Abstract
In this research, the authors examine how consumers perceive the fruits of university–industry collaborations (i.e., new products codeveloped with universities). Eight studies document a positive university effect and highlight its practical significance and boundary conditions. An Instagram A/B test utilizing a video that refers (vs. does not refer) to the underlying university–industry collaboration results in higher click-through rates and ad engagement levels. Another field study demonstrates that university-codeveloped products are more attractive to consumers, even after an actual product trial. Further, several consequential studies reveal that consumers are willing to pay up to 65% more for products marketed as codeveloped with a university. The authors argue and show that collaborating with a university infuses the underlying firm with a stronger sense of scientific legitimacy, thereby making the resulting product more attractive to consumers. Congruously, the authors find that the effect is more pronounced when the scientific legitimacy engendered by universities is more important to the focal product (i.e., high tech vs. low tech), underlying company (i.e., new vs. established), or target customer (i.e., high vs. low belief in science).
Keywords
Firms collaborate with universities to access novel scientific knowledge and technological expertise (Bercovitz and Feldman 2007; Lavie and Drori 2012; Wirsich et al. 2016). In short, working with universities on innovation projects can facilitate the development of superior new products (Bianchi et al. 2016; Gretsch, Salzmann, and Kock 2019; Nieto and Santamaría 2007). For example, Angles90, a startup from Northern Italy, codeveloped the first dynamic training grips worldwide with faculty of strength training ergonomics from the Technical University of Munich, Germany. Before long, the company had successfully sold its patented innovation in more than 30 countries (Angles90 2020). Similarly, well-established firms, such as Adidas, also engage in university–industry collaborations. For example, prompted by research showing that body heat management of young children is less effective than that of older children or adults (i.e., younger children are physiologically unable to effectively keep their skin cool with evaporating perspiration), faculty from Loughborough University (2016) helped Adidas develop a range of “Climacool” training tops that are targeted at young children and offer full air flow during exercise.
Literature in this space has mainly examined how firms can best collaborate with universities from an innovation management perspective (e.g., Bercovitz and Feldman 2007; Wirsich et al. 2016). In our research, we take a different approach by asking how consumers perceive and experience university-codeveloped products. Specifically, will consumers react differently to the same product upon learning it has been codeveloped with a university? What will these perceptions depend on, and in which situations will the related inferential process yield more favorable results? How strong, and thus managerially relevant, are the effects?
At the core of our theorizing is the idea that collaborating with a university infuses the underlying firm with stronger scientific legitimacy, thereby making the resulting product more attractive to consumers. A firm with scientific legitimacy is viewed as being able to understand and effectively work “with the latest scientific ideas in the field,” and thus capable of developing cutting-edge technological innovations (Rao, Chandy, and Prabhu 2008, p. 61). Critically, we reason that this effect derives from the image of universities in general (vs. the specific image of a given university). 1 We further argue that university-codeveloped products might be more attractive to consumers when the scientific legitimacy afforded by universities is more relevant to the focal product, underlying company, or target customer.
Eight studies comprising two field studies, several incentive-compatible experiments, and a series of more nuanced studies based on data collected across four countries document a positive university effect and highlight its practical significance and boundary conditions. Our theorizing and findings imply that university–industry collaborations are promising from not only an innovation perspective, as suggested by prior literature, but also a marketing perspective. Put differently, firms may benefit by actively marketing products resulting from university–industry collaborations as such at the point of sale. We believe the related implications for firms are novel and potentially disruptive because thus far, companies rarely advertise their products as codeveloped with a university. 2 Two intuition studies support this conjecture (for details, see Web Appendix A). In the first study, we instructed 22 MBA students to develop a short product advertisement based on various background information about a company and its latest product, including the notion that said product was codeveloped with a university. Findings revealed that only 4 out of 22 managers (18.2%) used the university-codevelopment information when marketing the focal product. In the second study, we used a similar paradigm involving 42 Master of Science in Marketing students. Again, only a minority of participants (14.6%) decided to include the fact that the focal product was developed in collaboration with a university in their advertisement copy.
Our research offers a series of nuanced contributions to the theory and practice of university–industry collaborations. In short, we add a new branch to the managerial decision tree in that space. Once a firm has decided to codevelop a new product with a university (a decision that mostly depends on the innovation potential of a given university–industry collaboration), we highlight how and when actively marketing university-codeveloped products as such may yield incremental benefits. In addition, and as discussed in more detail in the “General Discussion” section, we also contribute to the related literatures on legitimacy (Rao, Chandy, and Prabhu 2008), cobranding and ingredient branding (Desai and Keller 2002; Pinello, Picone, and Destri 2022), source credibility (Pornpitakpan 2004), and belief in science (Farias et al. 2013; Rutjens, Sutton, and Van Der Lee 2018).
Open Innovation and University–Industry Collaborations
To develop promising new products, firms increasingly look beyond their own boundaries and engage in open innovation practices. Brunswicker and Chesbrough (2018) report that over two thirds of large firms surveyed in Europe and the United States indicate that they rely on open innovation. To illustrate, in 2011 Unilever invested approximately 60% of its research and development budget in open innovation projects (Warc 2012). Broadly speaking, open innovation involves moving from a closed innovation paradigm to one that is open in several directions (Chesbrough 2003); instead of relying exclusively on internal research and development staff, companies search for novel insights, stimulating ideas, technological problem solutions, and know-how partners within the firm's periphery.
As highlighted by the aforementioned examples, universities are frequent open innovation partners (Yusuf 2008), especially when technological and scientific expertise may be relevant to the focal innovation project (Colombo et al. 2021; George, Zahra, and Wood 2002; Soh and Subramanian 2014). In university–industry collaborations, companies interact with universities to enhance their innovation-related capabilities (Bekkers and Bodas Freitas 2008; Siegel, Waldmann, and Link 2003; for reviews, see Ankrah and Omar [2015] and Perkmann et al. [2013]). Jeff Hamner, P&G's former Vice President of Corporate R&D, argues that P&G's collaborations with universities are a means “to drive cutting-edge innovations that have the potential to impact a wide range of P&G's global product categories” (Business Wire 2011). Our literature review of this topic identified 42 empirical articles published between 2001 and 2022 (for an overview, see Web Appendix B). Although the collective evidence points to positive effects of university–industry collaborations, the extant literature has been hitherto silent regarding our research questions.
Most of the existing research focuses on firm-level innovation outcomes such as the number of patents generated (e.g., Soh and Subramanian 2014), but only seven studies have investigated effects on market success indicators. While George, Zahra, and Wood (2002) as well as Eom and Lee (2010) found no such effects, Belderbos et al. (2014) and Kafouros et al. (2015) found a significant positive relationship between university–industry collaborations and firm-level financial performance. Similarly, the cross-sectional investigation by Lööf and Broström (2008) found that collaborating with universities is associated with a 7% increase in firm-level innovation sales; moreover, Belderbos, Carree, and Lokshin (2004) and Chen, Vanhaverbeke, and Du (2016) demonstrated that collaborating with universities is positively related to firms’ new product sales.
Interestingly, none of these studies have investigated the effects of university–industry collaborations on a product level, and all the evidence presented is correlational in nature. Thus, it is not yet established whether working with universities has a causal effect on the resulting new products’ eventual market performance. Moreover, previous research fails to disentangle a potential objective effect of university–industry collaborations (i.e., a factually better product may emerge) from a more subjective one (i.e., consumers may perceive a given product to be better). In other words, the extant literature provides no direct insights into how consumers perceive university-codeveloped products.
Consumer Reactions to University-Codeveloped Products
Related Literature
There are several literature streams that suggest consumers may react differently to a given product if they know it has been codeveloped with a university. First, on a high level, consumer inference literature maintains that consumers often generate if–then linkages between the limited information available and their conclusions (Kardes, Posavac, and Cronley 2004). For example, the well-known price–quality inference suggests that consumers believe price and quality are highly correlated, and that a higher price (directly observable) implies a higher-quality product (often not directly observable; e.g., Huber and McCann 1982). A case in point, Plassmann et al. (2008) find that consumers experience a given wine differently (e.g., find it tastier) when framed as more (vs. less) expensive, resulting in different neural representations of experienced pleasantness. Second, research on cobranding (e.g., Apple Watch Nike+ [Pinello, Picone, and Destri 2022]) and ingredient branding (e.g., Intel inside Lenovo [Desai and Keller 2002]) point to potential benefits when firms collaborate with external partners. Such brand alliances describe “a marketing strategy in which two or more brands are presented simultaneously to the consumer as one product to create a sum of brand assets, that is greater than that of the individual brands” (Turan 2021, p. 911). As reflected in the concept's definition, the success of cobranding rests on the individual brand associations of the collaborating firms held by consumers (e.g., brand attitudes, brand equity, brand quality perceptions [Keller and Lehmann 2006, Simonin and Ruth 1998]). Third, research on source credibility demonstrates that the source of a given message affects its persuasiveness. Although different characteristics of a message source can potentially affect its persuasiveness (e.g., physical attractiveness or ideological similarity), a meta-analysis by Wilson and Sherrell (1993) shows that expert sources exert the strongest effects. Taken together, the related literature streams make our predictions plausible; however, it is important to note that none of these research streams have actively studied university–industry collaborations. As such, the effect, its strength, the underlying process, and related boundary conditions are not yet defined. 3
The Positive University Effect
We conjecture that consumers may react differently to a given product upon learning it was codeveloped with a university. Specifically, we predict that consumers will find products codeveloped with a university more attractive, compared with products developed solely by the underlying firm or codeveloped with another company. This prediction should hold even when the products’ objective characteristics remain constant, the university brand name is largely unknown to the focal consumer, or no specific university brand name is given. We argue that the positive university effect can be explained through different consumer perceptions of the underlying firm's scientific legitimacy.
Legitimacy is “a generalized perception or assumption that the actions of an entity are desirable, proper, or appropriate within some socially constructed system of norms, values, beliefs, and definitions” (Suchman 1995, p. 574). Achieving legitimacy is important for organizations because it results in attracting important resources and stakeholders as well as organizational growth (Maier et al. 2023; Shepherd and Zacharakis 2003; Zimmerman and Zeitz 2002). As indicated previously, scientific legitimacy refers to one's ability to understand and effectively work “with the latest scientific ideas in the field” (Rao, Chandy, and Prabhu 2008, p. 61). Therefore, with an eye toward innovation, a firm with high levels of scientific legitimacy presumably has the technological capabilities and know-how to develop successful new products. Indeed, scientific legitimacy has proved important for innovations and new products, and adopting strategies that provide firms with such legitimacy in the eyes of stakeholders has been associated with favorable stock market gains (Rao, Chandy, and Prabhu 2008). 4
The general and widely shared image of universities when it comes to scientific and technological expertise helps us predict that consumers may associate a university-collaborating firm with higher levels of scientific legitimacy. Put differently, universities may imbue the focal firm with scientific legitimacy. 5 Indeed, universities are uniquely known for being at the forefront of the generation and dissemination of the latest scientific knowledge, which is needed to develop new technologies; therefore, universities enjoy an unprecedented reputation for being essential to technological progress and innovation (Boulton and Lucas 2011; Etzkowitz et al. 2000; Huber 2016; Scott 2006). Recent research highlights the associated positive image of universities among lay consumers. For example, survey data from the United States shows that more than four out of five citizens (83%) see universities’ scientific advances as benefiting U.S. society, and more than two out of three citizens (73%) think that universities positively influence national prosperity and development (Drezner, Pizmony-Levy, and Pallas 2018).
A qualitative exploration of consumers’ associations concerning a product portrayed as codeveloped with a university supported our scientific legitimacy reasoning (n = 108; Mage = 21.4 years; 57.4% female, 42.6% male; for details, see Web Appendix A). Specifically, we heard many responses indicating that the product might have benefited from the latest scientific knowledge and technological know-how brought in by the collaborating university (e.g., “I find it good that the product was created with scientific knowledge” [Respondent 51], such that “the latest state of knowledge was incorporated into the development of the product” [Respondent 65]). Interestingly, many responses revealed that consumers presumed the product had been developed in a rigorous, scientific way, similar to the way in which academics are thought to conduct research at their universities (e.g., “the product was optimized for a long time in structured processes at the university” [Respondent 80], including “precise testing/research/many surveys or prototypes to achieve the perfect outcome” [Respondent 104]). As a result, participants frequently thought of the underlying product as particularly “sophisticated,” “technologically advanced,” and “innovative” (e.g., Respondents 11, 31, 71), or “objectively verified” and “trustworthy” (e.g., Respondents 4, 43, 57). The following quote provides an astute summary of consumers’ potential reactions to university-codeveloped products: “When a product has been developed together with a university, you have the feeling that the people who developed the product really know their stuff and exactly know what they are doing. This makes the product more attractive” (Respondent 29). Taken together, we reason that collaborating with a university infuses the underlying firm with stronger scientific legitimacy, thereby making the resulting product more attractive to consumers. More formally, we predict:
Three Boundary Conditions of the Positive University Effect
We do not maintain that university–industry collaborations will always yield positive consumer reactions. Instead, we consider three boundary conditions tied to our focal account: the product (i.e., high tech vs. low tech), the underlying firm (i.e., new vs. established), and the target customer (i.e., high vs. low belief in science). 6 First, not all products benefit equally from the scientific legitimacy engendered by universities. While the positive university effect might be strong for high-tech products where scientific legitimacy is key (i.e., scientific knowledge is indispensable to develop new technology), 7 it might be attenuated for low-tech products where access to the latest scientific ideas is less of a concern. To visualize, consider a padlock that opens with a conventional key (low tech) compared to its “smart” counterpart that utilizes digital fingerprint recognition (high tech). Indeed, research suggests that high-tech products “require technological sophistication” and “scientific know-how,” whereas “low-tech products depend less on science and technology” (Woolley, Kupor, and Liu 2023, p. 429; see also John, Weiss, and Dutta 1999). We thus predict:
Second, we reason that the positive university effect will be more pronounced for new (vs. established) firms. Newness can actually be a liability, as new firms often suffer from a lack of legitimacy in the eyes of consumers (Stuart, Hoang, and Hybels 1999). Specifically, consumers might be reluctant to purchase products from new firms with “no history” due to some uncertainty regarding the quality of their products (Shepherd and Zacharakis 2003; Singh, Tucker, and House 1986). In other words, because new firms lack internal legitimacy (e.g., a history of successful product launches), gaining external legitimacy through university–industry collaborations could be of great benefit.
For established firms, in contrast, collaborating with universities might be considered less impactful because such firms might already be associated with high levels of internal legitimacy (Rao, Chandy, and Prabhu 2008). That is, consumers may infer that established firms can draw from multiple, observable internal sources that validate their legitimacy, including previous successful product launches and an existing, satisfied customer base (Fisher et al. 2017; Maier et al. 2023; Rutherford and Buller 2007). Similarly, established firms might be associated with swift access to financial resources to fund expensive and potentially long-lasting in-house innovation projects, which might include sponsoring basic research and hiring scientists from universities (Lam 2007; Woolley, Kupor, and Liu 2023). Consequently, the positive university effect shall be weaker for established (vs. new) firms. We thus predict:
Third, we consider a situation wherein the valence of the signal might differ. We conjecture that the individual consumer's belief in science might moderate the positive university effect such that the effect is attenuated for consumers scoring low on this trait. “Belief in science” is defined as an individual's belief in the value of science as an institution and source of superior knowledge (Farias et al. 2013). In other words, belief in science can be described as one's “confidence and trust in the validity of scientific methods and outcomes” (Dagnall et al. 2019, p. 2). Individuals substantially differ in their belief in science. For example, while some people are receptive to scientific evidence with regard to specific topics (e.g., climate change, vaccination, genetic modification), others are more skeptical (Rutjens, Sutton, and Van Der Lee 2018). Although this trait has been extensively studied in the fields of cognitive and social psychology (Farias et al. 2013; Rutjens, Sutton, and Van Der Lee 2018), we are not aware of any marketing applications. We reason that one's general belief in science is positively related to both the belief that firms may benefit from collaborating with universities and the belief that science more broadly is critical for firms to generate innovations (a pretest confirmed these relationships; for details, see Web Appendix C). Thus, while individuals with a high belief in science may positively regard university-codeveloped products, the university effect shall be less pronounced for individuals with a low belief in science. More formally:
Overview of Studies
We report eight studies that test our hypotheses (for an overview, see Table 1). Study 1 is an Instagram A/B test assessing the positive university effect (H1a) using actual click-through rates and ad engagement levels as dependent variables. Study 2 is a field experiment that tests whether a product portrayed as codeveloped with a university is perceived as more attractive (H1a), even after an actual product trial. Study 3, an inventive-compatible choice experiment, tests the effect's underlying process; that is, whether university-codeveloped products are more attractive to consumers because of elevated perceptions of scientific legitimacy (H2). Study 4 uses an incentive-compatible willingness-to-pay (WTP) elicitation method to examine consumers’ WTP for university-codeveloped products, compared with products codeveloped by another company (H1b). The subsequent four studies aim to test whether the effect is more pronounced when scientific legitimacy engendered by universities is more important to the focal product (i.e., high-tech vs. low-tech product; H3; Study 5), the underlying company (i.e., a new vs. established firm; H4; Study 6), or the targeted customer (i.e., high vs. low belief in science; H5; Studies 7a and 7b).
Overview of Main Studies.
Notes: The studies reported in the main body of the article are complemented by several supplemental studies (see Web Appendix A). For example, Study S5 shows that the positive university effect also emerges against several other control conditions, Study S6 shows that political orientation may serve as a proxy for belief in science (i.e., the more conservative (liberal) one's political orientation, the lower (higher) one's belief in science), and Study S7 examines how the locus of collaboration (technology vs. aesthetics) moderates the focal effect.
Study 1: An Instagram A/B Test
Study 1 is a social media A/B test conducted on Instagram. The aims are twofold. First, we aim to provide an initial test of the positive university effect (H1a) using actual click-through rates and ad engagement levels as dependent variables. Second, we aim to assess the generalizability of the effect by involving different target populations with regard to university background. While current students and alumni are self-evident target groups for products codeveloped with universities, we argue that the image of universities, and their related scientific legitimacy, will be evident well beyond these target populations.
Method
The study was conducted on Instagram in cooperation with Angles90, the aforementioned startup, marketing the first dynamic training grips worldwide. We used a 32-second video from Angles90 featuring its dynamic training grips in action. Although the video was identical across experimental conditions, we worked with the Angles90 team to create two versions of text to accompany the video (see Web Appendix D). The text was in German because the campaign ran in Germany and Austria. In the treatment condition, the text revealed the focal university–industry collaboration (e.g., “the training grips have been developed by Angles90 together with ergonomics experts from the Technical University (TU) of Munich”). 8
In contrast, the control condition's text did not reveal the focal university–industry collaboration (e.g., “the training grips have been developed by Angles90”). Importantly, we posted the same description and advertisement cover in both conditions to avoid self-selection effects; that is, a user's decision to start watching the video was not subject to our treatment, because said treatment was implemented only shortly after the user started watching the video. A pretest confirmed that the two conditions did not differ in consumer perceptions regarding the ads’ realism, novelty, attitude, or surprise (for details, see Web Appendix C).
The ads were shown on the Instagram mobile app as feed posts (vs. stories). Instagram's advertising platform enables advertisers to target users by demographic, such as level of education. To assess whether consumers’ reactions to our experimental manipulations differed, we had to perform two identical campaigns aimed at different target groups, because Instagram did not allow us to target different populations while simultaneously comparing results across the selected target groups. In the first campaign, we targeted consumers with a university background (i.e., students or alumni), and in a second, otherwise identical campaign, we specifically targeted consumers with no university background. Moreover, in close consultation with Angles90, we chose three restricting characteristics for both target groups. We targeted (1) nonengagers (i.e., users who were not familiar with Angles90 before the campaign), (2) users between ages 20 and 50, and (3) sports enthusiasts. These restrictions correspond to the targeting variables typically used by the underlying brand for social media campaigns.
We followed recent best practice examples in the marketing literature (e.g., Hydock, Paharia, and Blair 2020; Paharia 2020) and configured our campaigns using automatic bidding in which Meta (Instagram’s parent company) determined the optimal bid for each 1,000 impressions. The cost per 1,000 impressions (CPM) is based on Meta's algorithm, which uses an auction mechanism for ad space at any given time. The metric CPM is commonly used by the online advertising industry to compare performances between different ads. To summarize, we created two separate Instagram campaigns with A/B testing, resulting in a 2 (university codevelopment vs. control) × 2 (target group: consumers with university background vs. no university background) between-participant study design. Each condition had a budget of $100 and was established on a CPM basis as described previously.
We set up our campaigns for optimal ad engagement. “Ad engagement” refers to the total number of actions that people take involving the ads while they are running (e.g., sharing, commenting, liking, or saving the post). Instagram only reports ad engagement levels in aggregate, such that one engagement “action” refers to one user having engaged with the ad. Ad engagement serves as a first dependent variable. Moreover, we use unique link clicks as our second dependent variable. “Unique link clicks” refer to the total number of individuals who click on the link provided in the ad. However, to effectively test whether our manipulations have an effect on the dependent variables, we need to take into account how many unique users have seen the ad in the first place; this is referred to as “reach.” 9 As noted by prior research (Hydock, Paharia, and Blair 2020; Paharia 2020), the number of individuals reached may vary considerably between conditions because of variations in marketplace conditions (e.g., different ad competitions).
Thus, the rows of our data file correspond to the individuals reached by our campaigns (see Figure 1). The columns capture the independent and dependent variables. In particular, the variables are coded as follows: university codevelopment = 1, control = 0; consumers with a university background = 1, without a university background = 0; dependent variable: action observed = 1, action not observed = 0. Therefore, we can predict the likelihood of observing the dependent variable action as a function of our manipulations. We can also calculate the share of consumers reached who engaged with the ad (ad engagement level) or clicked on the link (click-through rate).
Study 1 Findings.
Findings
We test our predictions using logistic regression analyses. First, we test for main effects by using both study factors as independent variables. Second, we test for a potential two-way interaction by adding the respective interaction term as a third predictor to the model.
Ad engagement
First, and in support of our predictions, we find a significant positive main effect of the university-codevelopment manipulation on ad engagement (β = .14, SE = .01, odds ratio = 1.16, Wald χ2 = 190.6, p < .001). Second, regarding consumers’ university background, we find a nonsignificant main effect (p = .22). Third, the respective two-way interaction proved significant (β = −.09, odds ratio = .91, Wald χ2 = 20.9, p < .001). However, by decomposing the sample according to university background, findings indicate that the positive university effect on ad engagement holds for individuals with a university background (β = .09, odds ratio = 1.10, Wald χ2 = 36.9, p < .001) and for individuals without a university background (β = .19, odds ratio = 1.21, Wald χ2 = 175.6, p < .001). Interestingly, the university effect is slightly stronger for consumers with no university background. Figure 1, Panel A, depicts the ad engagement levels for both samples and conditions.
Unique link clicks (Click-through rate)
Similar to ad engagement, we also find a significant positive main effect of the university-codevelopment manipulation on unique link clicks (β = .62, odds ratio = 1.86, Wald χ2 = 10.6, p = .001). Regarding consumers’ university background, we find a nonsignificant main effect (p = .59). Moreover, the respective two-way interaction proved nonsignificant (p = .81). Decomposing the sample by university background shows that the positive university effect on the clicks generated holds for individuals with a university background (β = .57, odds ratio = 1.77, Wald χ2 = 4.4, p = .035) as well as individuals without a university background (β = .66, odds ratio = 1.94, Wald χ2 = 6.1, p = .013). Figure 1, Panel B, depicts the click-through rates for both samples and conditions.
Discussion
Based on an A/B test conducted on Instagram, Study 1 provides initial evidence consistent with our primary prediction that consumers react more positively to products codeveloped with a university (H1a). In particular, we find that a product video yields higher ad engagement levels and click-through rates if the product is portrayed as codeveloped with a university, versus developed solely by the underlying firm. Further, we find that the positive university effect is not limited to the most obvious target group: current students and alumni. Instead, the effect is also significant for consumers with no university background, suggesting that the effect is indeed relevant to broader segments of the population. In the next study, we aim to corroborate these initial results by testing whether the focal treatment also affects consumer perceptions after an actual product trial.
Study 2: A Field Experiment in the City Center of Nuremberg
In Study 2, we test the robustness of the positive university effect by changing the test setting from online to offline and focusing on product evaluation and likelihood of recommending the product to others as dependent variables. We further aim to assess whether the positive university effect is also apparent in cases where consumers can actually try the underlying product themselves. This is an ambitious test because the objective product properties are experimentally held constant between conditions, and the product qualities can be directly assessed by study participants. Put differently, finding the effect in this setting would speak for the effect's strength and thus managerial relevance. 10 It would further suggest that the positive university effect might also hold for more central (vs. peripheral) processing (Petty, Cacioppo, and Schumann 1983). Our study sponsor was again Angles90.
Method
Design, procedures, and sample
We conducted a field experiment in the city center of Nuremberg, Germany, where we offered interested passersby the opportunity to test Angles90's dynamic training grips at a fitness pop-up booth. The study was conducted in December 2019 and lasted for four days (from 10a.m. to 6p.m. daily). We used official company merchandise, including a roller banner and official product flyers, provided by Angles90. We also allowed for an onstage product test by setting up a chin-up bar with the grips attached. Importantly, all the booth's features, including the standardized address used to recruit participants in the pedestrian area, were condition-unspecific. Therefore, condition-specific self-selection cannot plausibly be an alternative explanation in this study.
A total of 336 individuals agreed to participate in the product test. Upon agreeing to participate, participants were informed that the product was either codeveloped with a university or developed by the underlying firm alone. The focal treatment was changed every hour (with a quick break between sessions), ensuring that participants were unaware they were part of an experiment or that other versions of the product description existed. Participants received a condition-specific color handout. Similar to the video used in Study 1, the handout featured some standard information about the company and its product as well as the condition-specific manipulations (for details, see Web Appendix D).
We next encouraged participants to touch the grips to get a better feel for the product and test them on the full-height chin-up bar. After they tried the dynamic training grips, we asked participants to complete a short questionnaire to provide us with feedback on the product. Eleven participants in the university condition and ten participants in the control condition chose not to take the survey (p = .60). In addition, 18 participants in the university condition and 14 participants in the control condition left some survey questions unanswered (p = .45). Thus, any variations in the degrees of freedom reported subsequently are due to missing values. 11
11Findings are robust if we instead systematically remove participants with missing values from further analysis.
Measures
The survey contained a series of dependent measures followed by some control variables. The instrument, as well as the manipulations, was administered in German and translated thereafter. We first captured participants’ overall product liking with the following question: “How much do you like the dynamic training grips from Angles90?” (0 = “don’t like them at all,” and 10 = “like them a lot”). Second, we captured participants’ likelihood to recommend the product to others as follows: “How likely is it that you will recommend this product to (sports-interested) friends or colleagues?” (0 = “completely unlikely,” and 10 = “extremely likely”). Third, we captured perceived product sophistication using the following four-item scale (adapted from Brown and Dacin [1997]): (1) “This product is probably more advanced than any other product like it,” (2) “This product features advanced components,” (3) “This is a sophisticated product,” and (4) “This product has a lot of knowledge inside” (1 = “strongly disagree,” and 7 = “strongly agree”; α = .87). If our theoretical reasoning is correct—namely, that university–industry collaborations imbue companies with more scientific legitimacy—products codeveloped with a university should be perceived as more sophisticated. Fourth, we added an open-ended question to explore consumers’ unaided associations (“In your own words, what is your first impression of this product?”).
We next captured a series of controls including the following reading check question: “With whom did Angles90 develop the dynamic training grips?” (0 = “alone,” 1 = “together with a university,” 2 = “I don’t know”). Most participants answered this question correctly (76.9%). We decided in this and all other studies to retain all cases for further analyses. For brevity, we do not report the reading check question in the subsequent studies (the percentage of correct responses is comparably high throughout). The other control measures are summarized in Web Appendix D.
Findings
We first ran an analysis of variance (ANOVA) on overall product liking. In support of H1a, we find that participants in the university-codevelopment condition evaluated the product more favorably (M = 8.03, SD = 1.59) than those in the control condition (M = 7.54, SD = 2.05; F(1, 324) = 5.959, p = .015, η2 = .018). Second, we find participants more likely to recommend the product to others when it is purportedly codeveloped with a university (M = 7.21, SD = 2.09) than when it is merely developed by the underlying startup team (M = 6.28, SD = 2.47; F(1, 320) = 13.306, p < .001, η2 = .040). Third, and consistent with our theorizing, we find that participants perceived the university-codeveloped product to be significantly more sophisticated (Muniversity = 5.18, SD = 1.19; Mcontrol = 4.72, SD = 1.32; F(1, 327) = 11.072, p < .001, η2 = .033).
Discussion
Study 2 provides evidence for the prediction that consumers will evaluate a given product more favorably if they learn it was codeveloped with a university, versus merely developed by the underlying firm (H1a). The same result emerges for consumers’ likelihood to recommend the product to others. Notably, these effects are obtained in a test setting in which all participants were exposed to the same product and tested, thus personally experienced, the product.
Both Study 1 and Study 2 are characterized by high levels of ecological validity, such that the text describing the focal university–industry collaboration is very similar to the text used by the underlying startup (the same holds true for the more general marketing materials). However, one might argue that the resulting effects are in part due to specific associations with the university (TU Munich) and collaborators (ergonomic experts) themselves. To address this limitation, we ran an online follow-up study where we referred to the collaboration in general terms only (“the dynamic training grips were developed, tested, and improved by the Angles90 startup team together with a university”). Participants’ purchase intent served as the dependent variable. Findings are again supportive of H1a: Participants are more likely to purchase the dynamic training grips in the university-codevelopment (vs. control) condition (for details, see Web Appendix A).
Study 3: Scientific Legitimacy as Mediator
Study 3 aims to test our scientific legitimacy account postulated to underlie the positive university effect (H2). For this study, we employed a direct comparison design (Acar et al. 2021; Meyvis and Van Osselaer 2018), used actual product choice as the dependent variable, and changed our underlying product category to cameras. Finally, we recruited a sample of U.S. consumers to test the effect outside the German-speaking market.
Method
Design, procedures, and sample
Participants were 102 U.S. consumers (48.0% female, 49% male, 2% other, 1% prefer not to answer; Mage = 36.9 years; Prolific). We adapted the stimuli from previous research (Acar et al. 2021, Study 1c). Before stimuli exposure, we told participants that the aim of the study was to examine their interest in two products from two different companies: Brand A and Brand B (brand names anonymized for the purpose of the study). Moreover, we informed them of the chance to actually win the product of their choice (i.e., upon completion of the study we would randomly choose one winner to receive the product of their choice). Participants were then given detailed information about two cameras, presented side-by-side to facilitate an effective product comparison (for details, see Web Appendix D). We randomly assigned participants to one of two experimental conditions. For half the participants, we described Camera A as developed by Brand A together with a university (and Camera B as developed by Brand B); for the other half of participants, Camera A was described as developed by Brand A (and Camera B as developed by Brand B together with a university). Participants then indicated their product choice and completed a short survey. Upon completion of the study, we randomly determined the winner and finalized the lottery.
Measures
We first captured product choice with the following question: “Which of the two products do you prefer?” (Camera A, Camera B). Next, we captured our mediator, which was perceived scientific legitimacy. We employed the following four-item scale (based on Rao, Chandy, and Prabhu 2008): (1) “I think this company has access to the latest scientific ideas in the field,” (2) “This company is able to incorporate cutting-edge knowledge into their new product development activities,” (3) “I trust this company is capable of developing successful innovations,” and (4) “I think this company has achieved scientific legitimacy” (1 = “more true for Brand A,” and 7 = “more true for Brand B”; α = .93). We additionally captured two rival mediators. First, we measured perceived novelty with regard to the firm's innovation approach (i.e., one could argue that consumers perceive university–industry collaborations as more novel) with the following two items: (1) “I think the company's innovation strategy is common” (reverse coded), and (2) “The company's approach to develop new products is new” (1 = “more true for Brand A,” and 7 = “more true for Brand B”; r = .33, p < .001). Second, we measured perceptions of the firms’ attractiveness before developing the focal product (i.e., one could argue that higher-quality firms are more likely to convince external collaborators to work with them to begin with) with the following two items: “Think of the time before the companies have developed their products,” (1) “I think this company would have had an easy time to attract others to work with them on joint projects,” and (2) “I think many institutions would have liked to work with this company already at that time” (1 = “more true for Brand A,” and 7 = “more true for Brand B”; r = .45, p < .001).
Findings
Product choice
A logistic regression on actual product choice demonstrates that consumers have a significantly stronger preference for a given camera when it is presented as codeveloped with a university (β = .91, odds ratio = 2.46, Wald χ2 = 4.77, p = .029). To illustrate, when Camera B was presented as codeveloped with a university, 69% of participants chose Camera B (31% chose Camera A). Conversely, when Camera A was presented as codeveloped with a university, only 47% of participants chose Camera B (53% chose Camera A). In other words, the experimental treatment induced an incremental change in the focal product's choice share of 22 percentage points.
Scientific legitimacy
An ANOVA on our mediator shows that participants associate the company Brand B with significantly more scientific legitimacy when it is described as having codeveloped the product with a university (M = 4.79, SD = 1.08) than when described as having developed the product alone (M = 3.75, SD = 1.28; F(1, 100) = 19.774, p < .001, η2 = .165).
Mediation analysis
To test for mediation, we employed bootstrapping procedures (PROCESS Model 4 [Hayes 2017], 5,000 bootstrap samples as in the subsequent studies). Product choice served as dependent variable, the experimental treatment as independent variable (1 = codeveloped with a university, 0 = developed by the underlying firm alone), and perceived scientific legitimacy as mediator. The mediation analysis revealed a significant indirect effect of scientific legitimacy (b = 1.09, SE = .42, CI [95%]: .489, 2.121). We ran another mediation model, this time also including the two rival mediators. Findings once again revealed a significant indirect effect of scientific legitimacy (b = 1.09, SE = .50, CI [95%]: .463, 2.421) and, at the same time, no significant indirect effects of perceived novelty (CI [95%]: −.525, .079) or firm attractiveness (CI [95%]: −.098, .398). Thus, we find support for H2.
Discussion
Study 3 extends the previous findings in several major ways. First, in line with the results from our qualitative study, we find that scientific legitimacy mediates the identified positive university effect, thus confirming H2. Second, we find that the effect cannot be attributed to mere novelty and firm attractiveness perceptions. Third, we replicate the positive university effect in a behavioral choice experiment, in another product context, and with a U.S. sample.
Study 4: Consumers’ WTP for University-Codeveloped Products
Study 4 tests the generalizability of the positive university effect in several important ways. First, we utilized a different product (a smart fingerprint padlock). Second, we recruited a sample of U.K. consumers. Third, we used an incentive-compatible WTP measure as dependent variable, with the prediction that consumers will pay more for a given product when it is portrayed as codeveloped with a university. Fourth, to test H1b, we utilized a different control condition (codeveloped with another company).
Method
Design, procedures, and sample
Participants were 200 U.K. consumers (49.0% female, 49.5% male, 1% other, .5% prefer not to answer; Mage = 39.0 years; Prolific) who were randomly assigned to one of two experimental conditions (codeveloped with a university vs. codeveloped with another company). Before their exposure to the product stimuli, we informed participants of the incentive-compatible nature of the experiment and the purpose of the study, which was to gauge their interest in a smart fingerprint padlock. Specifically, they learned they would be asked about their WTP for the padlock and that they could win one of three $30 vouchers in a raffle. Participants also learned that if they are among the raffle winners, their WTP for the focal product would have real-life consequences (i.e., they would be able to purchase the padlock; for further details, see the following subsection). In both conditions, participants saw a screenshot of a shopping website featuring the padlock. The stimuli contained several pictures and product descriptions (from an actual shopping website that sells padlocks) that were, apart from the focal manipulation, identical across conditions (for details, see Web Appendix D). While participants in the university-codevelopment condition read that the padlock was developed in collaboration with a university, those in the control condition read that the product was developed in collaboration with another company.
Measures
Participants’ WTP for the padlock served as dependent variable. We adopted the procedure from Becker, DeGroot, and Marschak (1964), which is a commonly used marketing research method for eliciting consumers’ WTP in an incentive-compatible fashion (Wertenbroch and Skiera 2002). Specifically, participants submitted their bids for the product in $.1 increments with the maximum bid being $30 (the prize amount in the raffle). If participants had no interest in the product (i.e., they did not want to bid for it), we coded their WTP as 0.
Participants also learned about the auction procedure. Specifically, we informed them that if they won the $30, we would draw a random price upon completion of the study. If the price drawn was lower than or equal to their bid, the bid would be accepted, and they would receive the padlock at the randomly determined price. In that case, participants would also receive the remaining amount ($30 minus the randomly determined price). If, by contrast, the price drawn was higher than their bid, the bid would not be accepted, and they would not receive the padlock. Instead, they would receive the full $30. We adopted this double-lottery procedure from prior marketing research (e.g., Acar et al. 2021). Upon completion of the study, we randomly determined the three winners, drew the random prices, and finalized the auction (i.e., sent them the product/money).
Findings
An ANOVA on WTP provides support for the positive university effect (H1b). Participants were willing to pay significantly more for the padlock when it was described as codeveloped with a university versus another company (Muniversity = 12.00, Mcompany = 7.30; F(1, 198) = 12.20, p < .001, η2 = .058).
Discussion
Study 4 complements our previous findings in several major ways. First, we test the positive university effect among another study population (U.K. consumers). Second, we employ an incentive-compatible WTP measure as dependent variable. The effect found appears to be managerially meaningful: consumers are willing to pay, on average, almost 65% more for a product promoted as codeveloped with a university. Notably and in support of H1b, the effect emerges against a control group that portrayed the product as codeveloped with another company, suggesting that the effect is unique to universities as an external collaboration partner. This conclusion is further supported by a follow-up study reported in more detail in Web Appendix A, which replicates the positive university effect in another product context (airbag for cyclists), using a different dependent variable (purchase intent), and importantly, a series of control groups (product developed by the underlying firm alone, codeveloped with another company, a nonprofit organization, or a nonuniversity research institute). Having established the positive university effect in the first four studies, we next turn to testing the related moderators.
Study 5: Product Type as Moderator
In Study 5, we aim to test whether the positive university effect will be attenuated when the underlying product is low tech (vs. high tech). We test this prediction (H3) by using the same incentive-compatible WTP elicitation method as in Study 4.
Method
Design, procedures, and sample
Participants were 785 U.K. consumers (49.9% female, 49% male, 1% other, .1% prefer not to answer; Mage = 41.4 years; Prolific). The experiment used a 2 (codevelopment: university vs. another company) × 2 (product type: high tech vs. low tech) between-participant design. Before exposing participants to the product stimuli, we informed them of the incentive-compatible nature of the experiment (i.e., we informed all participants that they could win one of three $30 vouchers in a raffle) and purpose of the study, which was to gauge their interest in a smart fingerprint (vs. classical) padlock. In the high-tech condition, we used similar stimuli as in Study 4. In the low-tech condition, we showed participants a technologically less sophisticated, more conventional padlock (for details, see Web Appendix D).
Measures
As indicated, we employed the same WTP measure as in Study 4. Upon study completion, we randomly determined three winners, drew the random prices, and finalized the auction procedure (i.e., we sent them the product/money).
Findings
We performed a 2 (codevelopment: university vs. another company) × 2 (product: high tech vs. low tech) ANOVA on WTP. Findings reveal a significant main effect of the first experimental factor (F(1, 781) = 6.999, p = .008, η2 = .009); participants in the university-codevelopment condition were willing to pay significantly more (M = 9.52) than participants who evaluated the product codeveloped with another company (M = 7.82). Moreover, we find a significant main effect of product type (F(1, 781) = 14.352, p < .001, η2 = .018), such that participants were willing to pay significantly more in the high-tech condition (M = 9.91) than in the low-tech condition (M = 7.46). Most importantly and as expected, findings also reveal a significant interaction effect (F(1, 781) = 9.484, p = .002, η2 = .012; see Figure 2). Follow-up contrasts reveal that participants were willing to pay significantly more for the high-tech padlock when it was described as codeveloped with a university (Muniversity = 11.74) versus codeveloped with another company (Mcompany = 8.05; F(1, 781) = 16.204, p < .001, η2 = .020). However, for the low-tech padlock, participants reported no significant difference in their WTP (Muniversity = 7.32, Mcompany = 7.59; p = .758). Thus, we find support for H3.
Product Type as Moderator of the Positive University Effect (Study 5).
Discussion
In support of H3, Study 5 highlights a boundary condition of the positive university effect: said effect is attenuated when the underlying product is low tech (vs. high tech). Notably, the high-tech contrast closely replicates the findings obtained in Study 4. That is, the respective WTP increment increased to almost 46%, underscoring once again the managerial significance of the effect. In the next study we examine firm type (new vs. established) as a second moderating factor.
Study 6: Firm Type as Moderator
In Study 6, we test whether the positive university effect will be attenuated when the underlying company is an established (vs. new) firm. We test this prediction (H4) in the context of virtual reality (VR) glasses.
Method
Design, procedures, and sample
Participants were 444 U.S. consumers (44.8% female, 55.2% male; Mage = 36.8 years; Amazon Mechanical Turk). The experiment used a 2 (codevelopment: university vs. another company)×2 (firm type: new vs. established) between-participant design. Participants were randomly assigned to one of four experimental conditions. In the new firm condition participants learned about Brand A, a startup company new to the VR market. In the established firm condition participants learned about Brand A, an established company and one of the most experienced firms in the VR market. We further informed participants that Brand A recently developed the VR glasses Views and the company worked intensively with an external partner on the development of the product. Participants in the university (control) condition learned that “Brand A developed its new product in collaboration with a university (another company).” Next, participants received visual and verbal descriptions of the company's new VR glasses (for details, see Web Appendix D).
Measures
Participants’ purchase intent served as dependent variable and was captured with a three-item scale: “If you needed VR glasses now, would you consider purchasing the VR glasses from Brand A?” (1) “I would definitely consider purchasing the VR glasses from Brand A,” (2) “I would be very likely to purchase these VR glasses,” and (3) “I think purchasing these VR glasses would be an excellent choice for me” (1 = “strongly disagree,” and 7 = “strongly agree”; α = .94).
Findings
We performed a 2 (codevelopment: university vs. another company) × 2 (firm type: new vs. established) ANOVA on our dependent variable. We first found a significant main effect of the initial experimental factor (F(1, 440) = 4.027, p = .045, η2 = .009); participants in the university-codevelopment condition reported significantly higher purchase intentions (M = 5.04, SD = 1.35) than participants who evaluated the product codeveloped with another company (M = 4.76, SD = 1.52). Findings revealed no significant main effect of firm type (p < .94). Most importantly (and as hypothesized in H4), findings revealed a significant interaction effect (F(1, 440) = 9.042, p = .003, η2 = .020; see Figure 3). Follow-up contrasts revealed a positive and significant university effect in the context of a new firm (Muniversity = 5.25, SD = 1.14; Mcompany = 4.57, SD = 1.63; F(1, 440) = 12.684, p < .001, η2 = .028), but not an established firm (Muniversity = 4.83, SD = 1.51; Mcompany = 4.97, SD = 1.39; p > .48). Thus, we find support for H4.
Firm Type as Moderator of the Positive University Effect (Study 6).
Discussion
In support of H4, Study 6 demonstrates that the positive university effect is attenuated when the underlying company is an established (vs. new) firm. In particular, we find that the positive university effect is highly significant in the context of a startup; however, in the context of an established company (i.e., with pertinent experience relating to the underlying product market), the effect is attenuated. In the final three studies, we examine the predicted moderator regarding the consumer: namely, their belief in science (H5).
Study 7a: Consumers’ Belief in Science as Moderator (Measured)
The aim of Study 7a is to test whether the positive university effect is moderated by individual consumers’ belief in science, such that the effect will be particularly pronounced (attenuated) for consumers scoring high (low) in this regard (H5).
Method
Design, procedures, and sample
Participants were 594 U.S. consumers (45.6% female, 54.4% male; Mage = 38.0 years; Amazon Mechanical Turk). Participants were randomly assigned to either the university-codevelopment condition or the external company–codevelopment condition. We used an “airbag for cyclists” as the underlying product category (for details, see Web Appendix D).
Measures
We first captured participants’ purchase intentions with the same scales as in our previous studies (α = .91). Next, we measured participants’ belief in science using a ten-item scale taken from Farias et al. (2013). An exemplary item reads, “Science provides us with a better understanding of the universe than does religion” (1 = “strongly disagree,” and 7 = “strongly agree”; α = .94; for all items, see Web Appendix D).
Findings
Purchase intentions
An ANOVA on our dependent variable reveals that participants demonstrate significantly higher purchase intentions when the focal product is described as codeveloped with a university (M = 5.44, SD = 1.21) than when the focal product is described as codeveloped with another company (M = 5.18, SD = 1.29; F(1, 592) = 6.183, p = .013, η2 = .010).
Moderating effect of belief in science
To test for moderation, we used bootstrapping procedures (PROCESS Model 1 [Hayes 2017]); purchase intent served as a dependent variable, the experimental treatment as independent variable (1 = codeveloped with a university, 0 = codeveloped with another company), and the belief in science index as moderator. As expected, results reveal a positive interaction effect (b = .19, SE = .08, p = .013). To decompose this interaction, we used the Johnson–Neyman technique (Johnson and Neyman 1936) to identify the range of belief in science for which the treatment effect is (vs. is not) significant. We find a significantly positive university effect among participants with a belief in science score greater than or equal to 4.95 (this corresponds to 64.8% of participants; Mgrand = 5.19, SD = 1.11). Below that level, the effect becomes nonsignificant and even reverses direction with regard to very low belief in science scores (see Figure 4). Thus, we find support for H5.
Belief in Science as Moderator of the Positive University Effect (Study 7a).
Discussion
Study 7a highlights another moderator of the positive university effect. In support of H5, Study 7a demonstrates that the positive university effect is particularly pronounced (attenuated) for consumers who have a high (low) level of belief in science. To further corroborate the moderating role of consumers’ belief in science, we utilize a manipulated moderation approach in Study 7b.
Study 7b: Consumers’ Belief in Science as Moderator (Manipulated)
The aim of Study 7b is to extend the findings obtained in Study 7a by manipulating, rather than measuring, the focal moderator variable (i.e., belief in science).
Method
Seven hundred nineteen U.S. participants (49.7% female, 50.3% male; Mage = 37.9 years; Amazon Mechanical Turk) were randomly assigned to conditions in a 2 (codevelopment: university vs. external company) × 2 (belief in science: high vs. low) between-participant design. To prime high (vs. low) belief in science, we asked participants to carefully read one of two versions of a summarized article, ostensibly published in a newspaper with global reach. More specifically, in the high (low) belief in science condition, participants read that “science is (not) the holy grail of our society” and that science is (not) an important driver for our well-being (for details, see Web Appendix D). After reading the article summary, participants were asked to think of one to three reasons the author of this article might be correct; this in turn served to highlight the key advantage (disadvantage) of a predominantly science-driven society. We followed recent research (e.g., Acar et al. 2021) and used participants’ answers as a reading check. 12 We then asked participants to complete the same belief in science scale as used in Study 7a; their answers served as a manipulation check, which proved effective (Mhigh BIS prime = 5.34, SD = 1.16; Mlow BIS prime = 4.64, SD = 1.40; F(1, 647) = 47.698, p < .001, η2 = .069). In a purportedly unrelated study (Study 7b), participants subsequently completed the same study as in Study 7a, except a hoverboard was used as our underlying product category (for details, see Web Appendix D).
Findings
A 2 (codevelopment: university vs. external company) × 2 (belief in science: high vs. low) ANOVA on our dependent variable reveals two nonsignificant main effects (ps > .08) and, as expected, a significant interaction effect (F(1, 645) = 6.533, p = .011, η2 = .010; see Figure 5). In support of H5, follow-up contrasts demonstrate a significant positive university effect in the high belief in science prime condition (Muniversity = 5.17, SD = 1.42; Mcompany = 4.65, SD = 1.73; F(1, 645) = 8.736, p = .003, η2 = .013), but not in the low belief in science condition (Muniversity = 4.64, SD = 1.67; Mcompany = 4.75, SD = 1.52; F(1, 645) = .423, p = .516, η2 = .001). Moreover, follow-up contrasts reveal that participants in the high belief in science prime condition reported higher purchase intentions for the university-codeveloped product than those in the low belief in science prime condition (F(1, 645) = 9.079, p = .003, η2 = .014). The respective difference proved insignificant for the product described as codeveloped with another company (F(1, 645) = .364, p = .547, η2 = .001).
Belief in Science as Moderator of the Positive University Effect (Study 7b).
Discussion
Study 7b provides further evidence for the moderating role of belief in science (H5). In accordance with Study 7a, we find that the positive university effect is particularly strong (attenuated) for consumers with a high (low) belief in science. Although belief in science can be readily measured by market researchers (see Study 7a), it is hardly known a priori (e.g., social media platforms do not [yet] offer it as a targeting variable). Against this backdrop, we were interested in identifying a managerially accessible proxy for belief in science.
In particular, previous research showed that political orientation is a robust predictor of belief in science (e.g., Brzezinski et al. 2020; Gauchat 2012; Rutjens, Sutton, and Van Der Lee 2018). Accordingly, we predict that the positive university effect will be particularly pronounced (attenuated) for liberals (conservatives). We performed a follow-up study to test this prediction (for details, see Web Appendix A). Findings are affirmative: while the positive university effect is significant among liberals, it is attenuated (and directionally even reversed) among conservatives (see Figure 6). Thus, marketing university-codeveloped products as such might be particularly promising when targeting the focal product to liberals.
Political Orientation as Moderator (Study S6).
General Discussion
Summary, Theoretical Contributions, and Practical Implications
Our studies offer three major insights. First, we identify a positive university effect: consumers perceive a given product as more attractive when it is portrayed as developed in collaboration with a university. Second, consistent with our theorizing, we find that collaborating with a university infuses the underlying firm with a stronger sense of scientific legitimacy, thereby making the resulting product more attractive to consumers. Third, the positive university effect is found to be more pronounced when the scientific legitimacy conferred by universities is more important to the (a) focal product (i.e., high tech vs. low tech), (b) underlying company (i.e., new vs. established), or (c) target customer (i.e., high vs. low belief in science).
In short, we provide a novel perspective on university–industry collaborations and show that collaborating with universities might be beneficial for not only innovation but also marketing purposes. The practical implications are immediate (see the decision tree depicted in Figure 7): firms that engage in open innovation practices with universities might leave some economic value on the table if they fail to broadly communicate said collaboration to prospective customers. In particular, managers should consider using labels such as “codeveloped with a university” or “university knowledge inside” and thereby incrementally increase the underlying product's market performance. Indeed, the strength of the effects found underscores the managerial significance of the positive university effect. For example, participants in Study 4 were willing to pay, on average, 65% more for the same product when it was portrayed as codeveloped with a university. Furthermore, the boundary conditions identified help managers anticipate when actively marketing university–industry collaborations will be more (vs. less) effective. Marketing products as codeveloped with a university might be particularly promising for new firms, when the underlying product is high tech, or when the target customer scores high on belief in science. Regarding the latter, we have utilized Farias et al.’s (2013) ten-item scale. If managers think the scale is too long, they can employ a shorter version. Indeed, we reanalyzed our Study 7a data by relying on the top one (top three) item(s) in terms of item-to-total correlation (see Web Appendix D). Findings were affirmative: the focal interaction emerges when only one (three) item(s) are used. Relatedly, we identified a proxy variable that is managerially accessible: consumers’ political orientation. That is, because belief in science is markedly related to political orientation, we found that the positive university effect emerges strongly for liberals but not for conservatives. Thus, marketing university-codeveloped products as such might be particularly promising when targeting the focal product to liberals.
Decision Tree for Managers.
In addition to contributing to the theory and practice of university–industry collaborations, our findings also bear notable implications for the following literatures:
Legitimacy. In contrast to Rao, Chandy, and Prabhu (2008), our findings highlight that scientific legitimacy is relevant to not only investors but also consumers. In addition, we show that university–industry collaborations are an effective means to drive scientific legitimacy perceptions, suggesting that project-based collaborations (vs. the more formal approach of having academic scientists on the firm's board) might also bolster a firm's scientific legitimacy. Furthermore, our analysis effectively isolates the subjective positive university effect from one that is based on factual differences of the resulting innovation. Finally, while the analysis of Rao, Chandy, and Prabhu suggests that new ventures might benefit more from forming alliances with other companies versus hiring scientists, our findings suggest the opposite: collaborating with a university might be preferable to working with another firm. Of course, there are several factors that could potentially account for these different conclusions (e.g., we kept objective product quality constant, we focused on consumers); however, one notable advantage of our analysis is that we effectively isolated the source from the type of collaboration. In sum, we see our research as complementary to theirs, and stimulating for further investigations pertaining to legitimacy. Cobranding and ingredient branding. While existing research (Desai and Keller 2002; Pinello, Picone, and Destri 2022) has mostly studied specific brands and other firms as cobranding partners, we point to the opportunity to investigate effects of general types of institutions. Put differently, success of cobranding has been thought to be contingent on the specific brand associations held by consumers (e.g., Intel inside Lenovo). Our research highlights that cobranding might go beyond firms, and the type of collaboration partner per se (e.g., company vs. university) might matter in shaping the respective consumer response. Source credibility. Research (e.g., Pornpitakpan 2004) suggests that source credibility effects are particularly pronounced in the context of indirect (vs. direct) experiences (i.e., when the information is peripherally processed; Petty, Cacioppo, and Schumann 1983). Put differently, when consumers “have direct experience with the object, source credibility tends to have little effect on persuasion” (Pornpitakpan 2004, p. 255). Interestingly, our Study 2 findings imply that the positive university effect also unfolds after an actual product trial in the context of direct experience. In tandem with our other findings (e.g., more pronounced ad engagement and click-through rates in the course of an Instagram field experiment, strong effects on consumers’ actual WTP), we believe this suggests that the focal effect is substantial and withstands a cognitive processing check on the part of the consumer. One unique feature of the source (university) in this study is that it causally affected the object (codeveloped product); this differs from most other expert source effects in the literature where the source often comes as an afterthought (e.g., this product is recommended by an expert). From this perspective, as noted previously, our setting is similar to the one studied by Plassmann et al. (2008) and Lee, Frederick, and Ariely (2006). If there is a comprehensible connection between source and object (a wine's price, a few drops of vinegar in a beer, a university-codeveloped product), effects might be particularly pronounced and even visible in case of central processing and direct experience.
13
Belief in science. Research in this space has been primarily the purview of cognitive and social psychology (Farias et al. 2013; Rutjens, Sutton, and Van Der Lee 2018). We conjecture that belief in science might be relevant to marketing research even beyond the scope of the present investigation. By demonstrating that belief in science can be not only chronically measured but also temporarily activated (i.e., primed), we offer hands-on possibilities for investigating the construct's relevance to consumer behavior and marketing. For example, in the context of the COVID-19 pandemic, research suggested that belief in science is positively related to physical distancing (Brzezinski et al. 2020), mask-wearing behavior (Stosic, Helwig, and Ruben 2021), and vaccination intentions (Bleakley et al. 2022). Our research opens the construct's potential scope of application by highlighting that belief in science affects marketing-relevant downstream variables.
Limitations and Avenues for Future Research
Because our theorizing centered on the brand image of universities in general, apart from Studies 1 and 2, we tested and identified the positive university effect without providing a specific university brand name (e.g., “this product has been codeveloped with a university”). Thus, we did not actively distinguish between different university “brands” (e.g., Harvard University vs. less prestigious universities) or different faculties within universities (e.g., engineering vs. humanities). Going forward, scholars might take up these nuances to see if they impact the size (and maybe sign) of the effects reported here.
Relatedly, it might be interesting to investigate the type of innovation tasks for which universities are considered more (vs. less) helpful. For example, will consumers perceive universities as helpful in collaborating on aesthetics (vs. technology) of new products? Our initial investigation on this topic (see follow-up study S7, Web Appendix A) suggests that the effect indeed depends on the locus of innovation: specifically, the positive university effect emerges strongly when the focal collaboration focuses on technology, whereas it fully reverses when the focus is on aesthetics. What about services (vs. products), digital (vs. physical), or incremental (vs. radical) innovation?
Another promising direction for future research might be the contrast of university–industry collaborations with other contemporary modes of open innovation. For example, recent research shows that involving users in the design of new products via crowdsourcing might yield promising effects from both an innovation and marketing perspective (e.g., Nishikawa et al. 2017). As demonstrated by Schreier, Fuchs, and Dahl (2012), marketing crowdsourced new products as “user-ideated” resonates well with consumers for low-complexity products; however, as complexity increases, consumers’ trust in other users vanishes. Against this backdrop, and in light of the findings presented in this research, we argue that the positive university effect has unique qualities. To start exploring this notion, we also included a “codeveloped with users” condition in our follow-up study S7 (Web Appendix A). Indeed, we find that if the critical task is to operate a high-tech product, university–industry collaborations are more convincing for consumers compared to codevelopment efforts with users. Conversely, however, working with users resonates comparatively better with consumers if the collaboration centers on aesthetics.
Future research might also build on our initial belief-in-science finding. In particular, apart from political orientation, what other proxies are available for managers who aim to effectively target university-codeveloped products to the respective consumer? Rutjens, Sutton, and Van Der Lee (2018), for example, find religiosity to be another important belief-in-science predictor. Findings from our prestudy for Study S6 (for details, see Web Appendix C) confirm this relationship: the more religious a given consumer is, the lower their belief in science (and vice versa, r = −.580, p < .001). Future research could thus test whether religiosity is another managerially accessible moderator of the positive university effect. Relatedly, prior research points to between-country variance with regard to belief in science. In the Netherlands, for example, people tend to trust science and its institutions more than various other institutions, including media, government, and courts of law (Rutjens et al. 2018). In contrast, there are other countries, such as Guatemala or Bosnia and Herzegovina, with a very low belief in science, generally speaking (Gallup 2019). Against this backdrop, another interesting question would be whether the positive university effect documented in this research systematically differs between countries as a function of their citizens’ belief in science.
Future research could shed more light on the exact mechanism linking scientific legitimacy perceptions of the underlying firm to product attractiveness. In particular, as suggested by our qualitative exploration, university-codeveloped products may be perceived differently; that is, they may be perceived as more sophisticated (as shown in Study 2) and trustworthy, ceteris paribus, and these perceptual effects may help explain the favorable consumer response documented in this research (for an initial exploration in this direction, see follow-up study S8, Web Appendix A). Relatedly, future research might examine the dynamic effects following a given university–industry collaboration. Would consumers evaluate a subsequent new product by the underlying firm differently, even if said product was developed by the underlying firm alone?
Similarly, it would be interesting to extend the dependent variables investigated in this research. For example, the social media A/B test reported in Study 1 was designed to test effects on ad engagement and click-through rates. Would the focal treatment also trigger more interest in the ad and the brand's products to begin with? If so, this would bear further implications for the active marketing of university–industry collaboration across various touch points in the customer journey.
Finally, and more broadly, our literature review on university–industry collaborations suggested that there are still ample research opportunities centered on core innovation questions. While there is suggestive evidence that firms might benefit from university–industry collaborations (in terms of yielding objectively better innovations), the existing evidence is merely correlational and mostly based on firm-level (vs. product-level) data. Thus, it is not yet established whether working with universities has a causal effect on the resulting new products’ innovativeness and eventual (long-term) market performance. More specifically, while research has looked into how to engage universities (Ankrah and Omar 2015), there is less research available that prescribes when to expect the highest returns (see Web Appendix B). The moderators identified in this research (e.g., high-tech vs. low-tech products, technology vs. aesthetic collaboration focus, new vs. established firms) provide a fruitful starting point for future research in this direction.
Supplemental Material
sj-pdf-1-jmx-10.1177_00222429231185313 - Supplemental material for University Knowledge Inside: How and When University–Industry Collaborations Make New Products More Attractive to Consumers
Supplemental material, sj-pdf-1-jmx-10.1177_00222429231185313 for University Knowledge Inside: How and When University–Industry Collaborations Make New Products More Attractive to Consumers by Lukas Maier, Martin Schreier, Christian V. Baccarella and Kai-Ingo Voigt in Journal of Marketing
Footnotes
Acknowledgments
The authors thank Angles90, in particular Simon Sparber, for collaborating with us on the field studies and students enrolled in the Research Seminar (Fall 2019) at Friedrich-Alexander-Universität Erlangen-Nürnberg for their help in data collection (Study 2). They further thank the JM review team for their constructive feedback and guidance, as well as WULABS and the Dr. Theo and Friedl Schöller Research Center for generous financial support.
Associate Editor
Roland Rust
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by WULABS (Vienna University of Economics and Business) and the Dr. Theo and Friedl Schöller Research Center.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
