Public perceptions of the US innovation system: moderate support but compelling need for reform

Abstract

Science and innovation policy in the USA often frame publics as the beneficiaries of new technologies, but little research has yet engaged publics on their views of the innovation system (IS)—the combined efforts of government, industry, and universities to produce and promote new technologies. Based on a national public survey (n = 3,010), we identify three dimensions of public judgments about the IS with public policy implications: (1) US publics hold moderate confidence in the IS to produce benefits for them and to respond to public input; (2) they are slightly more critical of innovation-related environmental harm and the accrual of benefits to large corporations; and (3) they strongly support reforms to ensure safe, responsible, and affordable technological innovation. Multivariate regressions indicate variance of judgments by social location and worldviews, finding equity and justice aspects particularly salient in views on the IS. We discuss implications for innovation policy.

The USA has long seen its science and technology innovation system (IS) as the key engine for economic development and geopolitical competitiveness (Chang 1994; Appelbaum, Parker, and Cao 2011; Appelbaum et al. 2018; Andreoni and Chang 2019). By IS in this work, we mean the system of people, regulations, and organizations that combine efforts to develop and promote new technologies. Despite their intended benefit, diverse publics have been notably absent from key national conversations about investment in and rollout of new technologies. Thus, publics are arguably positioned in this system as what sociologist Adele Clarke refers to as “implicated actors,” those who are involved in and affected by a social “situation” but lack meaningful presence or agency (Clarke 2005). The adoption of more inclusive responsible innovation governance frameworks elsewhere aims to remedy the lack of meaningful public agency in scientific innovation by incorporating components of societal ethics in new technology research and development across the innovation cycle, but the USA has not yet implemented such a uniform and comprehensive framework.

In Europe, the European Commission has promulgated since 2011 a policy for “better management of the relationship between science and society” called responsible research and innovation (RRI) (Declich, Berliri, and Alfonsi 2022). RRI is a normative framework that stipulates an IS that both anticipates current and future downstream societal effects of new research and technology and prescribes meaningful engagement with those who may be affected (Owen, Macnaghten, and Stilgoe 2012; Owen, Bessant, and Heintz 2013; Stilgoe, Watson, and Kuo 2013). It specifically calls for an IS that is anticipatory about its downstream effects, participatory for all potentially affected parties, reflective about its own processes and effects, and responsive to such inputs in meaningful ways (Owen, Bessant, and Heintz 2013; Stilgoe, Watson, and Kuo 2013) and has been incorporated programmatically as a normative framework across the European Union (EU) Research Horizons. A number of recent studies assess the varying effectiveness and impacts of the framework in practice (Stahl et al. 2021; Declich, Berliri, and Alfonsi 2022; Michali and Eleftherakis 2022; González-Esteban, Feenstra, and Camarinha-Matos 2023) and the implications for particular research fields, such as gene editing (Nawaz and Kandlikar 2021). The USA has thus far used a more diffuse and selective approach of ethical, legal, and social implications research in a number of technological development streams, beginning with the Human Genome Program (Fisher 2005; Dolan, Lee, and Cho 2022) and reaching its peak during the first decade and a half of the National Nanotechnology Initiative (2000–15). Concern about science and technology in society is currently strongly on the rise again in the USA, particularly in reference to balancing the economic benefits and social risks of artificial intelligence (AI) (Zhang 2023) and other emergent technologies (President’s Council of Advisors on Science and Technology 2023).

Throughout these approaches, the direct engagement of US publics in the IS has most often been portrayed as a potential barrier or threat to successful innovation by government, industry, and scientists and engineers alike (Pidgeon, Kasperson, and Slovic 2003; Cormick 2009; Seifert and Fautz 2021). Although publics are consistently framed as the intended beneficiaries of the IS, they are also viewed as potential obstacles to overcome and often assumed by science and engineering elites to be potentially hostile to new innovations, through ignorance of the science or values in opposition such as religious views or political ideologies (Motta 2019; Kunovich 2023). Given this apprehension, there is surprisingly little evidence-based research on US public perceptions of the IS itself and its capacity to deliver public benefits. The best available proxies for such research are extant studies on public attitudes in Europe and the USA toward science and technology more broadly.

In the USA, the National Science Foundation “Science and Engineering Indicators” national survey most recently (2022) found that a majority saw more benefits than harms from science, that they support federally funded research, and that confidence in the “scientific community” is quite high relative to views of other institutions, such as congress, corporations, and the media (Besley and Hill 2020; Southwell and White 2022). Even so, public trust and confidence in scientists have dwindled over time, particularly following the COVID-19 pandemic (Kennedy, Tyson, and Funk 2022; Kennedy and Tyson 2023). Additionally, US public opinion diverges from that of scientists in some key respects. Overall, publics tend to be more concerned than scientists about specific technological applications, including nuclear energy and genetic modification, and about industrial pollution and pesticide use (Funk 2015; Besley and Hill 2020).

Across the Atlantic, the official EU public survey panel, the “Eurobarometer,” found in 2021 that large majorities of Europeans in over thirty-eight countries believe that “science and technology” has a positive overall influence on society and view certain technological applications such as energy or medical applications as particularly beneficial (European Commission 2021). At the same time, a majority of surveyed EU publics were reported to think that science and technology mostly benefits wealthier individuals, that governments should ensure that new technologies benefit everyone, and that lay publics should be included in the innovation process to ensure that societal needs and values are met (European Commission 2021), suggesting high public demand for responsible innovation.

However, public attitudes about science and technology and about scientists are arguably distinct from views on the complex national investment and IS through which technologies are produced. For one thing, the IS, as we have defined it, takes an organizational and institutional approach to understanding how science and technology is actually developed (Appelbaum et al. 2018). Thus, we anticipated that there may be differences between the way publics judge science and technology generally versus how it is developed through the nexus of university, industry, and government action. In our view, the IS as a concept also evokes scientific and industrial policy and governance frameworks because as a system it includes regulations that structure how innovation takes place. Publics may see science and technology as a positive (or negative) force in society, particularly with a focus on its products, yet still retain distinct perspectives on how the IS works to develop and produce science and technology.

This research thus set out to examine public perceptions of different aspects of the IS as a whole. Because no prior research that we have been able to identify has yet evaluated such public attitudes, we constructed a new set of measures designed to do this (see the Methods section for more details). Drawing from mental model approaches (Morgan et al. 2001), we anticipated that respondents who were unfamiliar might form judgments on the IS that are anchored in and hence mirror their views on particular technologies (Tversky and Kahneman 1974; Furnham and Boo 2011) and that affective responses may influence their judgments (Loewenstein et al. 2001). Risk perception and public attitude research demonstrate that, in general, individuals in the USA tend toward “techno-optimism” in their judgments regarding technological benefits, reflecting what has recently been described as “an unwavering belief in the power of technology” (Smith 2014; Hochschild and Sen 2015; O’Mara 2019). We note that there are significant caveats to this generalization, for example, that Americans are notably pessimistic about some particular technologies (e.g. nuclear power; Slovic 1987), and that trust has been shown to profoundly mediate risk and benefit perception (Siegrist 2000). Nonetheless, we hypothesized that this optimistic orientation toward technological benefits would translate into more positive assessments of the IS itself. At the same time, past research has shown that when publics are presented with specific scenarios and applications for given technologies, even those technologies judged positively overall, they tend to garner more ambivalence and concerns about specific risks and harms (Pidgeon et al. 2005, 2009; Satterfield et al. 2009, 2013; Harthorn, Shearer and Rogers 2011; Harthorn, Satterfield, and Pidgeon 2021). Does such specificity also affect perceptions of the potential risks and harms of the broader IS? We anticipated that US publics, despite general techno-optimistic tendencies, would form concerned judgments around specific risks and harms of the IS (i.e. they would reflect such ambivalent views).

Furthermore, we are interested in contributing to new understanding of the diversity of views held by US publics on key aspects of the IS. This set of interests is motivated by a commitment to participatory democracy (Pateman 2012; OECD 2020) and the need for meaningful engagement by science and IS actors with the broad reaches of society, as well as furthering the aims of RRI and what Rodrik and Stantcheva (2021) have referred to as “inclusive prosperity.” In order to attend to such distinct perspectives, we examine how perceptions of the IS differentiate across social location, such as respondents’ racialized, gendered, and classed identities, age, and individual worldviews and values (for in-depth discussion of hypothesized relationships, see Supplementary data). The IS definition we provided respondents was intentionally very broad, including people, regulations, and organizations that develop and market new technological innovations (see the Methods section for exact wording).

This study thus aims to respond to the absence of research on public views on the IS by contributing new social survey research with a large representative sample (n = 3,010) that provides empirical evidence about how diverse US publics view aspects of the IS. Our analyses identify three dimensions that broadly characterize respondent views on the IS: confidence, critique, and need for reform. Results show that, on average, US publics hold tempered but positive confidence in the IS to produce technological and economic benefits for the public and to respond to public input, while at the same time, they are slightly more critical of innovation-related environmental harm and the accrual of IS benefits that they see as mainly going to large corporations. Importantly, our results demonstrate that Americans strongly and unambiguously support reforms to safety regulations and product affordability. Furthermore, this research illuminates the social contours of these views in some surprising ways: identities and worldviews that are commonly believed to permeate almost every social perception, such as race, political ideology, and religion, are here found to be either not as strongly associated or not in the direction expected. On the other hand, we found that concerns around equity and justice, particularly environmental justice, were, on average, strongly associated with public perceptions of the need for reform to the IS.

1. Methods

1.1 Survey design and sampling

In this research, we designed an original US nationally representative survey on public views concerning the IS, among a number of other technological risk and benefit perception topics covered. Survey design began in 2020, drawing on qualitative interview data from a preceding study by the same team. This research was conducted after approval from the University of California, Santa Barbara Human Subjects Committee (Institutional Review Board) in 2021. We contracted with the survey research firm Qualtrics to recruit our sample. Data collection began and ended in summer 2021. We note that this period was likely influenced by the COVID-19 pandemic. Qualtrics screened respondents for English proficiency and a minimum age of 18 years, and they compensated participants with USD $6 for a complete response.

There was a total of 7,183 recorded responses, of which 3,095 were complete quality responses (attrition rate = 57 per cent). This was an online nonprobability sample, which can potentially limit the participation of some segments of the population, including older, low-income, and rural individuals who may have limited access to the internet. In order to address this, we took several steps to ensure a representative sample, allowing us to make inferences to the US population. First, we quota sampled age, gender, and region and oversampled Black and Latina/o respondents due to our interest in issues of social justice and inequality in technology innovation. According to the American Association for Public Opinion Research report on nonprobability samples (Baker et al. 2013), quota sampling can help reduce bias in estimates. Based on quota sampling alone, our sample closely matches US population parameters on age, gender, and region, while race is slightly skewed toward people of color due to intentional oversampling.

We weighted our sample using iterative proportional fitting to bring it further in line with population parameters. Utilizing auxiliary data from the American Community Survey 2019 5-year estimates, we weighted our sample to US population totals in gender, age, region, race, Latina/o ethnicity, and educational attainment (our proxy for socio-economic status). Due to missing data in race and region and because some respondents did not fit binary gender categories used in the auxiliary data, eighty-five respondents were dropped from the sample because they could not be weighted. Our final analytic sample size is 3,010 (Harthorn and Budge, n.d.).

1.2 Outcome variables

The survey measures were constructed through an iterative and interpretive process. Public views on the IS reported here were measured using an eight-item question set (see Supplementary Table S1 in Supplementary data for measures of central tendency and spread and Supplementary Fig. S1 in Supplementary data for survey question wording). These questions were based in part on knowledge gained in past research in the Center for Nanotechnology in Society at the University of California, Santa Barbara (UCSB) on the global nanotechnology IS (Appelbaum et al. 2018). However, the questions were also partly experimental and novel. We provided all our respondents the following definition just prior to the IS question set. This definition was first vetted with the project’s science and engineering collaborators. You can think of the innovation system as the system of people, regulations, and organizations, such as universities, industries, and the government, that combine their efforts to develop and promote new technologies, such as synthetic cells.

The question set by design focused on a diverse series of issues, including opportunities for public input, product affordability, public benefits, scientific authority, and regulations. Item order was randomized for respondents. The question set was presented toward the end of a longer survey that also gauged attitudes on science and technology, views of the risks and benefits of developing bottom-up synthetic cells, and attitudes about technology governance. We provided respondents a four-point agreement scale, from strongly disagree to strongly agree, with no midpoint by design, in order to encourage respondents to lean one way or another.

1.3 Explanatory variables

We utilize a range of explanatory variables in our analyses (see Supplementary Table S1). These include sociodemographic variables of age (range 18–92 years), gender (0 = male, 1 = female), and self-reported race/ethnicity (coded as Non-Latino/a White, Asian or Asian American, Black or African American, Latino/a any race, Native American-Hawaiian-Alaskan-Pacific Islander, and two or more races). We also include education (coded as high school or less, some college or associate, and bachelor’s or higher), political ideology (1–7 scale, 1 = more liberal, 7 = more conservative), and religiosity (0 = not religiously affiliated, 1 = religiously affiliated).

We utilize three personal views in our multivariate regression analyses: life satisfaction (0 = not satisfied, 1 = satisfied), agreement that “I often feel discriminated against” (0 = disagree, 1 = agree), and agreement that “I have very little control over risks to my health” (0 = disagree, 1 = agree).

We also dichotomously coded our environmental justice variables (0 = disagree, 1 = agree) measuring agreement with the following statements: (1) “I think hazardous facilities are more common in minority communities,” (2) “Minority communities lack the political clout to stop hazardous facilities from being located near them,” and (3) “The government should restrict the placing of hazardous facilities in minority communities” (Satterfield, Mertz, and Slovic 2004; Conti, Satterfield, and Harthorn 2011).

We then employ several measures of respondent views on science and technology generally and nonspecific to the IS. We measure dichotomous agreement or disagreement with the following statements: (1) “Scientists are unbiased”; (2) “New scientific discoveries are good for everyone, not just the rich”; and (3) “I trust my own observations more than I trust scientific information” (0 = disagree, 1 = agree). We also examined respondent attitudes toward five biotechnologies: synthetic biology, synthetic cells, animal cloning, genetically modified crops, and laboratory-grown meat. We asked respondents to rate how acceptable or unacceptable the risks these technologies pose to humans are on a scale of 1–5, where 5 is very acceptable and 1 is very unacceptable. We then created an additive index of acceptability of all five measures to use in our analyses. This index is scaled to measure from 0 (least accepting of the risks of these biotechnologies) to 20 (most accepting) (rescaled to begin with 0 rather than 5).

1.4 Multiple imputations

Because “Don’t Know” responses are generally treated as missing data in analyses such as multivariate regression, we employed multiple imputations for our regression models. Although the two outcome variables have low rates of “Don’t Know” responses (<7 per cent), when combined with many more variables in multivariate regression, the percent of “Don’t Know” responses can reach almost 50 per cent. The “Don’t Know” responses can be modeled using observable characteristics measured in the survey, suggesting that the pattern is missing at random. We include these observable characteristics in the imputation model to estimate response values for these respondents. We performed ten imputations with chained equations (Multiple Imputation with chained equations or MICE) in Stata 16.1 with 100 burn-in iterations (University of California, Los Angeles: Statistical Consulting Group n.d.). MICE allows us to estimate different imputation models for variables with different response outcomes (e.g. dichotomous versus continuous variables). We imputed nineteen variables with missing data. These include all eight IS items, views on biotechnologies, political ideology, views on science and scientists, environmental justice views, and feelings of discrimination and powerlessness (see Supplementary Table S1 for unimputed descriptive statistics).

These imputations are partially predicted using demographic characteristics and relevant attitudes that have no missing data. These are race, Latino/a ethnicity, gender, age, region, education, income, religious affiliation, life satisfaction, risk versus benefit of synthetic cells, and comfort with synthetic cell development. We used sampling weights in the imputation model. We inspected the diagnostic trace plots (see Supplementary Fig. S3) and concluded that there was no discernable trend in the summaries of the imputed values, suggesting that the imputation model is adequate. The imputed datasets are used only in the regression analyses where the missing data problem is otherwise significant. We use Ordinary Least Squares (OLS) regression with the multiply imputed data because the outcome variables have been imputed continuously on a 1–4 scale.

1.5 Analytic strategy

As we do not have a priori assumptions about how the IS items relate to each other, we utilize principal components analysis (PCA) to analyze the IS variables by reducing the number of dimensions in our eight-item question set (varimax rotation for maximum interpretation). This data-driven process identifies underlying data patterns that capture variation in responses, demonstrating how the eight items inter-relate. We believe that the PCA is essential for identifying the three core concepts we discuss—“support,” “critique,” and “reform.” We interpret the components, describe what underlying dimensions they measure, and descriptively compare them. Based on the PCA identifying the two “reform” items and the descriptive analysis demonstrating they have the highest agreement, we then employ nested OLS multivariate regression to analyze the relationships between respondent views on safety and affordability reforms and explanatory variables, such as social location and underlying beliefs. For information on measures of central tendency and spread in these variables, see Supplementary Table S1 in Supplementary data.

2. Results

2.1 PCA

The PCA results in three components that meet our criteria for acceptance (Kaiser criterion where eigenvalues > 1; Kaiser-Meyer-Olkin test value = 0.71; see Supplementary Fig. S4 in Supplementary data for the scree plot of eigenvalues). The three components collectively explain 62 per cent of the variance among respondents in the IS question set, which is considered a strong finding for survey research of this kind. We now qualitatively interpret how items from the question set load onto each component.

Component 1 reflects support for the idea that the IS takes into account public input and concerns, that it leads to more jobs for people, that the public benefits from government investment in scientific discoveries, and that scientists should be the ones to decide what kind of research to pursue (loadings > 0.3; see Fig. 1). Based on this data-driven finding, we describe this component as measuring an underlying dimension of “support” (for) or “confidence” (in) the US IS. In particular, higher scores on this measure reflect stronger confidence that the US IS produces benefits for Americans (e.g. new scientific discoveries and job creation), that it is responsive to the public, and that they trust scientists to direct the research, while lower scores reflect less confidence in these.

Component 2 has the strongest association with agreement that there should be better regulations to ensure safe development of new technology and that drugs developed with taxpayer funds should be affordable (loadings > 0.3; see Fig. 2). There is also a smaller but potentially important association with two other items: agreement that the IS mainly benefits large corporations and disagreement that it takes into account public input and concerns (loadings < 0.3). This suggests that this component measures agreement with wanting the US IS to be more responsive and fair and thus agreement with “need for reforms” to the IS.

Component 3 is most strongly associated with negative views of the IS: that it mainly benefits large corporations and that it causes too much environmental harm (loadings > 0.3; see Fig. 3). We believe that this component measures an underlying dimension of “critiques” of the IS, where higher scores are associated with more critical views of the IS and agreement that it produces social and environmental harms.

In order to compare public responses to these different constructs, we create three composite variables measuring agreement with support/confidence in the IS (four items), need for reforms to the IS (two items), and critiques of the IS (two items). Normally, a component includes three or more items, but we record these here as they capture discrete ideas—support, critique, and reform—and explain such a large total proportion of variance that we deemed these sufficient for analysis. We additively combine and rescale these three composite variables to a 1–7 scale, where 1 is least agreement, 7 is most agreement, and 4 is a midpoint.

2.2 Descriptive analyses

Support for the IS has an average agreement of 4.6 [µ = 4.61, df = 2,045, n = 2,046, 95 per cent confidence interval or CI (4.56, 4.68)], only slightly above the midpoint of 4 (see Fig. 4). This suggests that US publics, on average, have a positive view of the IS, but that this support is lukewarm. This positive view supports our expectation that US publics are, in general, techno-optimistic and will largely judge the science and technology IS as delivering benefits. Alongside this weak confidence, there is slightly higher agreement with critiques of the IS [µ = 4.79, df = 2,170, n = 2,171, 95 per cent CI (4.73, 4.86)], which suggests that US publics have concerns about how specific benefits and harms are distributed. In contrast to both these, the highest agreement among US respondents is for reforms to the IS, which has an average agreement of 5.8 [µ = 5.84, df = 2,760, n = 2,761, 95 per cent CI (5.79, 5.89)].

This need for reform variable is composed of two items from the question set: that drugs developed with taxpayer funds should be affordable and that we should enact better regulations to ensure safe development of new technologies. These two items individually have the highest level of agreement among the eight items, demonstrating strong support for affordability and regulations (see Fig. 5). We also see that agreement that public input is taken into account is the lowest averaging item in this set of eight. We further contextualize these results by examining the “Don’t Know” responses. The two reform items have the lowest “Don’t Know” response rates, demonstrating that they have not only the highest level of agreement but also the lowest level of uncertainty about their answers (see Fig. 6).

2.3 Multivariate regression

Because the two reform variables have the highest agreement and lowest uncertainty (see Figs 5 and 6), we employ multivariate regression to understand the diversity of views underlying this finding. We are motivated to understand who is more or less likely to support these reforms, as this has important implications for science policy. For example, does everyone support these reforms? Can we “explain” support by certain groups through underlying perceptions of related topics, such as views on life, environmental justice, science, and technology?

The full regression results for our two outcome variables are presented in Tables 1 and 2. We provide an overview of the main findings here. We use two-tailed t-tests at a significance level of 0.05 in the following results, but we also describe results that fall between alpha levels of 0.05 and 0.10. In Table 1, we examine the outcome variable measuring agreement with better safety regulations. Contrary to our expectations, we do not find evidence that gender, race, or educational attainment is associated with support for regulatory reform (Model 1). Interestingly, we do find limited evidence that Black and Latino/a respondents are less supportive, on average, of better safety regulations compared to White respondents after controlling for environmental justice views (Model 3). The coefficient sizes are not large, and the P-values are close but not below .05 (Black respondents β = −0.08, P = .083; Latino/a respondent β = −0.09, P = .069). This sensitivity to the addition of environmental justice views suggests that some of the support for better safety regulations among Black and Latino/a respondents may be partially explained by environmental justice views.

Age is significantly positively associated with the dependent variable, meaning that older respondents are more supportive of better safety regulations for new technologies (β = 0.003, P = .001). This coefficient suggests that a 75-year-old would, on average, score 0.15 higher than a 25-year-old. However, this finding diminishes in subsequent models with the addition of environmental justice (EJ) views, again suggesting that EJ views may partially explain the association between age and support for regulatory reform.

Somewhat surprisingly, respondents who identify with a religious group appear to not be statistically distinguishable from nonreligious respondents (i.e. atheists, agnostics, and “nothing in particulars” who together account for about 29 per cent of our sample). Political ideology is initially significantly associated with the outcome variable, such that more conservative respondents agree less with need for regulations (Model 1, β = −0.03, P < .001). The difference between the most liberal and most conservative respondents would, on average, translate into 0.18 lower agreement for conservatives. However, the addition of environmental justice views in Model 3 eliminates the statistical significance of political ideology, suggesting that environmental justice views may “explain” the association between political ideology and support for safety regulations or that it acts as a proxy for some other unmeasured variable(s).

Some environmental justice views themselves have robust relationships with support for regulatory reform. For instance, both belief in the existence of power asymmetries (“Minority communities lack the political clout…”) and support for government intervention on behalf of environmentally vulnerable communities (“The government should restrict…”) are statistically and substantially significant (Model 3). It is worth examining the coefficient sizes for these two variables. Belief in power asymmetries (β = 0.18, P < .001) and support for governmental intervention (β = 0.33, p < 0.001) are substantial effects for support for better safety regulations as these are measured on a four-point scale. If respondents agree with both statements, this is essentially a half point higher.

The results in Table 1 are also notable for the lack of statistically significant results for some expected relationships with support for better safety regulations. In addition to those described earlier, we found no credible evidence that variables measuring life satisfaction, feelings of discrimination, and feelings of powerlessness correlate with support for safety regulations (Model 2), similar to views on science or biotechnologies (Models 4 and 5).

In our next series of regression models, we examine agreement that drugs developed with taxpayer funds should be affordable (Table 2). We do not find evidence that gender is significantly related to support for affordability, but we do for race and educational attainment (Model 1). Specifically, Asian (β = −0.13, P = .009) and Black (β = −0.22, P < 0.001) respondents are both less supportive of affordability compared to White respondents, on average. There is weaker evidence that Latino/a respondents are also less supportive; however, the coefficient size is relatively small and P-value falls in between alpha levels 0.05 and 0.10 (β = −0.07, P = .087). Descriptive analyses show that all racialized groups largely support drug affordability, but that White respondents are particularly supportive. We also find that higher educational attainment is linked to more support for affordable taxpayer-funded drugs (some college or associate’s β = 0.09, P < .007; bachelor’s or higher β = 0.12, P < .001). These findings for race and educational attainment are robust for Models 1–5, demonstrating that their relationships with affordability are not explainable by other variables we include.

Age is significantly associated, where older respondents are more supportive of affordable drugs (β = 0.01, P < .001). Using the example from before, this coefficient suggests that a 75-year-old respondent would score 0.5 higher on average than a 25-year-old respondent. This is a half point higher agreement that taxpayer-funded drugs should be affordable. Religious affiliation is negatively associated with affordability, meaning that religiously affiliated individuals are less supportive compared to nonreligious individuals (β = −0.12, P < .001). We also do not find evidence that political ideology is related to support for affordable drugs (Model 1). This is a surprising finding and contrary to our expectations. On the other hand, we find that life satisfaction (β = −0.09, P = .003) and feelings of powerlessness (β = −0.07, P = .025) are significantly negatively related to support for affordable taxpayer-funded drugs (Model 2).

Environmental justice views are also important correlates of support for affordability. In particular, support for government intervention (Model 3, β = 0.21, P < .001) and belief in power asymmetries (β = 0.10, P = .017) are linked to greater support for affordable taxpayer-funded drugs. There is also weaker evidence that belief in environmental racism is linked to support (β = 0.09, P = .055). Surprisingly, we do not find evidence that perceptions of science and biotechnologies are linked to support for affordability (Models 4 and 5), with the exception that we find a relationship with belief that scientific discoveries are good for everyone (β = 0.13, P = .013). In the next section, we discuss these results and their implications.

3. Discussion

We find that US publics are able to form judgments about the IS, and we argue that these judgments reveal an ability to discriminate among different aspects of IS. The definition we provided them was intentionally very broad and included all sectors and timeframes from early conception to finished products. We intended the items in the scale to cover a range of societal and individual benefits in order to understand what constituted “public benefit” to our respondents. As the strong PCA results indicate, US publics simultaneously hold lukewarm confidence in the IS to respond to public input and deliver benefits generally while also being critical of environmental harms and corporate profiteering. Among the three components, respondents demonstrate a strong desire for reform—for an IS that heightens public benefit while mitigating harms, at least according to the specific aspects provided in this exploratory question set. Respondents strongly support the affordability of taxpayer-funded drugs and the enactment of new safety regulations, and they do so with very low uncertainty. This result highlights the public’s desire for public investment in scientific and technological development to produce safe, well-regulated, tangible, and accessible public benefits, including affordable drugs. These findings should be seen as a resounding public endorsement of the ethical precepts of the RRI framework—that the US science and IS and science policy should take into account and respond to public views and concerns.

However, democratizing science and innovation policy also requires attending to a diversity of views (OECD 2020). Our findings point to a broader diversity of views among Americans about these reforms than prior research on specific technologies would suggest. For example, we find differing results dependent on respondent social location and sometimes not in the expected direction. These results provide an important counterpoint to arguments about bimodal polarization of views on science and technologies, particularly along political and religious dimensions.

Our findings with respect to racialized and gendered identities did not consistently conform to our expectations. We anticipated that respondents from more marginalized identities would be more supportive of better safety regulations and more affordable taxpayer drugs compared to more educated White male respondents (Finucane et al. 2000; Shearer et al. 2013; Weisberg, Badgio and Chatterjee 2017). Instead, we find evidence that suggests there is no difference or even that the reverse may be true. For example, we find no significant differences by gender in either set of models and that educational attainment is associated with more rather than less support for affordability reforms. We also find that White respondents are more supportive of better safety regulations compared to Black and Latino/a respondents, on average, and they are more supportive of affordable taxpayer drugs compared to Asian, Black, and Latino/a respondents, on average. This may reflect privileged beliefs around access, where more educated and White respondents think that they should be able to access these drugs, while non-White respondents do not hold such expectations, or more educated and White respondents identify more closely with the term “taxpayer” and thus feel more entitled to affordability. Alternatively, the lower average agreement with these reforms may reflect a skeptical belief among some non-White respondents that reforms will not deliver the needed change. In particular, Black Americans may be understandably skeptical of science’s likely benefits, given historic injustices (e.g. phrenology, eugenics, US Public Health Service syphilis experiments on Black men in Tuskegee, and the ongoing use of Henrietta Lacks cells without permission), a relationship which might be fruitful to explore in future research. Even so, regardless of relative differences, all racial/ethnic groups and both men and women largely support these reforms, as measured by weighted averages. Furthermore, the relationship between racialized identities and support for better safety regulations may be partially explained or mediated by environmental justice views, a finding which echoes other research (Rivers, Arvai, and Slovic 2010). Given our decision to oversample Black and Latino/a respondents in our survey, we believe that these findings are not the result of under-representation and skewed samples within these groups.

Studies show that older individuals tend to view emerging technologies, among other things, through the lens of affordability and security (Zhang 2023). In this research, increasing age is associated with both support for better safety regulations and affordable taxpayer-funded drugs, suggesting that safety and affordability more broadly may be important factors for older adults. We find that political ideology is not significantly related to support for affordable taxpayer-funded drugs or support for better safety regulations, after controlling for environmental justice views. We also find that religiously affiliated individuals are less supportive of affordable taxpayer-funded drugs compared to religiously unaffiliated individuals, but we do not find this relationship with support for better safety regulations. Our (null) findings regarding political ideology are particularly interesting. Despite a widely reported politically polarized landscape in the USA (Pew Research Center 2014, 2022), political ideology is notable for not being an important predictor of these judgments about reforms. By and large, it appears that US publics regardless of political ideology are in favor of safe technological development and more affordable taxpayer-funded drugs. Somewhat surprisingly, other worldviews or values, such as on environmental justice, appear to be more salient than political ideology in the explanation of public views toward reforms.

Respondents’ personal views on satisfaction, discrimination, and powerlessness figure into their support for affordable taxpayer-funded drugs. Respondents who feel satisfied with life are less supportive of the need for affordability, perhaps because they feel satisfied with the status quo. Those who feel discriminated against and those who feel they have little control over risks to their health also agree less that taxpayer-funded drugs should be affordable. This puzzling response may reflect that these respondents are more likely to feel alienated from the benefits of the IS as a whole and medical innovations in particular. Additional qualitative in-depth interviews present an ideal direction for future research to examine more closely how these kinds of respondent worldviews may influence their views on IS reforms.

Surprisingly, we find null relationships between views on science and heretofore controversial technologies, such as biotechnologies, and judgments about the IS. Contrary to our expectation, individuals who are more risk-averse toward biotechnologies are not more likely to embrace safety reforms to the science and technology IS, indicating that risk aversion is not related to safety regulations but to some other factor. We find that respondents who agree that scientific discoveries are good for everyone, not just the rich, are on average more supportive of the need for affordable drugs. This may reflect a principled stance on making drugs more affordable in order to align with the belief that scientific discoveries are a public good and that they are for everyone, not just those who can afford costly new technologies. Such concerns about distributive justice have been reported with regard to other specific emerging technologies, including nanotechnologies (Pidgeon et al. 2009), unconventional oil and gas (Thomas et al. 2017; Harthorn et al. 2019), geoengineering (Pidgeon et al. 2013), genomics (Reardon 2013, 2017), and a number of other new technologies.

Indeed, concern around justice and equity appears to be a major component of judgments about the IS and the need for reform: we found that environmental justice beliefs are powerful correlates with views on the IS. Support for governmental intervention on behalf of environmentally marginalized communities of color was the strongest correlate with both agreement with need for better safety regulations and more affordable drugs. Belief in power asymmetries and, to a lesser extent, belief in environmental racism, also correlated with the need for reform statements. We interpret this as demonstrating that values of justice and equity permeate respondent perceptions of the IS and see this as resonant with other recent literature (Acemoglu and Johnson 2023; Reardon 2013). As noted earlier, we have not identified directly comparable findings on European public views of the IS itself, but we note that the wider EU RRI policy framework encourages a more participatory, reflective, and responsive IS. We would suggest that these US public views are reflective of generally more positive views about likely societal benefits from technological innovation than their EU counterparts, tempered by perhaps distinctly American patterns of concern about social risks and justice driving the desire for reform.

While the results of this exploratory research demonstrate strong support for better regulation of new technological developments, we cannot specify from these results exactly what kind of regulations or policies are desired or if there are specific technologies that respondents particularly want to be regulated. These findings provide an excellent opportunity for future research to expand on by probing engaged publics in depth on their views on needed reforms to the IS. One possible limitation includes priming effects of preceding tasks in the survey itself, such as discussion of synthetic cells and other technological developments, that may have inadvertently increased respondent sensitivity to the issues. Furthermore, this question set was experimental and exploratory, and therefore more validation and refinement in future research will be valuable. The PCA results are indicative of how to proceed. More research on perceived benefits, harms, and needed reforms in the IS including novel survey research, in-depth interviews, or deliberations that explore a broader array of questions on the IS would provide crucial feedback to science and innovation policymakers. US public concerns about the labor force and job security in the face of such technological development are particularly of concern (Zhang 2022) and merit further attention in future research. Regarding synthetic cells, research in progress by the authors builds on the work on dichotomies of natural and artificial in public perceptions and ambivalences about medical technologies by Reich (2016, 2020). We note, too, that in thinking about the public benefits of the IS, Rodrik and Stantcheva (2021) have provided a framework for thinking about “inclusive prosperity” in a far more nuanced way across income groups and production stages that future research could fruitfully explore.

These results demonstrate that US publics do not simply endorse the current IS. While these data reflect moderate confidence in the IS as is, US publics strongly support the need for reforms, Studies that show declining trust in scientists are a flashing light for the US scientific enterprise (Kennedy, Tyson, and Funk 2022; Kennedy and Tyson 2023). We propose that continuing to expand and incorporate RRI precepts in the USA provides a necessary first step toward rebuilding trust in scientists and scientific institutions. In particular, generating democratically agreed-upon rules and ethical norms for responsible innovation may be most effective in creating a responsible IS for the future (Weber 2010). The IS must engage with and incorporate a diversity of public views, not only those of already engaged publics. The current US national experience with AI demonstrates the pressing need for heightened public engagement and anticipatory approaches to technological development (Zhang and Dafoe 2019). We endorse calls to enhance the public input infrastructure in the USA (Kuzma 2021), particularly around how reforms take place, as the current structure is inadequate to sustaining genuine engagement with diverse publics. Expanding how public input is incorporated into the IS provides a potential resolution to the “implicated actor” problem, where publics are directly affected by the IS but lack meaningful presence or agency. Addressing this may also require government, industry, and university-based IS actors to not only offer new and better avenues for meaningful public input but also reach out and engage with marginalized communities to enhance broad-based participation in the US science and technology IS.

Acknowledgements

The researchers thank all collaborating scientists and engineers on both synthetic cell projects (lead Principal Investigators Allen Liu, University of Michigan, USA, and Cheryl Kerfeld, Michigan State University and Lawrence Berkeley Lab, USA) and their students and postdocs for their many generous and thoughtful contributions. The authors also thank Richard P. Appelbaum, UCSB, for critical feedback on a draft; Mary B. Collins, Stony Brook University, for methodological advice; 3 anonymous reviewers for their helpful comments; and the anonymous public participants who contributed their ideas to this research.

Supplementary data

Supplementary data is available at SCIPOL Journal online.

Conflict of interest.

None declared.

Funding

This work was supported by the US National Science Foundation (Collaborative Grants #BIO-EF-1935184 and #BIO-EF-1935231 to B.H.H.).

Data availability

The data underlying this article are available in the Harvard Dataverse at https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/TTKTAP (doi: 10.7910/DVN/TTKTAP).

References

Author notes