (Frewer 2004).

Studies have also provided an understanding of cognitive determinants of the perception of technology development, like personal interest (Pardo et al. 2002) and knowledge (Shaw 2002). The extent to which an individual is interested in and informed about gene technology developments has been shown to be a predictor of the perceptions of gene technology applications (Pardo et al. 2002). Personal interest has a significant effect on knowledge, with interest having a positive effect; people with more interest acquire more knowledge. In the case of GM foods, it is clear that higher levels of knowledge coincide with higher levels of acceptance (Moerbeek and Casimir 2005). On the other hand, knowledge may be just one of the many factors that influence the opinions concerning GM foods (Cuite et al. 2005). Others have found that knowledge increases critical opinions about biotechnology (Bauer and Gutteling 2006).

Research on other new technologies has shown that experience with the technology (that is, having contact with the technology at the transaction level) can influence knowledge and interest (Grunert et al. 2003). In the EU, direct experience with gene technology in everyday life, such as gene-tech food or medical applications, was still rather rare several years ago (Henneman et al. 2004), but this is changing rapidly.

Recent risk literature distinguishes between cognitive and affective components of perception (Slovic et al. 2004). The literature review by Pin and Gutteling (2006) identified both cognitive and affective determinants of the perception of gene technology and its consequences. The “cognitive system” uses algorithms and normative rules, with knowledge being one of its most important motives. This system is relatively slow, requiring effort and conscious control. Relying on images and associations linked by experience to emotion and affect (a feeling that something is good or bad), the “affective system” is intuitive, fast, mostly automatic, and does not depend on conscious awareness. These two systems operate in parallel, seeming to depend on each other for guidance (Slovic et al. 2004).

Various concepts that can be attributed to the systems can be found in the literature. Trust is a multidimensional concept that has frequently been identified as an affective determinant1 for perception and reactions to gene technology (Frewer et al. 2003, Tanaka 2004). “Social trust” refers to people's willingness to rely on experts and institutions when managing risks and technologies as a generalized attitude over particular issues and institutions. Trust in regulators, science, and industry is particularly important when the public perceives itself as having no control over a particular event or activity, having to leave the responsibility for ensuring consumer protection or public welfare to others – as is arguably the case with genetically modified foods (Frewer 2004). Increased trust in regulators in the gene technology field decreases perceived risk and increases perceived benefit (Siegrist 2000, Frewer 2004). Higher levels of trust produce a more positive attitude toward gene-food and make it more likely that people will accept it (Gutteling et al. 2006).

Frewer et al. (1996) found that information sources are associated with different characteristics that influence the extent to which they are trusted by the public. Expertise by itself does not lead to trust but must be accompanied by other characteristics, such as accountability. Sources with a moderate degree of accountability tend to be trusted more than those with complete freedom (Finucane and Holup 2005). Siegrist (2000) demonstrated that trust in the companies and scientists conducting research on gene technologies has a strong effect on the risks and benefits perceived to be associated with those technologies. It follows that the more an individual trusts that the biotechnology industry and scientists studying gene technology have the interests of the general population and environment at heart, the less risk and more benefit the individual will perceive to be associated with their scientific endeavors. Other factors, such as ethical and moral considerations or uncertainties and concerns about possible unintended effects, are also determinants of consumer acceptance or rejection of emerging technologies and their products. An important issue, therefore, is exactly how to include societal values when making decisions about regulation and innovation (Frewer 2004).

A generic model describing the reactions to gene technology

In contrast to most models (e.g. Urban 1996, Moon and Balasubramanian 2004, Mucci et al. 2004), in this study we developed and tested a generic model describing the relationships between the variables identified in the literature review (Pin and Gutteling 2006) (see Figure 1). This model is not specific to any particular application of gene technology. In the diagram, the perception of gene technology was given a central position, and arrows indicate potential causal relationships. Experience with gene technology, personal interest in gene technology, and trust in actors were considered to be determinants of the perception of gene technology, whereas the intention to buy gene-tech products, concern about the abuse of gene technology and desired influential actors were considered to be dependent on this perception.

Figure 1. Results of testing the generic model.^a Notes.

The literature review identified determinants of the perception of gene technology and its consequences, reflecting cognitive processes, affective processes, and the interaction between the two (Slovic et al. 2004). In the model, experience with gene technology is hypothesized to determine personal interest in and knowledge about the field. Personal interest in gene technology is in turn hypothesized to determine the perception of gene technology. Trust in actors is also considered to be an important determinant of the perception of gene technology.

The central position given to the perception of gene technology does not rule out the possibility that the determinants can also directly influence the consequences; this explains the arrows from interest and trust to the intention to buy gene-tech products, concern about the abuse of gene technology, and desired influential actors. Experience with gene technology and trust in actors were assumed to be unrelated.

Method

Respondents

In November 2005, we conducted telephone interviews with a randomly drawn sample of 3119 Dutch individuals, aged 18 and older. Of these, 1010 respondents participated in the research (33% response rate). Those who declined to participate mainly gave as reasons “don't feel like it”, “no time”, or “don't see the use of it” and 15% indicated that the subject did not appeal to them.

Of the 1010 respondents, 39.5% were male and 60.5% female. Age varied between 18 and 90, with a mean of 49 years of age. As to education: 37.9% of the respondents had a high level of education, while 36.2% reported a medium level of education and 25.9% reported a low education level. Household composition, socio-economic position, religious affiliation and the residence of the respondents resembled the Dutch adult population: 78.7% had children, 57.8% were in paid employment, and 23.8% said they visited church, mosque, or temple regularly (Statistics Netherlands [CBS] 2006).

Questionnaire

This study was based on an earlier survey conducted in 2002, which aimed to reveal trends in the public reaction to applications of gene technology.2 In designing our questionnaire, we built upon this existing measurement instrument (Pin and Gutteling 2005).

Dependent variables: reactions to gene technology

Three different reactions to gene technology were measured as dependent variables. The intention to buy gene-tech products (e.g. Mucci et al. 2004) was assessed with two items: “Would you buy fruit or vegetables that are genetically modified?” and “Would you buy food that contains a genetically modified ingredient, for example margarine with genetically-changed soy-protein?” The items were measured on a five-point scale (definitely not – definitely yes). The correlation between the two items was 0.79.

The intention to buy or not buy a gene-tech product is one way for the public to indicate their reactions toward this technology. In line with the theory of emotion (Bandura 1989, Lazarus 1993), we assume that concern about the technology is another way of expressing these reactions. Therefore, concern about the abuse of gene technology was measured with three statements relating to the abuse of DNA material by three different actors: assurance companies, employers, and judicial authorities. The items were measured on a five-point scale (not at all concerned – very concerned). The three items had a Cronbach's alpha of 0.78.

In addition to direct expressions, the public may also express their reaction toward gene technology by supporting societal organizations (NGOs, consumer organizations, or patient organizations) in monitoring gene technology developments and influencing decisions relating to them. In previous studies, we found that large segments of the public trust these organizations (Gutteling et al. 2006). So, the public's desire that different actors be able to influence gene technology development was measured by asking which of 12 actors (see Table 1) should be able to influence the boundaries of research on genes and heredity (two-point scale: yes – no).

Table 1. Results for the dependent variables (n = 1010), and correlation with demographic variables.

		Relation with demographic variables^a
Construct	Description and items of scale	Gender (male)	Age	Education	Having children	Being religious
Intention to buy gene-tech products	5-point scale (1 – 5); α  =  0.89; M = 2.25; sd = 1.22	+***	–*	+***
Buy GM fruit or vegetables?
Buy food with GM ingredients?
Concern about abuse of gene technology	5-point scale (1 – 5); α = 0.78; M = 3.18; sd = 1.07			+**
Insurance companies
Employers
Justice
Desired influence of actors	2-point scale (yes – no); M = 0.57; sd = 0.22	–**	–*	+**	+***
Doctors/dieticians
Scientists
Universities
Ethical committees
Patients associations
Me, as a citizen
Environmental organizations
Consumer organizations
Governmental organizations
Politics
Pharmaceutical industry
Food industry
Note: *p < 0.05; **p < 0.01; ***p < 0.001 (all two-tailed significant levels). M = mean item score.
^aGender: 0 = female; 1 = male. Age: 17–19; 20–39; 40–64; 65–79; 80+. Education: 1 = low level; 2 = medium level; 3 = high level. Having children: 0 = no; 1 = yes. Religious activity: 0 = no frequent attendance at church/temple/mosque; 1 = frequent attendance at church/temple/mosque.

Determinants

The perception of gene technology was measured with a series of statements relating to 11 developments in the gene technology field, and participants were asked how they evaluated them (see Table 2). The items were measured on a five-point scale (very positive – very negative). The items had a Cronbach's alpha of 0.80.

Table 2. Results for the predicting variables (n  =  1010), and correlation with demographic variables.

		Correlation with demographic variables^a
Construct	Description and items of scale	Gender (male)	Age	Education	Having children	Being religious
Perception of gene technology	5-point scale (1–5); α = 0.80; M = 3.08; sd = 0.64	–***		+**		+**
Human cloning
Obligatory genetic tests for insurance
Genetically changing animals for food production
The use of gene technology to simplify food production
GM crops to decrease environmental impact
Use of embryo tissue to research treatment of diseases
Storing DNA in databases
GM crops to counter food allergy
Modifying rice for dry areas
Use of gene technology against intestinal cancer
Mapping DNA to prevent diseases
Knowledge about gene technology	2-point scale (incorrect–correct answer); M = 0.65; sd = 0.25	+*	–***	+***		–*
The chimp is the animal species genetically closest to mankind						–**
If someone has an increased risk of a hereditary disease, his or her children will also have this risk
Only genetically modified bacteria have genes, whereas normal bacteria don't
Plants have some of the same genes as human beings
Trust in actors	5-point scale (1–5); α = 0.71; M = 3.16; sd = 0.60		–*
As an information source; family doctor/specialist
As an information source: scientists
As an information source: science/health documentaries on television
As an information source: news and current affairs programs on television
As an information source: newspapers
As an information source: government service announcements
Trust in compliance with legislation on genetic research
Trust in supervision by the government
Personal interest in gene technology	4-point scale (1–4); α = 0.68; M = 2.72; sd = 0.51			+***
DNA identification for forensic purposes
Diseases and their treatment
Genetic research and heredity
GM food
Genetic modification
Cloning
Experience with gene technology	2-point scale (yes–no); α = 0.64; M = 0.14; sd = 0.30
Have ever been offered genetic tests or genetic research
Have recently been offered a genetic test
Note: *p  <  0.05; **p < 0.01; ***p < 0.001 (all two-tailed significant levels). M = mean item score.
^aGender: 0 = female; 1 = male. Age: 17–19; 20–39; 40–64; 65–79; 80+. Education: 1 = low level; 2 = medium level; 3 = high level. Having children: 0 = no; 1 = yes. Religious activity: 0 = no frequent attendance at church/temple/mosque; 1 = frequent attendance at church/temple/mosque.

To measure the trust in actors, respondents were confronted with six actors relevant to decision-making in gene technology. They were asked who they would trust to give honest information about gene technology. All items were measured on a five-point scale (little trust – lots of trust). We also asked about the compliance with legislation on gene-research and supervision by the government. These eight items had a Cronbach's alpha of 0.71 (see Table 2).

Knowledge about gene technology was assessed with four items (answers either correct or incorrect) derived from a study by Gaskell et al. (2006). For each respondent, the number of correct answers was counted and divided by four, resulting in a single knowledge score per respondent (min 0, max 1). The four knowledge items are listed in Table 2.

Personal interest in gene technology was measured with six statements relating to interest in developments in gene technology. These items were measured on a four-point scale (not interested at all – very interested). The six items had a Cronbach's alpha of 0.68.

Experience with gene technology was assessed with two questions: “Have you been offered a genetic test at the moment, or do you know someone who has?” and “Have you, or someone you know, ever been offered genetic testing?” For each respondent, the number of “yes” answers was counted, resulting in a single experience score per respondent (min 0, max 2).

Finally, to assess their demographic background, we asked respondents their gender, age and educational level. Furthermore, we asked whether they had children and whether they were actively religious, as we expected reactions towards gene technology may be influenced by religious background or, for example in the case of prenatal screening and human cloning, having children.

Results

Reactions to gene technology

An overview of the results for the dependent variables is shown in Table 1. A small majority of the respondents did not intend to buy gene-tech products (M = 2.25, sd = 1.22). The respondents also had some concern about the abuse of gene technology (M = 3.18, sd = 1.07). When asked if actors should be able to influence decisions about research on genes and heredity, many respondents responded slightly above the mean (M = 0.57, sd = 0.22).

When we looked at the correlations among these three dependent variables and the demographic variables, we noticed that men, younger people, and people with higher levels of education are more inclined to buy GM foods than women, older people, and people with lower levels of education. People with higher levels of education are more concerned about the possible abuses than people with lower levels of education. Men and younger people are less inclined to grant influence to other actors to make decisions about gene technology, whereas people with higher levels of education and those with children are more inclined to do so (all r  <  0.25).

The determinants of reactions to gene technology

The perception of gene technology (M  =  3.08, sd  =  0.64) is multifaceted. Some gene technology applications are seen as negative by the majority of respondents, while others are seen as positive. Together they represent a balanced view of gene technology.

Table 2 indicates that the level of knowledge about gene technology (M  =  0.65, sd = 0.25) and trust in actors (M = 3.16, sd = 0.60) was slightly above the mean of the scale. The level of personal interest in gene technology was slightly below the mean of the scale (M = 2.72, sd = 0.51). Experience with gene technology was rather low (M = 0.14, sd = 0.30).

Looking at the correlations between the determinants for reactions to gene technology and the demographic variables (Table 2), we notice that men have a more positive perception of gene technology, and people with higher levels of education and those who are more religious have more negative perceptions. Generally, younger people, people with higher levels of education, and those who are less religious have more knowledge about gene technology. Older people trust the actors less, people with higher levels of education have more personal interest in gene technology, and women and younger people have more experience with gene technology (all r values are between 0.06 and 0.32).

Testing a generic model describing the reactions to gene technology

The model visualized in Figure 1 was tested using AMOS 5.0 (Byrne 2001).3 This model hypotheses that experience, interest, knowledge, the perception of gene technology and trust are determinants of the intention to buy gene-tech products, concern about abuse and desired influential actors. The parameters were estimated using the maximum-likelihood (ML) method. The analysis was carried out for scale means, and the measurement error was not included, meaning that the path analysis version of the SEM was used (Kline 2005).

The results show that the generic model provides a plausible explanation of the public's reaction to gene technology (RMSEA = 0.020; chi square = 17.98, df = 14, p = 0.21; CFI = 0.989; SRMR = 0.028).4 All paths were found to be significant and the signs of the path coefficients were all in the expected direction.

The variables in the model explained 27% of the variance in the intention to buy gene-tech products. The perception of gene technology was the main determinant (β = –0.45, t = 13.56). However, trust (β = 0.12, t = 3.67), knowledge (β = 0.11, t = 3.28) and personal interest (β = 0.09, t = 2.75) also explained significant portions of the variance.

The perception of gene technology was explained to a lesser degree (8%). There was one highly significant determinant: trust in actors (β = –0.26, t = 7.19). Personal interest in gene technology (β = 0.10, t = 2.66) also contributed significantly to the prediction of the perception of gene technology.

The explained variance in all other endogenous variables was less than 5%. For the concern about abuse, the explained variance amounted to 4%. Trust in actors (β = –0.12, t = –3.17), perception of gene technology (β = 0.11, t = 2.89) and personal interest in gene technology (β = 0.10, t = 2.62) were all found to be statistically significant determinants.

The explained variance in desired influential actors was also small (4%). Personal interest in gene technology (β = 0.16, t = 4.37) and trust in actors (β = 0.12, t = 3.37) were both highly significant determinants. The explained variance in the knowledge about gene technology (3%) and personal interest in gene technology (1%) were small. All hypothesized determinants were found to be statistically significant.

Discussion

The aim of this study was to identify some building blocks for a model of public reactions to applications of gene technology. Previous studies have focused on single-application models, but a more generic model describing the underlying determinants for reactions to various gene technologies has not yet been developed. Using the data available in this survey, we have made a first attempt at presenting such a generic model.

This model shows that it is possible to identify the determinants that lead individuals to form a perception related to gene technology and act accordingly. However, some limitations have to be mentioned. The quality of measurement is crucial in predictive analyses and theoretical modeling. In this study, the constructs were operationalized according to items used in prior research (IBT Marktonderzoek 2002). Although most scales have acceptable internal consistencies, the measuring instruments were not specifically designed for the path analyses we performed.

The path model gives a plausible explanation of the relationship between various dependent and independent variables in the model. All hypothesized paths were significant, and the signs of the path coefficients were as expected.

We found that experience and knowledge played a negligible role. Neither of these variables directly predicted the perception of gene technology. Experience predicted a small percentage of the variance in interest and knowledge, but it was not significantly related to any of the three reactions: buying intention, concern and desired influence. Knowledge, however, did significantly predict buying intention, but had no significant relationship with concern and desired influence. This is in accord with Bauer and Gutteling's (2006) findings. Personal interest is significantly related to all three dependent variables, with betas ranging from 0.09 to 0.16.

We observed more pronounced results for the role of the factor trust; trust does correlate with the perception of gene technology, as well as to all three dependent variables. The largest path coefficients were those predicting the buying of gene-tech products from the perception of gene technology (β = –0.45, t = 13.56), and predicting the perception of gene technology from trust in actors (β = –0.26, t = 7.19). The model can be summarized as follows: People develop a perception of gene technology based on their trust in relevant actors in the field, and this affects their intention to buy gene-tech products. These findings correspond to the literature reviewed earlier (Frewer et al. 1996, Peters et al. 1997, Slovic 2000, Hansen et al. 2003, Trumbo and McComas 2003).

An important question with the model we analyzed concerns the direction of causality between variables such as the perception of gene technology and concern about abuse. In the path analysis, the fact that a model fits the data does not mean that it is the only fitting model. The plausibility of a given model needs to be determined by evaluating other hypothetical models that might explain the data. We tested a model in which the concern about abuse predicted the perception of gene technology. This model fitted the data almost equally as well as the model in which the perception of gene technology predicted the concern about its abuse.5 We also tested a model in which the concern about abuse determined the interest in gene technology. This model also provided a plausible representation of the data.6

A final issue that needs to be addressed is whether a generic model for reactions to gene technology can be used as a framework to structure determinants for specific applications of gene technology. Are the results applicable to various fields of gene technology, or are they specific to particular applications? From this research we conclude that a generic model for reactions to gene technology can be stated. In this model, three separate reactions to gene technology are represented where the socio-psychological process can be structured in a general way. Further research should investigate if general models like these can in fact be applicable to the public response on other new technologies and other domains. Also, one would expect other determinants, like involvement, lifestyle, or media influence to play a substantial role in this process. Exploring the whole range of determinants in more detail will increase our understanding of the public reactions to the development of emerging technologies.

Acknowledgements

This article is based on research funded by the Centre for Society and Genomics, The Netherlands, a national centre funded by the Netherlands Genomics Initiative.

Notes

Note: ^a RMSEA = 0.020; chi square = 17.98, df = 14, p = 0.21; CFI  =  0.989; SRMR  =  0.028. ^# p ≤ 0.05; ^## p ≤ 0.001; ^### p extremely small (6.0 ≤ t).

The authors particularly refrained from identifying the trust determinant included in our model as either strictly cognitive or affective, but referred to the broad discussion on the role and character of trust (Frewer et al. 2003, Tanaka 2004).

Overall, no major differences for the variables reported in this study were found between 2002 and 2005 (IBT Marktonderzoek 2002).

The net response of completed and usable interviews for statistical analysis with Amos was 725. The dropout can be attributed to randomly missing data. List-wise deletion led to excluding 285 incomplete cases. The demographic make-up of the 725 is comparable to the 1010 participants.

Many different indices can be used to assess model fit. Following Kline (2005), we reported four indices: RMSEA, which has been recognized as one of the most informative criteria (Byrne 2001), chi square, CFI and SRMR. The following criteria were used: RMSEA: when RMSEA is below 0.08, the model fit is acceptable and when RMSEA is below 0.05, the fit is good; CFI: when CFI is higher than 0.90, the model is acceptable; SRMR: values of less than 0.10 are favorable; and chi square: when the significance level of chi square is larger than 0.05, the model is acceptable. It should be noted that there are problems attached to the use of chi square, and that the main reason for reporting the coefficient is because many other fit statistics depend on it (Kline 2005).

RMSEA = 0.020; chi square = 17.94, df = 14, p = 0.21; CFI = 0.989; SRMR = 0.028.

RMSEA = 0.024; chi square = 20.06, df = 14, p = 0.13; CFI = 0.983; SRMR = 0.029.