In the last decades, the Public Understanding of Science (PUS) approach has faced the challenge of responding to an increasing number of questions regarding the relationship between science and society. One of the underlying concerns in this area has been to elucidate which variables encourage public engagement in science. In order to resolve this question, PUS surveys have become a useful and habitual tool since they were first implemented in the late 1970s (see Bauer, 2009 for a revision). These types of instruments have provided empirical material of great value to test main PUS hypotheses, to check the relationship between different perception variables, to make comparisons between public engagement in different countries and to put the data in context with other types of contextual indicators (Durant et al., 1989; Godin and Gingras, 2000; Bauer and Durant, 1999; Shukla, 2005; Allum et al., 2008; Bauer et al., 2012; Guenther et al., 2018). However, it is also true that they still suffer from certain deficiencies, such as a lack of periodicity, of a body of comparable questions, and of theoretical support for some of the measuring instruments (Bauer et al., 2007).

Most of these approaches have been dedicated to analysing public perception of science in the general population, emphasising in some cases their differences, for example, by age, gender or nationality. However, with notable exceptions (Prpić, 2011, for example) there are hardly any studies that focus on key sectors of modern societies, such as the political, judicial, financial or business classes. Therefore, the question is: Why should we limit the target of PUS studies only to the general public? Why not expand focus to agents, as business managers, who are in fact largely involved in the execution of research and the appropriation and transfer of scientific knowledge. Recently, Bauer (2014) pointed out the need to continue expanding the framework ‘to reveal the diverse relations that different sections of the modern public have with science’. In this research, we propose to approach the understanding and perception of science by one of these key audiences, the business sector, traditionally neglected in the PUS field.

A brief review of the canonical journal in this area, Public Understanding of Science, shows a symptomatic fact: between 1992 and 2019, none of their published articles included the words ‘industry’, ‘firm’, ‘company’ or ‘business’ among their keywords -In fact, the first article in Public Understanding of Science with ‘industry’ as a keyword was published during the writing of this paper, in October 2019-. Despite this, there are some articles in the journal in which companies appear as a main or secondary actor in the paper. However, when this happens, they usually appear in relation to their role in scientific controversies that concern public opinion, such as genetically modified organisms (GMOs), climate change, biotechnology or chemical pollution. That is, there is a clear bias in the PUS literature-and Science, Technology and Society (STS) literature in general-that usually relates the business sector with a special ability to ‘manufacture doubt’ about science when it threatens their corporate and special interests (Stocking and Holstein, 2009). Not surprisingly, the Science and the Public report (OST and Wellcome Trust, 2001) revealed that there is a negative perception of the power of industry to set the scientific agenda along with certain concerns about the preservation of the scientific arena from corporate interests.

However, this gap in the literature contrasts with a growing role of industry in the European scientific policies related to a shift in the production model towards an intensive knowledge-based economy (Saisana and Munda, 2008). In fact, since the detection of the so-called ‘European-Paradox’ (European Commission, 1995) -the conjecture that EU countries play a leading global role in terms of scientific output, but lag behind in the ability of converting this strength into innovations (Dosi et al., 2006)- concern about the relation between research and innovation is increasing (Frenz and Ietto-Gillies, 2009). In this scenario, a large volume of data about the inputs and outcomes produced by the relationship between science and business is available: that is, how much money they invest in research and development (R&D), how many research projects they execute, how many patents they register, how many doctors they hire, etc. -see, for example, the collection of the Science and Technology Database of Eurostat (Eurostat, 2020)-. Nevertheless, the relationship of company managers with science remains black boxed. We still have a theorical and empirical gap about which type of perception variables could influence the implementation of pro-scientific behaviours in companies. This study aims to open that ‘black box’ on the premise that the imagery about science and technology that an individual harbours depends on -or rather is mediated by- the social role and position that he/she occupies (Godin and Gingras, 2000; Sturgis and Allum, 2004).

Gonçalves et al. (1996) and Prpić (2011) showed their surprise about the scarce surveys of perceptions of science conducted among business managers and other power elites who have a possible influence on scientific and technological policies. But little has been done since then to resolve this gap, although Prpić showed that managers reveal views about science and its implications that are significantly different from those of the wider public. In general, elites express ‘a remarkably lower level of traditionalism’, and they are ‘more inclined to idealize science’ than is the general population (2011). In the same year, López-Navarro et al. (2011) also confirmed that, in the Spanish context, the perception of science in the wider public and among entrepreneurs and self-employed workers was significantly different, particularly regarding institutional trust and risk management

In a recent work, we found that the image of science in the business sector is shaped by ‘entrepreneurs’ and business managers’ perception of science, their interest in and knowledge of science and technology, and their willingness to take action regarding science, R&D and innovation’ (Rey Rocha et al., 2019). We thus propose that business engagement in R&D is related to entrepreneurs’ and managers’ perception of science. Particularly, the aim of this study was to test this relationship and to measure the effect that certain PUS variables have on it.

With respect to these direct research antecedents, we have considerably expanded the number of dimensions included in our survey and the targeted sample, including companies of all economic sectors and sizes. Finally, our study not only describes business sector perception of science, but attempts to provide empirical evidence about its relationship with R&D execution.

This research is based on the results of the Scientific Culture at Enterprises 2016 survey (Rey-Rocha and López-Navarro, 2016), the first research tool specifically designed to investigate science perception in the business sector. We have included, on the one hand, classic dimensions traditionally used in most science perception surveys, which count on a considerable volume of previous literature and a high consensus about its implementation: namely, knowledge, attitude, interest and information. On the other hand, less usual dimensions -related to the recent ‘science and society’ paradigm (Bauer et al., 2007)- have also been included: institutional trust, closeness and appropriation.

The following is a review, which is not intended to be exhaustive, about the main consensus and vanishing points in relation to the use of each of these variables in the PUS field. The methodology used for this work is presented, followed by the main results of our research. Finally, the capabilities and limitations of the perception of science variables to explain research engagement in the business sector and their implications related to science policy are discussed.

Previous literature has speculated about what is scientific culture when we talk about the business sector. For example, Godin and Gingras (2000) pointed out that ‘for industrial executives and managers, it could be the capacity to invest wisely in research, and to evaluate and select from a group of new technologies, as well as to provide for adequate employee training and proper equipment maintenance’. Elzinga and Jamison (1995) pointed out the relevance of the cultural dimension of scientific policy, including the bureaucratic, academic, civic and economic cultures, in their conceptual typology. The economic culture or approach to scientific policy was typical of industrial companies and it was focused on the technological appropriation of science and a more accounting perspective of R&D. In the theoretical approach to innovation adopted by Quintanilla (2004), the propensity to engage in R&D is related not only with the economic capability of the firm to do so but also to its attitude toward science and technology.

However, in spite of theoretical approaches, little is known in the empirical field about the perception of science in companies. Prpić (2011) used the above-mentioned typology proposed by Elzinga and Jamison (1995) to investigate the perception of science of different key social actors. She showed that managers´ views of science were less traditional and more optimistic than of the wider public. However, a gap has been found in the literature review about the relation between perception of science and appropriation of science in the business sector. In contrast, we have a huge empirical baggage if we talk about the general public instead of business managers or entrepreneurs. Therefore, it is convenient to review the results of the use of the main variables of the PUS approach in order to transfer them into an empirical model that explains the relationship between perception of science and business engagement in R&D activities. In our review, we attend to the PUS traditional variables of knowledge, attitude, interest and information and a secondary group of variables with a more recent presence in the public perception of science surveys and that also have a certain empirical and conceptual relevance.

The main aim of this study was to test the relation between perception of science and R&D engagement in the business sector. We consider R&D engagement, as suggested by Godin and Gingras (2000), as a feature of scientific culture of the company, as a form of appropriation of science. Considering the aforementioned, we hypothesise that this form of appropriation -measured by the decision to carry out R&D activities in the company- should be positively related to the perception of science held by firm managers. In this sense, it would be predictable that not only traditional variables (such as knowledge, attitude, information and interest) contribute to explain R&D engagement but also that third generation variables (such as appropriation, closeness to science and institutional confidence) play a role in this relation. Our second objective in this study was to assess the weight of each one of the variables in their relationship with R&D engagement.

Finally, we have to consider previous contributions from other academic fields, especially the economic and innovation management area. This area has been providing empirical evidence for decades on the influence of economic and market structure variables on business research activity (Den Hertog, 1993; Van Dijk et al., 1997; Lee, 2003; Máñez et al., 2015; Doloreux et al., 2016). In the same way, the PUS field has shown the certain influence of individual characteristics, such as gender or age, on the public perception of science (Evans and Durant, 1995; OST and Wellcome Trust, 2001; Noy and O’Brien, 2019). For that reason, we decided to check our hypothesis controlling for individual characteristics of CEOs and structural and economic characteristics of the company.

A detailed description of the methodology used in this research has been published elsewhere (Rey-Rocha et al., 2019; González-Bravo et al., 2020). However, to facilitate the comprehension of the present article, the most relevant aspects of our method and data analysis are summarized below.

Our analysis approach consisted of logit regression models made to gradually add different variables to each of them to test their effect on R&D engagement. This step-by-step approach allows us to understand a) the sole effect of science perception variables (innovative and traditional); b) the added effect of individual control variables; and c) the added effect of structural and economic control variables.

The coefficients obtained for explanatory variables with suitable levels of significance estimate their relationship with R&D engagement in the business sector. Expβ coefficients above 1 indicate that an increase in the explanatory variable is related to an increase in the likelihood of a firm performing R&D activities (Table V).

Statistical analyses were done with the Statistical Package for Social Sciences (SPSS) v. 25.

A summary of the results for the regression models is provided in Table V, and the details of each analysis are shown in the Appendix. Model 1 includes perception variables, thus showing the probability of engaging in R&D in relation to the firms’ perception of science. Models 2 and 3 incorporate the interaction terms to test the moderation effects involving individual characteristics of respondents and firm-level variables, respectively. The models accurately explain a relatively large proportion of variance in the explained variables (Nagelkerke R² values between 0.511 and 0.541). The insertion of control variables in the model barely generated changes in R² and in the significance and direction of the effect of science perception variables in the model.

The results show that institutional knowledge -that is, the capacity of CEOs to recognise scientific institutions in their environment-, together with the regular use of scientific institutions as information sources to keep up to date about science and technology in the company, are the variables that present the highest predictive capacity in the model (they show the highest beta coefficients, thus the highest odds ratio increases). When a company’s CEO knows at least one scientific institution, the probability of he/she belonging to a company engaging in R&D activities increases by almost 200%. For every additional source of scientific information used, the probability of engaging in R&D activities increases by more than 100%. Interest in science and technology, and the level of closeness to science, are also significantly and positively related to business engagement in R&D. The level of confidence in scientific institutions is also significant but in an inverse direction. That is, the lower the confidence expressed by the managers of a company in scientific institutions when it comes to dealing with their company’s issues related to science and technology, the more likely it is that the company will carry out R&D.

Models 2 and 3 show that the relationship between engagement in R&D and perception of science remains even after controlling for individual characteristics of the entrepreneurs and business managers and for economic and structural characteristics of the firms. Both firm capabilities to generate resources internally (ROA) and a certain firm size (above 10 employees) are positively related with R&D engagement; they also nullify the effect of the age of respondents when individual but not firm characteristics are considered.

The results reported here confirm the potential we had already anticipated (Rey-Rocha et al., 2019; González-Bravo et al., 2020) of the Public Understanding of Science approach to light up issues related to scientific culture, not only in the general population but also in new scenarios such as the business sector.

The results of the first survey on scientific culture, perception and attitudes towards science and innovation in the business sector provide empirical evidence of a significant association of PUS variables and business engagement in R&D activities. A mix of traditional (institutional knowledge, interest, information) and third generation PUS variables (institutional confidence and closeness level) are significantly related with research decisions of the company, even when individual characteristics of business managers and economic and structural characteristics of firms are controlled.

The institutional knowledge of respondents about science and research systems, and the use of scientific sources of information were the variables more strongly associated with firms engaging in R&D activities. This result points to the relevance of the connection between the business sector and the innovation system, through an appropriate knowledge of research institutions by managers, to foster firm engagement in R&D activities. In other words, it suggests the importance of institutional and contextual knowledge -more than formal contents of scientific knowledge- when it comes to promoting engagement in R&D in the business sector. In fact, we found in our previous research that knowledge of scientific institutions was a better predictor of knowledge of science by CEOs than their level of formal education (Rey-Rocha et al., 2019). This contextual or institutional knowledge is related to the notion of tacit knowledge (Collins, 2010), a type of ‘expert’ and ‘focused knowledge’ that does not refer to the corpus of scientific knowledge itself but to the social and institutional system in which scientific knowledge is inserted as well as its management model. Familiarity with the institutional framework of science is thus essential for a closer approach to research by companies. Unfortunately, this type of knowledge has the particularity of being very difficult to transfer due because it is difficult to codify. This fact makes this ‘institutional know-how’ very dependent on the concrete person who holds the management position in the company/who the R&D decisions in the company. Like any other ability, this knowledge is not likely to be written or verbalised. For this reason, this type of skill usually goes with the manager when he/she leaves the firm or changes his/her position in the firm, revealing the volatile nature of this knowledge. In the light of the results obtained, it would be worthwhile to pay more attention to institutional knowledge in future analyses, as its inclusion in national perception of science surveys, or even in innovation studies, is still deficient.

The information dimension is closely related to institutional knowledge since the use of scientific sources of information implies certain previous knowledge about the organisations that produce it. This dimension has received increasing attention in PUS studies. Science communication, and the conditions under which it is produced, plays an increasing role in public engagement of science (Vraga and Bode, 2017; Bolsen et al., 2019). Our study confirms that, in the same way, the information dimension also makes an important contribution to explain business engagement in R&D.

Confidence in scientific institutions when addressing issues in the company related to science and technology is also significantly associated with R&D engagement but, surprisingly, in an inverse direction. This type of relation could suggest a kind of system failure related to the lack of confidence among innovative companies regarding the performance of universities and public research institutions. Our results show that intensive R&D companies were the most suspicious. That is, distrust would be related to their experience with this type of environment, not to the lack of contact with it. This result is consistent with previous work about cross sector collaboration and the distrust that generates public administration in the business sector (Diaz-Catalan et al., 2019) In fact, the Spanish business sector has traditionally developed their R&D internally and firms prefer this option over externalisation or collaboration with academic agents to obtain new knowledge (Industry, Economy and Competitiveness Ministry, 2017).

An inverse relation between institutional confidence and R&D engagement would also be interpreted as the reverse of one STS insight: ‘distance lends enchantment’ in science (Collins, 1985, 145). In this case, proximity to science -indeed, to scientific institutions- could lend certain criticism or distrust among the business sector. This interpretation is also in accordance with recent findings about the public perception of science in the general public carried out by Cámara et al. (2017). They identified ‘critical engagers’ as a segment of the population that holds an overall positive attitude towards science but is not reticent about expressing concern regarding particular scientific applications. The authors interpreted this result as a sign of a mature and conscious society, far from naïve positions towards the effects of science and technology applications. In the same way, our results suggest the existence of business as critical engagers; that is, firms that engage in R&D but are not reticent about expressing concerns about public research institutions.

Previous economic and innovation literature about the propensity of companies to carry on R&D activities very rarely has seen this disposition as a part of their scientific culture or, at least, as a form of scientific appropriation. With some exceptions (see, for example, Alam et al., 2019; Lorca and de Andrés, 2019), these academic approaches usually looked at market and economic variables to find the reasons that would explain R&D intensity (i.e., size of the company, economic sector, internal resources, debt, etc.) (Xu and Sim, 2018). For that reason, control variables have a particularly relevant role in this study, and their estimations/coefficients led us to confirm that all science perception variables remain necessary to explain R&D engagement in the business sector. Moreover, the explanatory power of the model does not substantially improve with the inclusion of structural and economic variables. However, results show that, on the one hand, engagement in R&D is positively associated with a firm’s capability for generating income and, on the other hand, a minimum size is required to address these types of activities. Regarding demographic control variables, neither sex nor gender has a significant association with science engagement in firms when all variables are included in the model. This result is consistent with PUS literature on science funding support. If we understand R&D investment in the business sector as a similar phenomenon as science funding support in the general public, the recent work of Besley (2018, 97) has pointed out that previous literature only found ‘limited relationships between demographics and support after controlling for more proximate variables’. In the same way, Prpić (2011) found that managers´ perception of science did not vary according to their gender. Only significant, yet small, differences were found in older respondents who tend to be more sceptical about science.

In 2007, Bauer et al. mapped the path of the PUS approach since its inception and anticipated ‘a fertile period for survey research on public understanding of science (…) albeit within the wider framework of science and society’ (2007, 79). The present study, inspired in the proposal mentioned by Bauer, et al. (2007), formulate a cross fertilisation between the studies on business innovation and innovation management and the public understanding of the science approach. That is, a new actor for a consolidated -and revisited- academic approach.

The relevance of the inclusion of new actors lies in the idea that, not only were there many publics (Einsiediel, 2000) but also multiple perceptions of science, bearing in mind that each perception is mediated by the social role of individuals and groups (Godin and Gingras, 2000). The particular position that the business sector has in the society -closely related to knowledge generation and transfer, appropriation of science and science and technology policies- could contribute to complete the complex puzzle of collective S&T culture (Rey-Rocha et al., 2019).

Finally, situated in the business scenario, the PUS approach has the potential to pose new answers -based on science perception indicators- to old unresolved questions related to business innovation and R&D in order to generate more accurate S&T policies. That is, the incursion into this new scenario positions the PUS approach in a strategic place to address the challenge of improving the relationship between science and the business sector. Our results confirm that the PUS approach is a valid interlocutor, among many others, to talk about R&D engagement: what distinguishes a firm that is proactive toward R&D engagement is primarily the institutional knowledge of its managers about the scientific ecosystem in which the company is inserted and the variety of scientific sources they habitually consult. When we put PUS variables in the equation, money also matters but in a modest way (see also González Bravo et al., 2020).

Turney (1996, 1087) stated that ‘we need to improve the scientist’s understanding of the public’. Today, maybe we could apply this recommendation to point out that not only managers and CEOs should have a better understanding of and engagement with science, but academics and science policy makers must have a better knowledge of the business sector as a particular public. In summary, PUS studies need to expand their agenda (Bauer et al., 2007) since they have the capacity to light up certain dark areas that the economy or innovation studies cannot solve alone when it comes to dealing with business engagement with science.

Further research is also needed to counteract the limitations of the present study. Due to questionnaire limitations, the knowledge variable could not be measured in a direct manner, so we had to use level of formal education as a proxy for this dimension. In future research, it is recommended to include in the survey a specific question for this purpose. Second, the economic indicator used as a control variable leaves out more specific indicators, such as leverage of the firm or asset turnover, that could refine our results. Third, it would ideally be recommendable that this study could open a new line of research in which more specific studies -centred in a particular sector or business size- could be replicated and compared with this general sample. This would allow us to test whether the relations found in this study vary in particular contexts.

It is particularly important to note that the cross-sectional nature of our data constrains the possibility of testing causal links between PUS variables and R&D engagement. We have been especially careful in the paper not to impute this type of relation in the interpretation of our results. Future research is also needed for a better explanation of the relation among variables. As López-Cerezo and Cámara-Hurtado (2007) proposed, the relation among science perception variables should be understood as a ‘non-linear process of a gradual, reciprocal and recursive character’. This would give us a better understanding. of the function of the behaviour of these types of variables and it would avoid the temptation to fall into linear interpretations.

This research constitutes an initial attempt to study the relationship between the perception of science and R&D, in combination with firms’ structural, economic and financial characteristics, and their engagement in R&D. The results provide sundry openings for further research to reinforce them. Some of the perception dimensions invite further and more detailed study, through different quantitative and qualitative research approaches and instruments.

Confidence and trust are particularly important in relationships of companies with science, scientists and scientific institutions. They invite further research, which should investigate factors as attribution of scientific institutions’ effectiveness when dealing with particular problems for the company.

Finally, in regard to our explained variable, it would be interesting to distinguish between the three more common categories included in R&D (internal, external and collaboration) in order to prove whether particular perception variables change its direction or its intensity. In the same way, in addition to R&D engagement, engagement in innovation activities must be investigated as explained variables in this line of research.