A few previous studies have specifically considered stakeholder attitudes towards RRI within United States biotechnology innovation systems. Two studies used interviews with academic researchers. Doezema and Guston (2018) interviewed United States biotech innovators within a single university research institute. In this study, RRI was framed according to areas of practice that could be put into place at the institute—that is, ethics, science education, open science, societal engagement, gender equality, and diversity. The study found that although ethics was of interest to biotech researchers at this institute, RRI was conceived as traditional “research ethics” such as reproducibility and misconduct, rather than according the Stilgoe et al. (2013) RRI principles. A second study (Glerup et al., 2017) interviewed eleven United States academics working in synthetic biology and nanotechnology using a Socio-Technical Integration Research (STIR) protocol where STS researchers embed themselves in laboratories (Fisher and Schuurbiers 2013). They found that the researchers thought of “responsibility” in more traditional ways, such as producing good science and taking care of employees, rather than broader obligations and responsivity to society (Glerup et al., 2017). Neither of these studies employed survey methods or empirically compared United States biotechnology stakeholder-group attitudes as we do in this article.

Other reports focus on attitudes of innovators in the EU, Canada, and United Kingdom towards RRI. Marris et al., 2015 identified engineers and natural scientists’ conceptions of RRI within the synthetic biology community in the United Kingdom. They found that seeking greater public acceptance of synthetic biology was the primary motivation for RRI. Similarly, Hartley et al. (2017) interviewed university researchers across STEM fields working in the United Kingdom and found that various actors espouse different meanings of RRI, although a predominant theme was the protection of scientific research from politics. For example, several researchers felt that public inclusion and engagement would best serve to increase public understanding of science and thus garner support for GM work (Hartley et al., 2017); rather than serve to give publics a “voice” in the conduct of GM work, like “inclusion” is meant to do in Stilgoe et al. (2013).

Along similar lines, Carrier and Gartzlaff, 2020 interviewed 80 researchers and research executives across the EU in a variety of technological fields across social and natural sciences, humanities, and engineering to investigate their understandings of RRI. While they found a welcoming attitude towards RRI in general, the interviewees were concerned about granting societal actors influence on the direction of research and innovation given public “ignorance and bias,” the additional expenditures that may be required to engage societal actors, and the potential loss of autonomy for science. Another study in Canada interviewed 31 people who design, develop and commercialize health innovations about practices of RRI (Rivard and Lehoux 2020). The study found that although innovators generally agreed on the desirability of several principles of RRI, they were concerned about the feasibility of meaningful implementation of them. The findings in these last two studies are consistent with our previous study with United States biotechnology stakeholders (Roberts et al., 2020), in which we found greater agreement on RRI principles (i.e., anticipation, responsivity, reflexivity, and inclusion) among diverse stakeholder groups than on specific RRI practices for implementing these principles.

RRI for biotechnology has also been investigated in multiple case studies using anthropological approaches to observation (see Macnaghten, 2016). For instance, Macnaghten, 2016 used ethnographic work to identify RRI attitudes towards GM food crops in Brazil, Mexico, and India as well as among symposium attendees in the United Kingdom and EU (Carro-Ripalda and Macnaghten 2015). These studies report the cultural, institutional, and social challenges to enacting RRI and provide greater identification of the practicality of enacting RRI within synthetic biology research and innovation (Macnaghten, 2016). In the context of GM crops, they found that in situations where the crop was not culturally significant to the country, like soybean in Brazil, scientists had “clear and unqualified optimism … on the role of GM crop technologies, with little evidence of a structured and sustained debate with wider society” (Carro-Ripalda and Macnaghten 2015; Macnaghten, 2016, p.284). In India, they heard from crop scientists who argued that India “could not afford the risk of falling behind in the development of biotechnology” and that anti-GM groups were “ignorant” (Carro-Ripalda and Macnaghten 2015, p 25). These results are similar to the barriers identified in our previous work with focus groups of United States biotechnology stakeholders (Roberts et al., 2020). Here we found 1) “cynicism” among innovators with regard to the public’s ability to engage in informed conversation and 2) the predominance of “academic capitalism” in United States culture and institutions, through which any process such as RRI that might slow innovation down would reduce competitiveness and be seen as undesirable.

All the studies mentioned above used qualitative methods of inquiry, and only a few focus on United States biotechnology innovation systems. In contrast, in this paper, we use quantitative surveys to investigate the relationships between demographic, sociographic, professional, or cultural factors and their influence on attitudes towards RRI. Furthermore, this paper breaks new ground by merging RRI scholarship and quantitative survey methods with policy process theory (namely the ACF), where we evaluate survey data with a larger sample (n = 110) of multi-sector stakeholders in United States biotechnology innovation. To our knowledge, our study is unique in these regards in the field of RRI scholarship.

A significant challenge faced by those attempting to legitimize and implement the RRI framework is establishing processes, strategies, and norms that create shared goals, while also facilitating coordination and cooperation between actors involved with innovation processes (Tait, 2017). ACF is a policy process theoretical framework that describes how actors engage in the policy process to translate their belief systems, which are simplified constructs used to make sense of the world, into public policy-making and action (Jenkins‐Smith et al., 2014). This paper uses the theoretical lens of the ACF framework to explore whether United States stakeholders share beliefs related to RRI and thus whether those beliefs may translate into shared policy action to implement RRI in United States biotechnology innovation systems.

The ACF provides a framework for understanding how coalitions of actors within a policy subsystem (e.g., biotechnology innovation) may interact to affect change or maintain the status quo (reviewed in Weible et al., 2009). The ACF provides a three–tier structure for describing the beliefs of actors (e.g., in this study, our biotech stakeholders) within a policy subsystem (e.g., in this study, biotechnology innovation) (Weible et al., 2009) (Figure 1). Deep core beliefs are the broadest category of beliefs, represent the most stable beliefs of actors, and are mainly normative. They transcend policy subsystems, or in other words, actors hold these beliefs across multiple policy areas. For example, deep core beliefs represent liberal and conservative political beliefs, whether responsibility for progress lies with individuals or communities, and beliefs about future generations. The next level in the ACF hierarchy of beliefs are policy core beliefs which are more moderate in scope and relate to the substance of particular policy subsystems (in our case, biotechnology innovation). These beliefs are thought to be important for forming more stable coalitions in policy subsystems, and although somewhat resistant to change, they are more malleable than deep core beliefs in response to new information and experiences. Policy core beliefs include how problems are defined, the ordering of priorities, and balance between values such as economics and ethics. At the most specific level of ACF beliefs are secondary beliefs (Figure 1). These are narrower in scope than policy core beliefs and are often formed in response to empirical information or experiences within the policy subsystem. Secondary beliefs are those related to specific public policy instruments used to achieve policy outcomes. The ACF predicts that secondary beliefs are the most changeable among coalitions and actors within them (Sabatier et al., 2007).

The advocacy coalitions that exist within a subsystem are aggregated groups of actors that coordinate to a non-trivial degree in the pursuit of policy change (Sabatier et al., 2007). Under ACF, a shared set of motivations and beliefs are what bind coalitions together. Deep core beliefs, and to a lesser extent policy core beliefs, are seen under the ACF to hold advocacy coalitions together. Applying these ideas to the biotechnology policy subsystem, in this paper we wanted to see whether attitudes towards RRI principles, which relate to the ACF’s level of policy core beliefs, and RRI practices, which relate to the ACF’s level of secondary beliefs, were influenced by deep core beliefs and whether stakeholder groups differed in their deep core, policy core, and secondary beliefs (Figure 1).

Given the importance of deep core beliefs in understanding the behavior of advocacy coalitions within a policy subsystem, ACF scholars have devised a way to conceptualize and measure deep core beliefs that captures their normative and ontological nature, while also being testable and generalizable (Jenkins‐Smith et al., 2014; Ripberger et al., 2014). These scholars used cultural theory (Douglas 1970; Douglas and Wildavsky 1982), to structure a scale for deep core beliefs that is generalizable across multiple policy subsystems. In this scale, twelve questions place survey respondents into four cultural archetypes--egalitarian, hierarchical, individualistic, or fatalistic (Ripberger et al., 2014). These cultural types, which are also considered “worldviews,” have been previously defined by the intersection of the dimensions grid and group (Douglas 1970; Douglas and Wildavsky 1982). The grid dimension is a measure of beliefs about how society should be structured. It represents the degree to which individual behavior should be regulated by group pressure and structural constraints. The group dimension is a measure of feelings of group membership within society; for example, a high measure of group indicates that individuals have a strong feeling of association with others. In this study, we use the Ripberger et al. (2014) validated scale to probe whether our United States biotechnology stakeholders fall into the four cultural archetypes, as a measurement of deep core beliefs according to the ACF (see Methods and Table 2).

In our previous study (Roberts et al., 2020), we found significant differences among stakeholder groups especially in their attitudes to secondary beliefs, or in other words, ways of implementing RRI in biotechnology innovation. Industry and trade organizations rated inclusion and responsivity practices, both which relinquish control to groups outside of biotechnology product development pipelines, less positively than government and advocacy groups. We found more agreement among stakeholder groups for the general principles of RRI, or policy core beliefs, applied to biotechnology innovation. As the ACF proposes that secondary beliefs are more flexible, and that shared policy core beliefs are important for stable coalition formation, we hypothesized that there are reasons to be optimistic about stakeholder groups in biotechnology innovation coming together to adopt RRI principles (which they share more agreement on) if better ways to implement them could be agreed upon (than those asked in the survey). However, in this prior study, we did not analyze whether deep core beliefs influence attitudes of biotechnology stakeholders towards RRI in the biotech innovation system. We also did not assess whether demographic (e.g., gender, race, age) or sociographic (e.g. income, education, experience in profession) factors influence RRI attitudes. We also did not address whether United States stakeholder groups share deep core beliefs and whether they are important for coming to agreement on RRI. Considering these gaps, this study addresses the following key research question:

We further describe our methods for addressing this key question below, followed by our results. Finally, we discuss the meaning of our results for the possibility of implementing RRI in United States biotechnology innovation systems.

We used a purposive sampling approach to recruit a diverse group of study-participants comprised of a variety of biotechnology professionals from different sectors. Participants were recruited from a sample of United States stakeholders in the greater Raleigh-Durham-Chapel Hill area (Research Triangle, NC). Recruitment was prioritized among professionals working in areas related to biotechnology in agriculture, food, or the environment. Many of the biotechnology developers recruited in the study work on emerging methods of genetic engineering, such as gene editing. The geographic region of the Research Triangle provides a host of diverse biotechnology and bioscience organizations. The Research Triangle is among the most active scientific and development regions in the United States and ranks second behind Boston Metropolitan region for life science expertise and development (Rose 2015). Participants reflect a fair representation of biotech stakeholders in this highly active biotech region. However, the participants do not comprise a statistical or geographic representation of the United States.

In our search we prioritized balanced representation from academe, advocacy groups (i.e., consumer and environmental groups), government, industry, and trade organizations (i.e., industry and conventional farming non-profit associations), (Figure 2). We chose to split non-governmental organizations (NGOs) into advocacy groups and trade organizations to better reflect the abilities of each group--where advocacy groups may influence policy change and biotechnology oversight through media communications and legal precedence while trade organizations often seek change through more direct lobbying efforts on behalf of their constituencies (Kuzma 2013). Government representatives were engaged in biotechnology arenas related to policy-making, regulatory analysis, and risk assessment, within the topics of agriculture, food, health and the environment.

Participants were first identified from databases and listserves collated by the Genetic Engineering and Society (GES) Center at NC State and included professionals working in the Research Triangle Area. We used this group along with other collaborators and key contacts in a snowball sampling strategy to provide more names to expand our sampling frame. We also sought out website databases for non-governmental and governmental organizations in the region whose work relates to agricultural and environmental biotechnology as the existing database was lacking in these areas. The final sampling frame including over 700 professionals. Participants for this study were first emailed by a research team member to introduce the project and outline the opportunity for participation as well and review their rights as participants following university IRB guidelines (Exempt, NC State IRB Protocol Approval #6157). A second email was sent to those who did not respond and targeted phone calls were made to fill stakeholder groups that were underrepresented in previous invitation phases. The final sample included 109 completed responses. Each participant was offered a $50 dollar gift certificate for their time although some participants accepted less money or none at all in accordance with their agency rules. All processes of this study followed the IRB agreement as exempted by the host university.

Table 1 reports the demographic composition of the participant groups. Age ranged from 21 to 70 years old (M = 50.23, SD = 11.76) and professionals held a mean length of experience in their profession of 15.65 years (SD = 10.73). 34% were female, 14% considered themselves non-white, 60% held doctoral degrees, and median household income ranged between $101,000 - $125,000. Figure 2 below also reports the sectors in which this group works. 100% of participants completed the agreed upon study and there was no apparent need for attention filter questions.

In order to assess our research question, data were collected using a pretested survey questionnaire (see Ripberger et al., 2014; Roberts et al., 2020). Survey questions were designed based on predominant elements of RRI (Stilgoe et al., 2013) and the application of cultural theory to test ACF deep core beliefs (Ripberger et al., 2014) as described above (Table 2). The survey was administered online using the Qualtrics platform. Participants were asked to complete the questionnaire which included the items reported on in this analysis (Table 2). The survey also elicited responses to open-ended qualitative questions relating to RRI, which we do not report in this paper. These qualitative results are being prepared for additional analyses.

Independent variables tested are listed in Table 1. Demographic independent variables (IVs) include age, gender, and race (white/non-white). Sociographic IVs included on the survey and tested were education and income. Professional sector variables tested were sector affiliation (Trade, Government, Advocacy, Academe, and Industry) within the biotechnology innovation system and years of experience.

ACF deep core beliefs were measured based on cultural theory and the previously reported and tested scale (Ripberger et al., 2014) (Table 2; Figure 1). This independent variable represents the personal and cultural-value orientation held by individuals. We used deep core beliefs (aka cultural worldview) as an independent variable to see whether they predicted RRI attitudes—either towards general RRI principles (policy core beliefs) or suggested RRI practices (secondary beliefs) (Table 2). We also subsequently assess if deep core beliefs align with certain United States stakeholder groups. ACF posits that deep core beliefs are important for tight policy coalitions and are the least malleable. Given the contentiousness of the agricultural biotechnology domain and our previous findings of differences in stakeholder attitudes towards RRI (Roberts et al., 2020), we expected to observe differences in deep core beliefs among stakeholder groups and that they would strongly influence attitudes towards RRI.

Four distinct deep core value-orientations were measured based on Ripberger et al. (2014): egalitarianism, fatalism, individualism and hierarchy (Figure 1). Note that the questions that Ripberger et al. (2014) and we use do not directly measure group and grid dimensions from cultural theory (Douglas 1970; Douglas and Wildavsky 1982) but rather use twelve questions to place people into these four cultural types. We measured the four worldviews by asking respondents to identify their level of agreement with three 7-point Likert-scale items (1 = “strongly disagree,” 7 = “strongly agree”). These were subsequently transformed into composite measures as below:

• Egalitarianism is the philosophical perspective which emphasizes equality and equal treatment of all people regardless of, religion, economic status, or political belief. People in this worldview seek strong group identities but prefer minimal prescriptions imposed from outside the group (high group, low grid). Consequently, they see value in more collective decision-making. The measures used to create this composite measure were: 1) What society needs is a fairness revolution to make the distribution of goods more equal, 2) Society works best if power is shared equally, and 3) It is our responsibility to reduce the differences in income between the rich and the poor (Ripberger et al., 2014). The three items were averaged to create a single composite measure of egalitarianism where a higher score indicated stronger identification of this philosophical belief (Table 1; M = 4.26, SD = 1.36; Cronbach’s α= 0.78).

For policy core beliefs and secondary beliefs about RRI, we developed, tested, and administered our own survey questions (Roberts et al., 2020; Table 2). Our policy core beliefs questions are based on the general principles of RRI (inclusion, anticipation, responsivity, and reflexivity) from Stilgoe et al. (2013), as they apply to biotechnology innovation as the policy subsystem (Roberts et al., 2020) (Table 2). For secondary beliefs, we designed questions to implement RRI principles according to ideas from Stilgoe et al. (2013). These represent specific policy practices that could be taken in biotechnology innovation to implement RRI (Table 2). The survey questions inTable 2 were pre-tested and used in a prior peer-reviewed study that investigated United States biotechnology stakeholder attitudes towards the four tenets of RRI (policy core beliefs) and ways to implement them (secondary beliefs) (Roberts et al., 2020).

Responses to the four questions of RRI policy-core beliefs or to the four questions of RRI secondary beliefs (Table 2) were compiled into two separate composite scores, and then each composite score was used as the dependent variable to examine whether demographic, sociographic, and professional factors, or deep-core beliefs from cultural theory (independent variables—Table 1) influenced attitudes about RRI principles or practices (Table 2). Responsible Innovation Policy-Core Belief Strength was measured using a composite from the four items (Table 2) (each item on 7-point Likert scale). These items were averaged to form the composite index, with higher scores indicating more agreement with the tenets of responsible innovation (Table 1; M = 5.53, SD = 0.92, Cronbach’s α = 0.69). Responsible Innovation secondary beliefs were similarly measured using four items on 7-point Likert scales (Table I). A composite index was created from these items where higher score indicates more agreement with the secondary belief statements (Table 1; M = 4.89, SD = 1.26, Cronbach’s α = 0.77).

We utilized ordinary least squares (OLS) hierarchical regression modeling, ANOVA, and Chi-squared analysis in SPSS software to test our research questions and explore relationships among the independent and dependent variables as discussed in the Results (Results).

Our study is limited in the number of participants (n = 110) and their geographical location as we sampled from stakeholders in United States biotechnology innovation located in the Research Triangle NC area (as discussed above in Recruitment and Survey Participants). Our study is also limited in the design of the survey questions (Table 2; Dependent Variables). All studies are limited by the choice and number of survey questions, and our study is no exception. When we designed the survey for RRI principles and practices (in 2016), to our knowledge, there were no survey instruments for assessing agreement with RRI in the literature. We focused on Stilgoe et al. (2013) in our survey design as it is the most highly cited paper when one searches for “responsible innovation” in Google Scholar. Specifically, we drew our questions from Stilgoe et al. (2013) textual descriptions of the four elements—anticipation, reflexivity, inclusion, and responsiveness--for the RI principle statements (aka policy core beliefs; Table 2) and from their “Indicative techniques and approaches” (Stilgoe et al., 2013, p. 1573) for the questions about secondary beliefs or practices (Table 2). We made particular choices for both sets of questions based on our understanding of the RRI literature and experience with the field of science and technology policy and RRI (see also Roberts et al., 2020 for discussion of this limitation). However, across the eight total RRI questions (Table 2), we feel the set captures the spirit and expression of RRI as articulated in Stilgoe et al. (2013).

Regardless, other choices could have been made for the survey questions. Thus, our results are constrained by the use of Stilgoe et al. (2013) to derive the principle questions and by our desire to use a reasonable set of questions for implementing RI. We mitigated this limitation by combining four questions into one construct for each dependent variable (8 questions total—4 for policy core beliefs or principles of RRI and four for secondary beliefs or practices of RRI; Table 2). We believe this to be a reasonable set for our novel, quantitative exploration of attitudes towards RRI and their relationships to demographic and cultural factors. In addition, the two 4-question constructs are the same ones used as dependent variables for testing all relationships with the independent variables of demographics, sociographics, affiliations and cultural (core) beliefs, so the comparative results about factors that influence RRI attitudes within this exploratory study are valid.

To investigate our research questions, we conducted two hierarchical ordinary least squares regression models. Specifically, we asked what demographic, sociographic, professional, and cultural-worldview (deep core beliefs from ACF) factors best explain the variance of observed stakeholder’s RRI 1) policy-core beliefs and 2) secondary beliefs? For the regression models, variables were entered into the model according to their assumed causal order by separate blocks and according to our research questions. Control variables (demographic variables) were included in the first block, whereas the second and third blocks were comprised of sociographic variables (educational level, household income) and professional variables (including sector affiliation and length of professional experience). The final fourth block incorporates deep core beliefs from ACF and cultural theory (Ripberger et al., 2014). As the IBM SPPS guide (2009) notes, this form of regression modeling adds these blocks in order to statistically control for the other variables, allowing researchers to evaluate the variables in concert with one another to note if “adding variables significantly improves a model’s ability to predict the criterion variable and/or to investigate a moderating effect of a variable.”

In Table 3 below, Model 1 pertains to responsible innovation policy-core beliefs (RRI principles) while Model 2 provides comparison among the same factors with regards to secondary beliefs (RRI practices). The factors tested in Model 1 accounted for 29.6% of the variance in RRI principle beliefs (policy core beliefs), whereas those factors tested in Model 2 explained over 43% of the variance in RRI practice beliefs (secondary beliefs) (Table 3).

ACF theory would predict that deep core beliefs influence attitudes towards policy-core beliefs, (in our case RRI principles applied to biotech innovation), and secondary beliefs (in our case RRI practices applied to biotech innovation). In fact, as predicted, in our hierarchical regression models, deep core beliefs significantly influenced both RRI principles (Model 1) and practices (Model 2). However, only one of the four worldviews, egalitarian views, was strongly and positively associated with stronger agreement with RRI principles (policy-core beliefs) (ß = 0.321, p = 0.006) and practices (secondary beliefs) (ß = 0.360, p = 0.001) while the other three worldviews, fatalism, individualism, and hierarchy, did not display prominent effects on RRI principles or practices. We further evaluate and discuss the role of egalitarian views vs. the other worldviews in subsequent tests below.

Professional sector-affiliation also showed significant correlation with attitudes towards RRI practices (secondary beliefs), although not with RRI principles (policy core beliefs). Participants affiliated with the biotech industry or industry-supportive trade organizations showed significantly lower agreement with the practices of RRI (secondary beliefs) (Table 3). Participants who work with trade organizations also held less favorable secondary beliefs and agreed less with RRI practices (ß =−0.662, p =0.05). However, the negative effect was even more pronounced for participants working with biotech companies (ß =−0.798, p = 0.041).

We also tested stakeholder group differences in RRI beliefs using ANOVA. ANOVA results confirmed the regression results, in that we did not find any significant differences among stakeholder groups regarding policy-core beliefs about RRI (RRI principles) (F = 1.64, p = 0.169), but there were significant differences among stakeholder groups with regard to secondary beliefs about RRI (RRI practices) (F = 6.39, p < 0.001). Industry and trade organizations held significantly more negative attitudes about RRI practices than government, academe, or advocacy groups according to ANOVA. The greatest magnitude difference was between advocacy and trade groups (mean difference = −1.36; p < 0.001) with the difference between advocacy and industry groups a close second (mean difference = −1.32; p < 0.0001). The lowest difference, yet still significant, was between academe and industry (mean difference = −0.83, p < 0.01). There were no significant differences between government, advocacy, and academic groups. These results are consistent with our prior results where a marked difference was found between two factions: industry + trade groups vs. government + advocacy groups, with academics affiliating with either faction depending on the specific facet of RRI (Roberts et al., 2020) (note: this previous work tested the four areas of RRI independently—i.e., inclusion, anticipation, responsivity, and reflexivity—and did not use a composite scale that combines all four like we do in this paper).

From the regression, we also found an association between years of professional experience and positive attitudes towards the principles of RRI (policy core beliefs) (ß = +0.223, p =0.07). However, this correlation was not significant for the practices of RRI (secondary beliefs).

In Figure 3, we present a visual model summarizing the regression results. In our model, the effect of sector affiliation on RRI secondary beliefs is about twice that of cultural beliefs (−0.798 for industry and −0.662 for trade, vs. +0.369 for egalitarian) (Figure 3). We then set out to examine the potential synergies between affiliation with biotech industry and deep core beliefs. Within industry-dominated sectors (industry or industry-supportive trade orgs), we wanted to investigate whether certain deep-core beliefs (i.e., non-egalitarian) could amplify negative RRI attitudes to explain these results. This would indicate both sectoral and cultural factors working together (i.e., not only what you belief, but also where you reside).

An association between a professional sector and certain cultural beliefs (deep core) could help to explain the strong effect on beliefs about RRI practices. If different stakeholder groups hold divergent deep-core beliefs, the ACF predicts that it would be more difficult to form stable policy coalitions transcending stakeholder groups (Jenkins‐Smith et al., 2014; Ripberger et al., 2014). In the context of our work, coalitions of stakeholders from different sectors would be more difficult to form and maintain in order to implement RRI if those stakeholders held different cultural world views. To evaluate if different stakeholder groups hold distinct cultural beliefs, we first did descriptive statistics on deep core beliefs (cultural beliefs) by sector (Table 4; Figure 4). We added up survey responses for each participant from the three 1-7 Likert Scale questions for each cultural archetype (Table 1; Methods)—Hierarchical (H), Individualistic (I), Egalitarian (E), Fatalist (F). We then averaged the results for each stakeholder group. The highest possible score would be 21 for each group average (all 7’s for the three questions), with the lowest as 3 (all 1s for the three questions). Results are shown in Table 4; Figure 4. This enables us to evaluate the relationship between these independent variables and also answer if sector affiliation and deep-core beliefs work together to influence attitudes about RRI practices (secondary beliefs). Given the significant positive relationship between RRI attitudes and egalitarian beliefs and the significant negative relationship between RRI attitudes and industry or trade stakeholder-group affiliation, we wanted to see whether participants from industry and trade organizations scored lower in the egalitarian belief scale.

Although all stakeholder groups scored relatively high on the egalitarian scale (all over 11 points), industry and trade organizations had the two lowest scores (M = 11.7 and M = 11.7). The highest score for trade organizations was in the individualistic category of deep core beliefs (M = 12.1). Individualists tend to hold free-market world views consistent with support of private industry and individual competition in innovation systems (Jenkins‐Smith et al., 2014; Ripberger et al., 2014), and the trade representatives came from groups supportive of biotech industry innovation. In contrast, advocacy and government groups held the strongest egalitarian views (Table 4; Figure 4) (M = 13.8 and M = 14.1 respectively). Thus, according to descriptive statistics, there appeared to be an association between stakeholder group and deep core beliefs that could explain the strong effect on RRI secondary beliefs (Figure 3).

We then set out to see if these group differences in cultural world views (deep core beliefs) were statistically significant using Chi-squared analysis. However, we found no statistical differences among stakeholder groups despite the differences we observed in the descriptive statistics (Figure 4; Table 4). This could likely be due to inadequate statistical power given the low cell size within the cross-tabulated grouping (e.g., n = 16–30) (Figure 2). Future work could be done with a higher number of participants in each stakeholder group to see if the lower egalitarianism scores we saw among our trade and industry groups is applicable in the wider United States biotech innovation system.

Next, we wanted to determine if other deep core beliefs (cultural beliefs) tended to be associated with attitudes towards RRI. Our regression model showed a strong correlation between egalitarian beliefs and positive attitudes towards RRI practices and principles. But what about the other cultural archetypes? They did not show significance in the regression model, but other variables could have masked their effect. To test if other archetypes of cultural or deep core beliefs relate to RRI attitudes, we split each participant into either a high or low score (relative to the mean) for each of the four cultural archetypes (H, E, I, F). We then conducted two-tailed independent Samples t-tests to determine if these groups differed in their attitudes towards policy core beliefs (RRI principles) or secondary beliefs (RRI practices). In other words, do individuals who are high or low on each cultural deep-core value scale hold distinct policy core or secondary beliefs?

For the policy core beliefs (RRI principles) we found significant relationships with deep-core beliefs for all the cultural archetypes except for fatalism (Table 5). For RRI secondary beliefs, we found significant relationships for all of the cultural archetypes (Table 5). Egalitarian beliefs continued to have a strong association with positive attitudes towards RRI principles and practices, as confirmed by the regression analysis (Table 3; Figure 3) and the sample t-test (Table 5). However, we were also able to uncover through the t-tests that fatalistic beliefs positively influenced beliefs about RRI practices (secondary beliefs) albeit to a lower extent. We were also able to uncover that hierarchical and individualistic beliefs had a statistically-significant negative influence on beliefs about both RRI practices (secondary beliefs) and principles (policy core beliefs) (Table 5). We note that this influence is stronger for individualistic world views than for hierarchical world views.

To summarize the results, we expand on our previous model from the regression alone (Figure 3) to incorporate the descriptive statistics, ANOVA, and Sample t-test results (Figure 5). We present Figure 5 as a hypothesis-generating exercise, recognizing that our limited study does not confirm these relationships. Larger numbers of stakeholders across wider geographic regions will be needed to do so. Nevertheless, we present novel findings of these empirical relationships that can provide insights into the formulation of attitudes towards RRI policies and practices in United States biotechnology innovation systems. In closing, we now turn to a broader discussion of the model (Figure 5) and its potential implications for building cross-stakeholder coalitions to advocate for and implement RRI principles and practices.