700
1
0
2024-12-19 16:52:07
2024-12-19
大学-企业互动如何影响企业创新速度？——以中国科技型中小企业为例
引用：
[1]Zhang J A, O'Kane C, Bai T. How do university-firm interactions affect firm innovation speed? The case of Chinese science-intensive SMEs[J]. Research Policy, 2024, 53(7): 105027.
[1]Zhang, J. A., O'Kane, C., & Bai, T. (2024). How do university-firm interactions affect firm innovation speed? The case of Chinese science-intensive SMEs. Research Policy, 53(7), 105027.
摘要
This study examines how university-firm (U-F) interactions affect innovation speed in science-intensive small and
medium-sized firms (SISMEs). We distinguish between formal and informal U-F interactions and build on dynamic capability theory to argue that (1) U-F的R&D alliances enhance innovation speed through firm-level
entrepreneurial orientation (EO), and (2) frequent U-F informal contacts weaken the effects of U-F的R&D alliances on innovation speed. Analyzing a sample of 268 SISMEs from 10 science parks in China, the results of the partial least squares structural equation modeling (PLS-SEM) support our hypotheses. Furthermore, fuzzy-set
qualitative comparative analysis (fsQCA) identifies various configurations of U-F的R&D alliances, U-F informal
contacts and EO, along with other organizational, science park and environmental conditions, that lead to higher
or lower innovation speed in SISMEs. Our findings offer valuable theoretical and practical insights, advancing
our understanding of the complex relationship between U-F interactions and innovation speed in SISMEs.
本研究考察了大学-企业（U-F）互动如何影响科学密集型中小企业（SISMEs）的创新速度。我们区分了正式和非正式的U-F互动，并建立在动态能力理论的基础上，认为（1）U-F的R&D联盟通过企业层面的创业导向（EO）提高了创新速度，（2）频繁的U-F非正式接触削弱了U-F的R&D联盟对创新速度的影响。分析了来自中国10个科技园的268个中小企业，偏最小二乘结构方程模型（PLS-SEM）的结果支持了我们的假设。此外，模糊集定性比较分析（fsQCA）确定了U-F的R&D联盟、U-F非正式联系和EO的各种配置，以及其他组织、科学园和环境条件，这些条件导致SISMEs中更高或更低的创新速度。我们的发现提供了有价值的理论和实践见解，推进了我们对SISMEs中U-F相互作用和创新速度之间复杂关系的理解。
文章分析了中小企业实施大学-企业互动对于企业1创新速度的影响，利用实证回归验证假设，用fsQCA探讨中小企业产学互动对于创新速度影响的具体路径。
Keywords: U-F的R&D alliances；U-F informal contacts；Entrepreneurial orientation；Innovation speed；Science-intensive SMEs
关键词：U-F研发联盟；U-F非正式联系创业取向创新速度；科学密集型中小企业
1.引言
Within the pharmaceutical industry, the development and launch of the COVID-19 vaccine underscores the critical role of innovation acceleration in securing strong market positions and generating substantial economic and social impacts (Cooper, 2021; Rosa, 2021). In fact, innovation speed is highly relevant in any science-intensive and hightech market, where being first-to-market is paramount (Chen et al., 2012; Markman et al., 2005). Despite captivating the attention of researchers and practitioners for some time, the factors driving innovation speed remain one of the least understood aspects in innovation literature (Ferreras-M´endez et al., 2022). Our study seeks to enrich this literature by exploring the influence of University-Firm (U-F) interactions on innovation speed in science-intensive small and medium-sized enterprises (SISMEs).
在制药行业，新冠肺炎疫苗的开发和推出强调了创新加速在确保强大的市场地位和产生重大经济和社会影响方面的关键作用。事实上，创新速度在任何科学密集型和高科技市场都是高度相关的，在这些市场中，率先上市是至关重要的。尽管一段时间以来吸引了研究人员和实践者的注意，但驱动创新速度的因素仍然是创新文献中最不为人所知的方面之一。我们的研究旨在通过探索大学-企业（U-F）互动对科学密集型中小企业（SISMEs）创新速度的影响来丰富这一文献。
新冠疫情案例引出创新速度很重要，什么驱动创新需要进一步研究
SISMEs strive for competitiveness through rapid innovations by leveraging unique technical knowledge (George et al., 2002). However, their small size constrains their ability to develop internal resources and capabilities, posing challenges to their innovation efforts (Vanacker et al., 2014). To compensate, SISMEs often turn to universities as major sources of advanced knowledge (Mindruta, 2013) and initiate formal and/or informal U-F interactions to acquire scientific knowledge and technological resources (Díez-Vial and Montoro-Sanchez, ´ 2016; Schaeffer et al., 2020). U-F formal interactions are established through contractual agreements, including R&D alliances and patent licensing (Azagra-Caro et al., 2017; Landry et al., 2010), whereas U-F informal interactions are comprised of non-contractual linkages, such as individual contacts between firm employees and university staff, and participation in academic events (e.g., seminars, conferences, workshops, etc.) (Azagra-Caro et al., 2017; Dahl and Pedersen, 2004; Di´anezGonzalez ´ and Camelo-Ordaz, 2019).
SISMEs利用独特的技术知识，通过快速创新争取竞争力。然而，它们的小规模限制了它们开发内部资源和能力的能力，对它们的创新努力构成了挑战。作为补偿，SISMEs经常求助于大学作为先进知识的主要来源，并发起正式和/或非正式的U-F互动，以获取科学知识和技术资源。U-F正式互动是通过合同协议建立的，包括R&D联盟和专利许可，而U-F非正式互动由非合同联系组成，如公司员工和大学员工之间的个人接触，以及参与学术活动（如研讨会、会议、讲习班等）。
中小企业是重要的创新主体，产学联合是重要的技术获取，创新速度提升手段
Among these channels, U-F的R&D alliances are most often utilized by science-intensive firms, and a substantial body of research acknowledges the significant relationship between such alliances and firm innovation (Caloghirou et al., 2021; Soh and Subramian, 2014). Nevertheless, the nature of this relationship remains ambiguous, yielding mixed findings that range from positive (Melnychuk et al., 2021; Soh and Subramian, 2014) to negative (Bruneel et al., 2010; He et al., 2021; Zhang et al., 2022b), with others suggesting a U-shaped relationship (Caloghirou et al., 2021). This inconsistency underscores both the complexity of the U-F的R&D alliances-innovation relationship in SISMEs, and the need to examine this relationship more closely. Furthermore, while a few previous studies highlight the pivotal role of F informal interactions in knowledge transfer (Apa et al., 2021; Díez-Vial and Montoro-Sanchez, ´ 2016), to date, we know little about how U-F informal interactions help to accelerate firm innovation. More critically, the interplay between these U-F informal interactions and formal R&D alliances in driving innovation speed remains underexplored (AzagraCaro et al., 2017; Schaeffer et al., 2020).
在这些渠道中，U-F的R&D联盟最常被科学密集型企业利用，大量研究承认这种联盟和企业创新之间的重要关系。然而，这种关系的性质仍然模棱两可，产生了从积极到消极的混合发现到消极，其他人提出了U形关系。这种不一致强调了U-F的R&D联盟与SISMEs创新关系的复杂性，以及更仔细地研究这种关系的必要性。此外，虽然先前的一些研究强调了U-F非正式互动在知识转移中的关键作用，迄今为止，我们对U-F非正式互动如何帮助加速企业创新知之甚少。更关键的是，这些U-F非正式互动和正式R&D联盟在推动创新速度方面的相互作用仍未得到充分探索。
文献梳理，文献回顾，点明研究缺口，以及本文研究意义
We attempt to address these issues by developing a fine-grained framework to investigate how U-F interactions affect firm innovation speed. Grounded in dynamic capability theory (Teece et al., 1997), innovation speed can be conceptualized as the firm's ability to accelerate the innovation process from idea generation to commercialization (Cankurtaran et al., 2013; Markman et al., 2005). Moreover, dynamic capability theory highlights the importance of organizational resourceoriented mechanisms through which firms integrate and reconfigure both external and internal assets to generate innovative outcomes and seize opportunities in rapidly changing environments (Teece, 2007). Building on this logic, we initially investigate how SISMEs harness U-F的R&D alliances to access advanced technical knowledge and employ entrepreneurial mechanisms to exploit these external resources effectively, thereby augmenting their innovation speed. We contend that firm-level entrepreneurial orientation (EO) represents such a mechanism because EO, characterized by innovativeness, risk-taking, and proactiveness, is a central function that allows firms to reconfigure and exploit critical resources inside and outside the organization (Brouthers et al., 2015; Wiklund and Shepherd, 2003) for innovation and changes to competitive positioning (Anderson et al., 2015; Covin and Slevin, 1991; Zhang et al., 2020). Subsequently, we explore the interplay between U-F informal contacts and U-F的R&D alliances, proposing their substitutive effects on innovation speed.
我们试图通过开发一个细粒度的框架来研究U-F相互作用如何影响企业创新速度来解决这些问题。基于动态能力理论，创新速度可以概念化为企业加速从创意产生到商业化的创新过程的能力。此外，动态能力理论强调了组织资源导向机制的重要性，通过这种机制，企业整合和重新配置外部和内部资产，以产生创新成果，并在快速变化的环境中抓住机会。基于这一逻辑，我们首先研究了SISMEs如何利用U-F的R&D联盟来获取先进的技术知识，并采用创业机制来有效地利用这些外部资源，从而提高他们的创新速度。我们认为，企业层面的创业导向（EO）代表了这样一种机制，因为以创新性、冒险和主动性为特征的EO是允许企业重新配置和利用组织内外关键资源的核心功能用于创新和竞争定位的变化。随后，我们探讨了U-F非正式联系和U-F的R&D联盟之间的相互作用，提出了它们对创新速度的替代效应。
Our study makes at least three significant contributions to the literature. First, we advance research on the U-F interactions-innovation relationship by emphasizing a specific indictor of innovation: innovation speed. Broadly, within the innovation literature, innovation can be comprised of two components and phases, early stage “innovation inputs” (e.g., results of R&D and human resources) and later stage “innovation outputs” (e.g., outcomes of product and process innovations) (Janger et al., 2017, p. 31; OECD, 2015). However, within this literature, prior research has primarily focused on how U-F的R&D alliances affect input proxies of innovation, such as R&D performance (Melnychuk et al., 2021), number of patents (Chai and Shih, 2016; George et al., 2002), and technology newness (Wirsich et al., 2016), with only a few studies examining their effects on output proxies of innovation, such as the number of product and process innovations (e.g., Caloghirou et al., 2021; Díez-Vial and Montoro-Sanchez, ´ 2016). Our study adds to this stream of literature by offering a novel perspective, namely the effects of U-F的R&D alliances on innovation speed, which encompasses both innovation input (e.g., idea generation) and output efforts (e.g., commercialization).
我们的研究至少对文学做出了三个重大贡献。首先，我们通过强调创新的一个具体指标：创新速度，推进了对U-F相互作用-创新关系的研究。广义地说，在创新文献中，创新可以是由两个组成部分和阶段组成，早期阶段“创新投入”（如R&D和人力资源的结果）和后期阶段“创新产出”（如产品和工艺创新的结果）。然而，在这些文献，先前的研究主要集中在U-F的R&D联盟如何影响创新的投入代理，如R&D绩效，专利数量和技术新颖性，只有少数研究考察了它们对创新产出代理的影响，如产品和工艺创新的数量。我们的研究通过提供一个新的视角，即U-F的R&D联盟对创新速度的影响，增加了这一文献流，这包括创新投入（如想法产生）和产出努力（如商业化）。
Second, our study provides a deeper understanding of the interplay between U-F formal and informal interactions (Schaeffer et al., 2020) by illustrating how the effects of frequent U-F informal contacts substitute for the effects of U-F的R&D alliances on innovation speed. While prior research shows that SISMEs emphasizing the utilization of specific knowledge and emerging technologies (Miozzo and DiVito, 2016) are more likely to benefit from the complementarity of formal and informal interactions with universities (Schaeffer et al., 2020), our analysis suggests an alternative perspective. Our findings show that a high frequency of U-F informal contacts may diminish the beneficial effects of U-F的R&D alliances on innovation speed in SISMEs. Our results provide robust evidence for the broader argument that frequent U-F informal contacts may generate strong “interpersonal commitments and social exchange” (Jiang et al., 2021, p. 1796), leading firms to be over-embedded in such relationships (Anderson, 2013), thereby reducing the efficacy of formal U-F interactions (Landry et al., 2010).
其次，我们的研究通过说明频繁的U-F非正式接触的影响如何取代U-F的R&D联盟对创新速度的影响，提供了对U-F正式和非正式互动之间相互作用的更深入理解。虽然先前的研究表明，强调利用特定知识和新兴技术的SISMEs更有可能从与大学的正式和非正式互动的互补性中受益，但我们的分析提出了另一种观点。我们的研究结果表明，U-F非正式接触的高频率可能会削弱U-F的R&D联盟对SISMEs创新速度的有益影响。我们的结果提供了稳健且更广泛的论点，即频繁的U-F非正式接触可能产生强大的“人际承诺和社会交流”，导致企业过度嵌入这种关系，从而降低正式U-F互动的功效。
Third, we contribute to the literature by identifying firm-level EO as an effective organizational mechanism for the U-F的R&D alliancesinnovation relationship. Specifically, our research extends beyondmerely examining U-F interactions-innovation speed and EO-innovation speed relationships. Instead, we broaden our understanding by emphasizing how firm innovation is accelerated through the integration of U-F的R&D alliances and EO, shedding light on how SISMEs “act entrepreneurially” (George et al., 2002) to reconfigure firm advanced technical resources to become competitive in the market. Our findings also help reconcile the debate regarding the relationship between U-F interactions and entrepreneurship (Dianez-Gonz ´ alez ´ and Camelo-Ordaz, 2019), as well as between EO and innovation speed (Ferreras-M´endez et al., 2022; Shan et al., 2016). We achieve this by demonstrating the beneficial effects of U-F的R&D alliances on EO and the positive impact of EO on innovation speed in the context of SISMEs.
Our analysis, drawing on data from 268 SISMEs located in 10 Chinese science parks, integrates symmetrical and asymmetrical approaches. Specifically, we initially employ the symmetrical approach in terms of the Partial Least Squares Structural Equation Modeling (PLSSEM) to test our conceptual framework. Subsequently, asymmetrical analysis is conducted using Fuzzy-Set Qualitative Comparative Analysis (fsQCA) to explore distinct configurations of antecedents of innovation speed. This methodological amalgamation allows us to gain a holistic understanding of the catalysts and mechanisms affecting innovation speed in SISMEs.
我们的分析利用了位于10个中国科技园的268个中小科技企业的数据，整合了对称和不对称方法。具体来说，我们最初采用偏最小二乘结构方程建模（PLS-SEM）的对称方法来测试我们的概念框架。随后，使用模糊集定性比较分析（fsQCA）进行不对称分析，以探索创新速度前因的不同配置。这种方法上的融合使我们能够全面了解影响SISMEs创新速度的催化剂和机制。
2.理论背景与假说发展
2.1.理论和概念框架
Innovation speed has garnered substantial attention in innovation research and various terms such as new product development speed, speed-to-market, time-to-market, and cycle time have been used to describe this concept (Bao et al., 2021; Chen et al., 2012; Clausen and Korneliussen, 2012). Despite varied terminologies, at its core, innovation speed captures the firm's ability to quickly move from initial idea generation to product commercialization (Cankurtaran et al., 2013; Markman et al., 2005). This capability-centric view can be underpinned by dynamic capability theory, where dynamic capabilities are defined as “the firm's ability to integrate, build, and reconfigure internal and external competencies to address rapidly changing environments” for competitive advantage (Teece et al., 1997, p. 516). Reflected in the firm's innovation abilities (Helfat and Peteraf, 2003), innovation speed is thus the firm's capability to compress time from idea generation through product development, production, to commercialization by reconfiguring both external and internal resources and capabilities (Williamson, 2016; Wu et al., 2017).
创新速度在创新研究中获得了极大的关注，各种术语如新产品开发速度、上市速度、上市时间和周期时间被用来描述这一概念。尽管术语各不相同，但其核心是创新速度抓住了公司从最初的想法产生快速转移到产品商业化的能力。这种以能力为中心的观点可以得到动态能力理论的支持，其中动态能力被定义为“公司整合、建立和重新配置内部和外部能力以应对快速变化的环境的能力”竞争优势。反映在企业的创新能力上，创新速度因此是企业通过重新配置外部和内部资源和能力来压缩从想法产生到产品开发、生产到商业化的时间的能力。
创新速度，科研转化为商品
Drawing on this capability perspective, prior research suggests that accelerated innovation enables a firm to reap the benefits of being the first mover (Fosfuri et al., 2013), maintain strong competitiveness and market positions (Cankurtaran et al., 2013; Kessler and Chakrabarti, 1996), and increase long-term survivability (Chen and Jin, 2023). Despite its recognized importance, research on the factors that drive innovation speed remains underdeveloped. Most studies focusing on elucidating such factors has been conducted at the project level (Ferreras-M´endez et al., 2022, p. 241), concentrating on specific project characteristics such as project investment (Menon et al., 2002), project complexity (Markman et al., 2005), technological newness and radicalness (Seidel, 2007), and knowledge sharing (Wang et al., 2016). Although insightful, prioritising project-level aspects may overlook the significance of organizational-level factors related to innovation speed, such as inter-organizational collaborations (Ma et al., 2012), strategic orientations (Clausen and Korneliussen, 2012), business models (Zhu et al., 2019), and absorptive capacity (Bao et al., 2021).
利用动态能力视角，先前的研究表明，加速创新使企业能够获得先行者的好处，保持强大的竞争力和市场地位，并增加长期生存能力。尽管其重要性得到了公认，但对推动创新速度的因素的研究仍然不发达。大多数侧重于阐明这些因素的研究都是在项目层面进行的，集中于具体的项目特征，如项目投资、项目复杂性、技术新颖性和广泛性以及知识共享。虽然现有研究很有见地，但优先考虑项目层面的方面可能会忽略与创新速度相关的组织层面因素的重要性，如组织间合作、战略方向、商业模式和吸收能力。
Moreover, due to their smallness, many SISMEs may only engage in one innovation project at a time, requiring them to dedicate a significant portion of their limited organizational resources to each initiative. In such scenarios, firm-level factors may have a more pronounced impact on innovation. To deepen our understanding of this phenomena, wedevelop a nuanced model that incorporates both external and internal firm-level factors as predictors of innovation speed in SISMEs. Specifically, we focus on the role of U-F interactions in accelerating SISMEs' innovation (Du et al., 2014). U-F interactions, serving as a critical external conduit for accessing advanced knowledge (Garcia-Perez-deLema et al., 2017; Jensen et al., 2007), are indispensable for SISMEs. They offer avenues for these firms to acquire and assimilate new technical knowledge (Melnychuk et al., 2021) and diverse technological resources (Wirsich et al., 2016) to improve their existing knowledge and resource bases (Caloghirou et al., 2021), ultimately bolstering the firm's innovative capabilities (Polidoro Jr et al., 2011; Polidoro Jr. et al., 2022).
此外，由于规模较小，许多SISMEs一次可能只参与一个创新项目，这要求他们将有限的组织资源的很大一部分用于每个计划。在这种情况下，企业层面的因素可能会对创新产生更明显的影响。为了加深对这一现象的理解，我们开发一个细致入微的模型，将外部和内部公司层面的因素作为SISMEs创新速度的预测因素。具体来说，我们关注U-F相互作用在加速SISMEs创新中的作用。U-F互动，作为获取先进知识的重要外部渠道，对SISMEs是不可或缺的。它们为这些公司提供了获取和吸收新技术知识和多样化技术资源的途径，以改善其现有的知识和资源基础，最终增强公司的创新能力。
Firms develop U-F interactions through either formal collaborations or informal contacts, distinguished by the presence or absence of contractual agreements (Azagra-Caro et al., 2017). U-F的R&D alliances, considered as the most utilized form of formal U-F interactions (Caloghirou et al., 2021; George et al., 2002; Soh and Subramian, 2014), typically focus on the creation of codified novel knowledge (GarciaPerez-de-Lema et al., 2017; Scillitoe and Chakrabarti, 2010). These alliances often lead to an increase in patents and products under development but do not necessarily lead to marketable products (George et al., 2002). Despite the numerous benefits of U-F的R&D alliances for innovation (Melnychuk et al., 2021; Soh and Subramian, 2014), they also pose challenges due to the mismatched orientations of firms, which aim for marketable innovation, and universities, which prioritize the creation of scientific knowledge (He et al., 2021). Moreover, it is argued that merely having network resources does not guarantee enhanced innovation, instead, it requires the effective configuration of external and internal knowledge and resources (Wei et al., 2012, p. 383). More importantly, innovation speed is often an “entrepreneurial and risky endeavor” that necessitates entrepreneurial mechanisms (Clausen and Korneliussen, 2012; Ferreras-M´endez et al., 2022, p. 241). Therefore, we propose that firm-level EO plays a pivotal role in connecting U-F的R&D alliances with innovation speed in SISMEs.
企业通过正式合作或非正式接触发展U-F互动，以是否存在合同协议为特征。U-F的R&D联盟，被认为是最常用的正式U-F互动形式，通常侧重于创造编纂的新知识。这些联盟通常会导致专利和开发中产品的增加，但不一定会导致适销对路的产品。尽管U-F的R&D创新联盟有许多好处，但由于以适销创新为目标的公司和优先创造科学知识的大学的方向不匹配，它们也带来了挑战。此外，有人认为仅仅拥有网络资源并不能保证增强创新，相反，它需要外部资源的有效配置以及内部知识和资源。更重要的是，创新速度往往是一种“创业和冒险的努力”，需要创业机制。因此，我们提出企业层面的EO在将U-F的R&D联盟与SISMEs的创新速度联系起来方面起着关键作用。
Firm-level EO is conceptualized as a strategic attribute and capability that encompasses specific “processes, practices, and decision-making activities”, including engaging in product-market innovation, venturing into somewhat risky enterprises, and exploring and exploiting opportunities (Covin and Slevin, 1991; Lumpkin and Dess, 1996, p. 136). EO manifests in three interconnected dimensions: (1) organizational innovativeness, marked by creativity and experimentation in new products/services introductions; (2) risk-taking, characterized by bold actions in uncertain environments; and (3) proactiveness, which involves pursuing novel opportunities and initiatives ahead of competitors (Covin and Slevin, 1989; Covin and Wales, 2012, 2019). Firms with a strong EO stand out for their ability to innovate, embrace risks, and proactively capitalize on opportunities, setting themselves apart from traditional firms (e.g., Brouthers et al., 2015; Wiklund and Shepherd, 2003), which prioritize stability, risk aversion, and adhering to established procedures. However, possessing EO does not imply a lack of rules or structure within a firm. Rather, it signifies a strategic posture and decision-making framework that guides organizational strategies to form supportive structures for entrepreneurial behavior and empower employees to contribute to innovation, thus improving firm performance (e.g., Engelen et al., 2014; Patel et al., 2015). As such, EO enhances a firm's capacity to integrate external and internal resources (Kollmann and Stockmann, ¨ 2014; Zhang et al., 2020) to forge new “internal firm capabilities” (Boso et al., 2013, p. 77), and ultimately, to drive innovation (Arzubiaga et al., 2018).
企业层面的EO被概念化为一种战略属性和能力，包括特定的“过程、实践和决策活动”，包括参与产品市场创新、冒险进入有一定风险的企业以及探索和开发机会。EO表现在三个相互关联的维度上：（1）组织创新性，以新产品/服务介绍中的创造力和实验为标志；（2）冒险，表现为在不确定的环境中大胆行动；（3）主动性，包括在竞争对手之前追求新的机会和计划。拥有强大EO的公司因其创新、拥抱风险和主动利用机会的能力而脱颖而出，将自己与传统公司区分开来，优先考虑稳定性、风险规避和遵守既定程序。然而，拥有EO并不意味着公司内部缺乏规则或结构。相反，它标志着一种战略姿态和决策框架，指导组织战略形成创业行为的支持结构，并授权员工为创新做出贡献，从而提高企业绩效。因此，EO增强了企业整合外部和内部资源的能力，以打造新的“企业内部能力”，最终驱动创新。
On the other hand, U-F informal contacts also contribute to knowledge transfer (Díez-Vial and Montoro-Sanchez, 2016). Unlike their formal counterparts, U-F informal contacts are not bounded by contracts but thrive within an informal doing, using, and interacting (DUI) mode (Jensen et al., 2007) through interpersonal bonds (Garcia-Perez-deLema et al., 2017). Such informal DUI modes prioritize learning of tacit knowledge elements such as “know-how” and “know-who”, which arepivotal for enhancing individuals' problem-solving capabilities (Fern´andez-Esquinas et al., 2016; Jensen et al., 2007). As such, informal contacts are often considered an essential component of innovation strategies (Garcia-Perez-de-Lema et al., 2017) that influence innovation outcomes in SMEs (Apa et al., 2021).
另一方面，U-F非正式接触也有助于知识转移。与正式接触不同，U-F非正式接触不受契约的约束，而是通过人际纽带。这种非正式的酒后驾车模式优先学习隐性知识要素，如“诀窍”和“诀窍”，这些要素是对提高个人解决问题的能力至关重要。因此，非正式接触通常被认为是影响中小企业创新成果的创新战略的重要组成部分。
Given the importance of both U-F formal and informal interactions in acquiring diverse knowledge and cutting-edge technologies from universities (Dahlander et al., 2016; Scandura, 2016; Soh and Subramian, 2014; Zhang et al., 2022b), understanding their interactive effects is crucial for firms focused on innovation (Azagra-Caro et al., 2017; Landry et al., 2010). However, the literature presents inconclusive findings regarding the interactive effects of U-F formal alliances and informal contacts. While some studies indicate that these interactions can generate complementary effects (e.g., Díez-Vial and Montoro-Sanchez, ´2016; Schaeffer et al., 2020), other studies suggest trade-offs in their effects on innovation (e.g., Toole and Czarnitzki, 2010). This divergence underscores a gap in our understanding of their interactive effects on innovation, and specifically whether they influence innovation in a complementary or substitutive manner. To this end, our study proposes that frequent U-F informal contacts may weaken the effects of U-F的R&D alliances on innovation speed. We detail these relationships as discussed above in our conceptual model, illustrated in Fig. 1.
鉴于U-F正式和非正式互动在从大学获取多样化知识和尖端技术方面的重要性，理解它们的互动效应对于专注于创新的企业至关重要。然而，关于U-F正式联盟和非正式接触的互动效应，文献给出了不确定的发现。虽然一些研究表明这些相互作用可以产生互补效应，其他研究表明它们对创新的影响是权衡的。这种差异凸显了我们对它们对创新的互动影响的理解上的差距，特别是它们是以互补还是替代的方式影响创新。为此，我们的研究提出，频繁的U-F非正式接触可能会削弱U-F的R&D联盟对创新速度的影响。如图1所示，我们在我们的概念模型中详细描述了这些关系。
2.2.创业导向作为U-F的R&D联盟与创新速度的中介机制
Building on the theoretical foundation explored previously, we suggest that EO serves as a pivotal mechanism through which U-F的R&D alliances boost innovation speed. This assertion stems from the goal asymmetry between U-F的R&D alliances and firm innovation. Specifically, while U-F的R&D alliances grant firms access to novel knowledge and inventions, these outcomes are often more scientific in nature and may not directly correspond with a firm's commercial goals (Soh and Subramian, 2014). In contrast, innovation speed, as a critical component of a firm's time-to-market capability (Bao et al., 2021), must align with the firm's commercial objectives. To benefit from novel technical knowledge gained through U-F的R&D alliances, firms need to strengthen their ability to assimilate and apply such knowledge towards commercial ends. Indeed, this ability is significantly bolstered by a firm's EO, which prioritizes innovativeness and value creation through exploration and exploitation of market opportunities (Lumpkin and Dess, 1996). Specifically, EO adds in the synchronization of R&D outcomes with commercial objectives. It does so by promoting the exploration of cutting-edge technical knowledge for creating market value and improving the integration of market insights and opportunities with R&D efforts (Hughes et al., 2021), thereby propelling rapid innovation (Mehrabi et al., 2019). Furthermore, the potential goal misalignment between U-F的R&D alliances and firm innovation also introduces uncertainty about the relevance and applicability of the acquired knowledge and resources for innovation. To navigate this uncertainty, firms need to adopt a trial-and-error approach to utilize these resources for innovation (Ganco, 2017), which requires a readiness to embrace risk. Thus, steering SISMEs towards risk-taking enhances their ability to quickly integrate and apply knowledge from U-F的R&D alliances with their existing knowledge base, leading to accelerated innovation.
在前面探索的理论基础上，我们认为EO是U-F的R&D联盟提高创新速度的关键机制。这一论断源于U-F的R&D联盟和企业创新之间的目标不对称。具体来说，虽然U-F的R&D联盟允许企业获得新知识和发明，但这些成果在本质上往往更具科学性，可能不直接符合企业的商业目标。相比之下，创新速度作为企业上市时间能力的关键组成部分，必须与企业的商业目标保持一致。为了从通过U-F的R&D联盟获得的新技术知识中受益，企业需要加强其吸收和应用这些知识以达到商业目的的能力。事实上，这种能力得到了公司EO的极大支持，EO通过探索和利用市场机会优先考虑创新和价值创造。具体来说，EO增加了R&D成果与商业目标的同步。它通过促进对尖端技术知识的探索来创造市场价值，并改善市场洞察力和机会与
研发努力，从而推动快速创新。此外，U-F的R&D联盟和企业创新之间潜在的目标错位也带来了所获得的知识和资源对创新的相关性和适用性的不确定性。为了应对这种不确定性，企业需要采用试错法来利用这些资源进行创新，这需要准备好接受风险。因此，引导SISMEs承担风险增强了他们将U-F的R&D联盟的知识与现有知识库快速整合和应用的能力，从而加速创新。
The mediating role of EO may also stem from the “potential risks of knowledge leakage and misappropriation” during knowledge transfer (Zhang et al., 2019a, p. 2640), especially when a university collaborates with multiple firms simultaneously. These risks potentially dilute the effectiveness of an individual firm's U-F的R&D alliances (de Leeuw et al., 2019). Knowledge leakage refers to instances where proprietary knowledge intended for one firm through its U-F的R&D alliances might unintentionally spread to and be utilized by other firms, including competitors, engaged with the same university, while misappropriation occurs when either party uses the other's knowledge without proper authorization (O'Dwyer et al., 2023). Such risks can discourage full engagement and effort in R&D alliances, “significantly jeopardizing the focal firm's innovation” prospects (Zhang et al., 2019a, p. 2639). Therefore, to help ensure U-F的R&D alliances to enhance innovation speed, SISMEs need to be cognizant of such risks. They must proactively evaluate the outcomes from these alliances and develop forward-looking mechanisms to gauge the extent to which they can integrate knowledge and resources from these U-F的R&D alliances with their own internal knowledge (Lumpkin and Dess, 2001). Given EO's emphasis on proactiveness, such adaptive and forward-looking mechanisms can be facilitated by adopting high degrees of EO. Thus, through EO, U-F R&D alliances may booster innovation speed more effectively.
EO的中介作用也可能源于知识转移过程中的“知识泄漏和盗用的潜在风险”，尤其是当大学合作时同时与多家公司合作。这些风险可能会削弱单个公司U-F的R&D联盟的有效性。知识泄露是指一家公司通过其U-F的R&D联盟获得的专有知识可能无意中传播到与同一所大学合作的其他公司（包括竞争对手）并被其利用，而当任何一方未经适当授权使用另一方的知识时，就会发生盗用。这种风险可能会阻碍充分R&D联盟中的参与和努力，“严重危及焦点公司的创新”前景。因此，为了帮助确保U-F的R&D联盟以增强创新速度，中小科技型企业需要认识到这种风险。他们必须主动评估这些联盟的成果，并开发前瞻性机制，以衡量他们可以在多大程度上将这些U-F的R&D联盟的知识和资源与他们自己的内部知识相结合。鉴于EO对主动性的强调，这种适应性和前瞻性机制可以通过采用高度的EO来促进。因此，通过EO，U-F的R&D联盟可以更有效地提高创新速度。
Previous research has also provided empirical evidence supporting the direct impacts of U-F alliances on EO (e.g., Di´anez-Gonz´alez and Camelo-Ordaz, 2019; Scillitoe and Chakrabarti, 2010) and, separately, the influence of EO on innovation speed (Clausen and Korneliussen, 2012; Ferreras-M´endez et al., 2022; Shan et al., 2016). These findings serve as a prerequisite for our argument regarding EO's mediating role in transforming the benefits of U-F的R&D alliances into increased innovation speed. Specifically, U-F的R&D alliances facilitate an increase in technical resource availability (Dianez-Gonz ´ alez ´ and Camelo-Ordaz, 2019) by enabling firms to tap into a wider pool of codified knowledge from universities (Díez-Vial and Montoro-Sanchez, ´ 2016). This enhanced resource availability can allow firms to more effectively pursue EO, as EO demands substantial resources as a strategic capability (Covin and Slevin, 1991; Wales et al., 2013, p. 1047), thereby leading to improved innovation outcomes (Bouncken et al., 2016). In turn, embracing EO enables firms to leverage the acquired knowledge and resources to achieve first-mover advantages (proactiveness), engage in the innovative and experimental introduction of new products/services (innovativeness), and explore new yet risky opportunities (risk-taking) (Wales et al., 2013; Wiklund and Shepherd, 2003). This approach ultimately culminates in accelerated innovation speed (Shan et al., 2016). Taken together, we hypothesize：
先前的研究也提供了经验证据，支持U-F联盟对EO的直接影响，以及EO对创新速度的影响。这些发现是我们论证EO在将U-F的R&D联盟的利益转化为提高创新速度方面的中介作用的先决条件。具体来说，U-F的R&D联盟通过使企业能够利用来自大学的更广泛的编纂知识库，促进了技术资源可用性的增加。这种增强的资源可用性可以让企业更有效地追求EO，因为EO需要大量的资源作为一种战略能力，从而提高创新成果。反过来，拥抱EO使企业能够利用获得的知识和资源来实现先发优势（主动性），参与新产品/服务的创新和实验性引入（创新性），并探索新的但有风险的机会（冒险）。这种方法最终导致创新速度加快。综上所述，我们假设：
H1.
EO mediates the relationship between U-F的R&D alliances and innovation speed such that U-F的R&D alliances facilitate EO, which in turn, leads to greater innovation speed in SISMEs.
EO调节U-F的R&D联盟和创新速度之间的关系，使得U-F的R&D联盟促进EO，这反过来又导致SISMEs更大的创新速度。
2.3.U-F非正式联系与R&D联盟对创新速度的交互作用
As previously discussed, while both U-F的R&D alliances and informal contacts alone can increase firm innovation speed, their interactive effects on innovation speed can be complementary or substitute (Landry et al., 2010). For SISMEs, we argue that their effects on innovation speed are substitutive. Specifically, we posit that frequent U-F informal contacts may weaken the effects of U-F的R&D alliances on innovation speed for several reasons. First, with a low frequency of U-F informal contacts, SISMEs might encounter limited options of informal channels for knowledge transfer, leading to increased reliance on formal interactions with universities to acquire the advanced knowledge and technologies necessary for innovation. As a result, the significance of U-F的R&D alliances on innovation speed is heightened when U-F informal contacts frequency is low. Conversely, a high frequency of U-F informal contacts offers firms alternative, possibly more cost-effective ways to access university knowledge and resources, potentially enhancing the resource base for innovation more efficiently (Dahlander et al., 2016; Dianez- ´ Gonzalez ´ and Camelo-Ordaz, 2019). Consequently, with abundant U-F informal contacts, firms may become less dependent on formal R&D alliances for accelerating innovation. This trade-off between engaging in U-F的R&D alliances and frequent U-F informal contacts is especially pronounced in individual entrepreneurs and SMEs, where participation in one type of knowledge transfer activity might limit the capacity to engage in another (Landry et al., 2010, p. 1390).
如前所述，虽然U-F的R&D联盟和非正式联系本身都可以提高企业创新速度，但它们对创新速度的互动影响可以是互补或替代的。对于SISMEs，我们认为它们对创新速度的影响是替代性的。具体来说，我们假设频繁的U-F非正式接触可能会削弱U-F的R&D联盟对创新速度的影响，原因有几个。首先，由于U-F非正式接触的频率较低，SISMEs可能会遇到有限的非正式知识转移渠道选择，导致越来越依赖与大学的正式互动来获得创新所需的先进知识和技术。因此，U-F的R&D的意义当U-F非正式接触频率较低时，联盟的创新速度会加快。相反，高频率的U-F非正式接触为企业提供了获取大学知识和资源的替代性、可能更具成本效益的方式，有可能更有效地增强创新的资源基础。因此，有了丰富的U-F非正式联系，企业可能会减少对正式R&D联盟的依赖来加速创新。参与U-F的R&D联盟和频繁的U-F非正式接触之间的这种权衡在个体企业家和中小企业中尤为明显，在这些企业中，参与一种类型的知识转移活动可能会限制参与另一种类型的知识转移活动的能力。
Second, and more significantly, U-F informal contacts may promote deeper learning and the exchange of tacit knowledge (Fernandez- ´ Esquinas et al., 2016), potentially accelerating the development of problem-solving competencies (Jensen et al., 2007). For example, U-F informal contacts may boost individuals' absorptive capacity in SISMEs, improving the effectiveness of knowledge transfer from universities (Díez-Vial and Montoro-Sanchez, ´ 2016; Uzzi, 1996). Such benefits could lead to reduced engagement in other formal relationships and diminish the advantages of diversification (Anderson, 2013). As a result, a high level of U-F informal contacts may decrease SISMEs' reliance on U-F的R&D alliances, thereby diminishing their role in facilitating university knowledge transfer for innovation purposes (Polidoro Jr. et al., 2022), and affecting innovation speed.
Third, while frequent U-F informal contacts offer significant benefits, an excessive frequency of U-F informal contacts may result in individuals becoming overly embedded in these interactions, which also hinders the pursuit of alternative partnerships (Elfenbein and Zenger, 2017; Uzzi, 1997). Specifically, while embeddedness, which denotes deep and cohesive relationships characterized by social and interpersonal ties (Granovetter, 1985; Greve et al., 2010), is generally beneficial for fostering innovation-related collaborations (Lioukas and Reuer, 2020; Uzzi, 1997), it may cause a cognitive lock-in for SISMEs (Lavie and Drori, 2012) when such relationships become too frequent or intense, resulting in over-embeddedness (Granovetter, 1985). This situation can lead to a constrained and dominant focus on acquiring and exchanging tacit knowledge through close interpersonal connections (Jiang et al., 2021), thus making firms hesitant to explore or leverage the benefits of other U-F formal interactions (Uzzi, 1997) essential for rapid innovation. Consequently, a high frequency of U-F informal contacts might reduce the beneficial effects of U-F的R&D alliances on innovation speed. Drawing from the above discussion, we propose the following hypothesis:
第三，虽然频繁的U-F非正式接触提供了显著的好处，但过度频繁的U-F非正式接触可能会导致个人过度嵌入这些互动中，这也阻碍了对替代伙伴关系的追求。具体来说，嵌入性，表示以社会和人际关系为特征的深层和有凝聚力的关系，通常有利于促进与创新相关的合作，当这种关系变得过于频繁或强烈时，可能会导致SISMEs的认知锁定，导致过度嵌入。这种情况可能导致对通过密切的人际关系获取和交流隐性知识的限制和主导关注，从而使企业犹豫是否探索或利用对快速创新至关重要的其他U-F正式互动的好处。因此，U-F非正式接触的高频率可能会降低U-F的R&D联盟对创新速度的有益影响。根据上述讨论，我们提出以下假设：
H2. Frequent U-F informal contacts weaken the relationship between U-F的R&D alliances and innovation speed in SISMEs.
H2.频繁的U-F非正式接触削弱了U-F的R&D联盟与SISMEs创新速度之间的关系。
3. 方法
3.1. 样本和数据收集
Following prior studies on U-F interactions (e.g., Caloghirou et al., 2021; He et al., 2021; Lee and Miozzo, 2019), we utilized survey data to test our conceptual model. We conducted a questionnaire survey targeting SISMEs located in 10 national science parks across Guangdong, Zhejiang, Jiangsu, and Fujian provinces in China. These science parks were selected for their strategic importance. First, they are in regions known for their rapid growth and active entrepreneurial climates, thus conducive to fostering innovation through new ventures (De Oliveira et al., 2022). Second, their proximity to elite universities and research institutions enhances the prospects of firms accessing the R&D resources and intellectual capital that are essential for innovation and business development (Armanios et al., 2017). Third, while prior research on China's science parks has predominantly focused on the Beijing Zhongguancun Science Park (ZSP), the inclusion of multiple parks aims to mitigate potential selection bias.
根据之前关于U-F相互作用的研究，我们利用调查数据来测试我们的概念模型。我们对广东10个国家科技园的中小科技型企业进行了问卷调查，中国的浙江、江苏和福建省。这些科学园因其战略重要性而被选中。首先，它们位于以快速增长和活跃的创业氛围而闻名的地区，因此有利于通过新企业促进创新。其次，它们靠近精英大学和研究机构，增强了企业获得对创新和商业至关重要的研发资源和智力资本的前景发展。第三，虽然先前的研究中国的科学园主要集中在北京中-关村科技园（ZSP），多个园区的纳入旨在减轻潜在的选择偏差。
样本选择的原因
Aligning with prior research on firms within Chinese science parks (Filatotchev et al., 2001; Zhang and Guan, 2021), our sample was comprised of SISMEs from science-intensive and high-tech industries, as officially categorized by China's High-Tech Industry Classification. This includes sectors such as pharmaceuticals, biotechnology, information technology, and more. These firms, often established as university spinoffs or founded by scientists, have a strong R&D focus, and emphasize the advanced skills and specialized knowledge of their employees. Moreover, we define SMEs as firms with 500 or fewer employees, reflecting that “about 99% of SMEs in China have less than 500 employees (OECD, 2022)” (Zhang et al., 2022a, p. 8).
与先前对中国科技园区内企业的研究一致，我们的样本由来自科学密集型和高科技产业的SISMEs组成，根据中国高科技产业分类的官方分类。这包括制药、生物技术、信息技术等行业。这些公司通常是作为大学的分支机构或由科学家创立的，具有强烈的研发重点，并强调员工的先进技能和专业知识。此外，我们将中小企业定义为员工500人或更少的企业，反映出“中国约99%的中小企业员工少于500人（OECD，2022）”。
The questionnaire for data collection, which was based on validated items from existing literature, was initially drafted in English. To ensure linguistic accuracy in the Chinese version, a translation-back-translation procedure was utilized. This translated questionnaire underwent a pretest by four Chinese academics and a pilot test with 70 CEOs or top managers of SMEs in two science parks, ensuring its reliability, validity, and clarity. Feedback from these tests led to further refinement of the questionnaire.
数据收集问卷基于现有文献中经过验证的项目，最初是用英语起草的。为了确保中文版本的语言准确性，采用了翻译-回译程序。这份翻译后的问卷经过了四位中国学者的预测试，并在两个科技园的70位中小企业首席执行官或高层管理人员中进行了试点测试，确保了其信度、效度和清晰度。这些测试的反馈导致了问卷的进一步完善。
For the initiation of data collection, a member of the research team visited each of the 10 science parks to engage with park administrators and discuss the research objectives. With their support, we employed the key informant approach (Kumar et al., 1993), sending personalized emails to the CEOs of SISMEs within these parks. Four weeks following the initial contact, a reminder email was sent to those who had not yet responded. The study resulted in 296 responses, from which a usable sample of 268 firms was derived after discarding responses with significant data missing for questions related to key constructs. This sample had firms of varied size and age and included firms from eight scienceintensive and high-tech industries. Descriptive statistics for these distributions are detailed in Table 1.
为了开始数据收集，研究小组的一名成员访问了10个科学园中的每一个，与园区管理人员接触并讨论研究目标。在他们的支持下，我们采用了关键线人方法，向这些园区内的SISMEs首席执行官发送个性化电子邮件。初次联系四周后，向尚未回复的人发送了一封提醒电子邮件。该研究产生了296份答复，在丢弃了与关键结构相关的问题中缺少重要数据的答复后，从中得出了268家公司的可用样本。这个样本有不同规模和年龄的公司，包括来自八个科学密集型和高科技行业的公司。这些分布的描述性统计数据详见表1。
3.2措施
To increase the validity of measurement, we used well-established items from prior research to measuring our key constructs. We elaborate on the detail of these measures below.
为了提高测量的有效性，我们使用了先前研究中公认的项目来测量我们的关键结构。我们在下面详细阐述了这些措施。
3.1.1创新速度
Drawing on prior research (Acharya et al., 2020; Markman et al., 2005; Shan et al., 2016; Wang et al., 2016), we measured innovation speed using three 11-point Likert scale items. These items assess the firm's ability to conduct rapid innovation in comparison to major competitors (α = 0.914), as detailed in Table 2.
借鉴以往的研究，我们使用三个11点李克特量表项目测量创新速度。这些项目评估了公司与主要竞争对手相比进行快速创新的能力（α=0.914），如表2所示。
3.1.2 U-F研发联盟
Informed by previous studies on R&D alliances, we developed three items to gauge U-F的R&D alliances. Participants evaluated the extent to which their firm engages with and partners with universities/research institutes for R&D projects based on a contract (e.g., Caloghirou et al., 2021), fosters and maintains constructive relationships with these entities (e.g., D'Este et al., 2019), and commits to U-F joint R&D projects.
根据以往关于R&D联盟的研究，我们开发了三个衡量U-F的R&D联盟的项目。参与者评估了他们的公司在多大程度上根据合同与大学/研究机构合作开展R&D项目，培养和保持与这些实体的建设性关系，并致力于U-F联合R&D项目。
3.2.3.U-F非正式接触频率
Drawing from prior research (Di´anez-Gonzalez ´ and Camelo-Ordaz, 2019; He et al., 2021; Scillitoe and Chakrabarti, 2010), U-F informal contacts frequency was assessed by querying the frequency of interactions firm managers and members have with universities, such as attending seminars/conferences, personal interactions, and communications with academics, etc. This was measured on a 5-point scale ranging from 1 = one or fewer contacts annually; 2 = on average, a contact every six months; 3 = on average, one contact every two months; 4 = on average, one contact every two weeks; 5 = one and/or multiple daily contacts.
借鉴先前的研究，U-F非正式接触频率是通过询问公司经理和成员与大学互动的频率来评估的，如参加研讨会/会议、个人互动以及与学者的交流等。这是在5分制下测量的，范围从1=每年一次或更少的接触；2=平均每六个月联系一次；3=平均每两个月联系一次；4=平均每两周联系一次；5=一个和/或多个日常联系人。
3.2.4.创业导向(EO)
We measured EO as a second-order construct composed of three dimensions: innovativeness, risk-taking, and proactiveness (FerrerasM´endez et al., 2021; George, 2011). We administered the widely adopted nine-item measure of EO developed by Covin and Slevin (1989) to measure these three dimensions with 3 items each on a 7-point Likert scale. The results in Table 2 show that an item was removed from each dimension given their poor factor loadings (< 0.40). Consequently, innovativeness (α = 0.867), risk-taking (α = 0.889) and proactiveness (α = 0.880) are each measured with 2 items. The second-order construct of EO is then measured by these three dimensions (α = 0.937).
我们将EO测量为由三个维度组成的二阶结构：创新性、冒险性和主动性。我们采用了Covin和Slevin（1989）开发的广泛采用的EO九项测量方法来测量这三个维度，每个维度有3个项目，采用7分李克特量表。表2中的结果显示，鉴于每个维度的因子负荷较差（<0.40），从每个维度中删除了一个项目。因此，
创新性(α=0.867）、冒险性(α=0.889）和主动性(α=0.880）各用2个项目测量。然后通过这三个维度测量EO的二级结构(α=0.937)。
3.2.5.控制变量
We included control variables at three levels: firm, science park, and industry level. At the firm level, as demonstrated in Table 1, we first controlled for firm age and measured it as the “logarithm of the difference between the year that the survey was conducted and the establishment year of the firm” (Caloghirou et al., 2021; p. 6; Soh and Subramian, 2014). Second, firm size was included, measured by the average number of full-time employees over the latest two years, categorized as follows: 1 ≤ 10 employees; 2 = 11 to 50 employees; 3 = 51 to 100 employees; 4 = 101 to 300 employees; 5 = 301 to 500 employees (e. g., Ferreras-M´endez et al., 2022). Third, due to its direct impact on innovation speed, we included R&D intensity and operationalized it as the ratio of R&D expenditures to total sales (Bao et al., 2021). Fourth, considering the influence of highly skilled employees on innovation in science parks (Filatotchev et al., 2001), we categorized the percentage of employees with a graduate degree relative to the total number of employees into 7 groups: 1 = 0%; 2 = 0% to 5%; 3 = 5% to 10%; 4 = 10% to 25%; 5 = 25% to 50%; 6 = 50% to 75%, and 7 = >75%). The fifth control variable, development stage, adopted from Armanios et al. (2017), is measured based on the firm's stage at the time of entry into the science park (e.g., “[1] R&D/planning...[5] growth: five percent or more profitability”) (p. 1380). The last firm-level control variable is export experience, coded as 1 if the firm has exported technology, products, or services, and 0 if not (Filatotchev et al., 2001).
我们在三个层面上纳入了控制变量：公司、科学园和行业层面。在公司层面，如表1所示，我们首先控制了公司年龄，并将其测量为“进行调查的年份与成立年份之间差异的对数”、“公司成立年”。第二，公司规模包括在内，通过创新性、过去两年全职员工的平均人数。第三，由于其对创新速度的直接影响，我们将R&D强度纳入其中，并将其操作为R&D支出与总销售额的比率。第四，考虑到高技能员工对科技园创新的影响，我们将拥有研究生学位的员工相对于员工总数的百分比分为7组。第五，控制变量，开发阶段，取自Armanios等人（2017），是根据公司进入科学园时的阶段来衡量的。最后一个企业层面的控制变量是出口经验，如果企业出口了技术、产品或服务，如果没有，则为0。
At the science park level, following Ng et al. (2019)'s study, which underscores the importance of the size of the science park for firms' business and innovation activities, we measured science park size based on the number of resident organizations: (1) <50, (2) 50 to 100, (3) 100 to 200, (4) 200 to 400, (5) 400 to 600, (6) 600 to 1000, or (7) >1000 (p. 721).
在科学园层面，跟随吴等人（2019）的研究强调了科学园的规模对企业商业和创新活动的重要性，我们根据常驻组织的数量来衡量科学园的规模：（1）<50，（2）50至100，（3）100至200，（4）200至400，（5）400至600，（6）600至1000，或（7）>1000
At the industry level, prior research on science parks (e.g., Zhang et al., 2019b) and innovation (e.g., Chaudhuri et al., 2023; Zhang et al., 2020) has indicated that industry and market environmental dynamism significantly affects innovation, including innovation speed (Bao et al., 2021). We therefore controlled for environmental dynamism using four items adopted from Jansen et al. (2006) on a 7-point scale (1 = strongly disagree to 7 = strongly agree) with good reliability and validity (α =0.882), as detailed in Table 2. Finally, industry dummies for various industries, as listed in Table 1, were included to account for potential heterogeneity related to industry types (Caloghirou et al., 2021)
在行业层面，先前对科学园的研究和创新表明，行业和市场环境动态显著影响创新，包括创新速度。因此，我们使用四种方法来控制环境动态。采用詹森等人（2006）的项目按7分制（1=强烈不同意7=非常同意）具有良好的信度和效度，详见表2。最后，各种行业如表1所列，包括行业是为了说明与行业类型相关的潜在异质性。
3.3.分析模型
We employed both symmetrical and asymmetrical approaches to develop a comprehensive understanding of the phenomenon—the relationship between U-F interactions, EO, and innovation speed (Belitski et al., 2019; He et al., 2021). Recent studies in innovation have frequently utilized PLS-SEM (e.g., García-Granero et al., 2020; Hung, 2017) and fsQCA (e.g., Renko et al., 2020; Speldekamp et al., 2020; Subrammanian et al., 2022) to address the complexity of causal conceptual models. Drawing from these studies, we initially utilized PLSSEM (Ringle et al., 2015) to test our hypotheses and then adopted fsQCA (Fiss, 2011; Ragin, 2000) to gain complementary insights (Pappas and Woodside, 2021).
我们采用了对称和不对称的方法来全面理解这一现象U-F相互作用、EO和创新速度之间的关系。最近的创新研究经常使用PLS-SEM和fsQCA来解决因果概念模型的复杂性。根据这些研究，我们最初利用PLS-SEM来检验我们的假设，然后采用fsQCA以获得互补的见解。
On the one hand, PLS-SEM as a symmetrical approach provides the most appropriate estimation for complex causal relationship frameworks with latent constructs (Hair et al., 2017; Hair et al., 2020). Being a variance-based approach, it offers “much greater flexibility compared to CB-SEM” (Hair et al., 2020, p. 102), enabling a bootstrapping approach to test complex moderated mediation relationships, as is the case in this study. Specifically, PLS-SEM allows us to estimate two sub-models: (1) the measurement model and (2) the structural model, addressing both measurement errors and issues associated with regression models (García-Granero et al., 2020). However, a drawback of PLS-SEM is that it is based on a set of mean-centered regressions, where explanatory variables compete in producing an outcome, leading to an “incomplete picture of the effects” (Rasoolimanesh et al., 2021, p. 1572). On the other hand, fsQCA, as an asymmetrical approach, accounts for causal conjunction, asymmetry and equifinality of configurations (Haefner et al., 2021). It posits that the same outcome can result from several non-linear configurations of conditions (Fiss, 2007). Therefore, it allows us to move beyond symmetric analyses to explore how multiple configurations of factors contribute to specific levels of innovation speed. In fact, the joint use of PLS-SEM and fsQCA not only aligns well with our research objectives but also provides us with opportunities to better assess our “model's predictive power” and generate more insightful “managerial recommendations” (Rasoolimanesh et al., 2021, p. 1573).
一方面，PLS-SEM作为一种对称方法，为具有潜在结构的复杂因果关系框架提供了最合适的估计。作为一种基于方差的方法，它提供了比“CB-SEM”更能够测试复杂的调节中介关系，如本研究中的情况。具体来说，PLS-SEM允许我们估计两个子模型：（1）测量模型和（2）结构模型，解决测量误差和与回归模型相关的问题。然而，PLS-SEM的一个缺点是它基于一组以均值为中心的回归，其中解释变量在产生结果时相互竞争，导致“影响的不完整画面”。另一方面，fsQCA作为一种不对称的方法，解释了构型的因果连词、不对称性和等价性。它假设相同的结果可以由条件的几种非线性配置产生。因此，它允许我们超越对称分析，探索多种因素配置如何有助于特定水平的创新速度。事实上，PLS-SEM和fsQCA的联合使用不仅可以很好地对齐,同时也为我们提供了更好地评估我们的“模型的预测能力”并产生更有见地的“管理建议”的机会。
4.结果
4.1.评估测量模型
We evaluated our measurement model in several ways. First, we examined the indicator loadings and weights (coefficients representing “each indicator's relative importance” in the regression of a construct on its indicators) and their statistical significance using a nonparametric bootstrapping procedure with 5000 subsamples (Hair et al., 2017, p. 323). The results, displayed in Table 2 and Fig. 2, indicate that all standardized indicator loadings to their corresponding constructs are greater than 0.708 (min(L) = 0.789) (Hair et al., 2020) with a significance level of p < 0.001, and all indicator weights are also strongly significant (p < 0.001). Together, these results ensure a significant contribution of items to construct score (Hair et al., 2020). We also checked outer variance inflation factors (VIFs) for all constructs to see if there is a collinearity issue. The results reveal that the highest VIF value among indictors is 4.318, which is below the threshold of 5 (Hair et al., 2017), indicating outer collinearity is not a concern in our measurement model.
我们通过几种方式评估了我们的测量模型。首先，我们检查了指标载荷和权重（代表结构回归中“每个指标的相对重要性”其指标）及其统计显著性，使用具有5000个子样本的非参数自举程序。结果如表2和图2所示，其相应结构的所有标准化指标负荷均大于0.708，显著性水平为p<0.001，并且所有指标权重也非常显著（p<0.001）。总之，这些结果确保了项目对构建分数的重大贡献。我们还检查了所有结构的外部方差膨胀因子（VIF），以查看是否存在共线性问题。结果显示，指标中最高的VIF值为4.318，低于阈值，表明外部共线性在我们的测量模型中不是一个问题。
Second, we assessed the reliability of both the first- and second-order constructs using Cronbach's alphas (α) and composite reliability (CR). All alphas' (min(α) = 0.867; max(α) = 0.937) and CRs' values (min(CR) = 0.918; max(CR) = 0.949) shown in Table 2 surpass the threshold of 0.70 and are below 0.95, suggesting construct reliability (Hair et al., 2020). Third, the results in Table 2 also demonstrate that the Average Variance Extracted (AVE) scores for all first- and second-order constructs (min(AVE) = 0.738) exceed the cutoff level of 0.5 (Fornell and Larcker, 1981), Confirming convergent validity.
其次，我们使用Cronbach α（α）和复合可靠性（CR）评估了一阶和二阶结构的可靠性。所有α（min（α）和CRs值超过了0.70且低于0.95，表明结构可靠性。第三，表2中的结果还表明，所有一阶和二阶结构的平均方差提取（AVE）分数（min（AVE）=0.738）超过了0.5的截止水平，证实了收敛有效性。
Fourth, we assessed discriminant validity by the Fornell-Larcker criterion and the heterotrait-monotrait (HTMT). As shown in Table 3, all square root values of AVE are greater than the corresponding correlations between constructs, adhering to the Fornell-Larcker's criterion. Moreover, the results in Table 4 demonstrate that the HTMT scores (max (HTMT) = 0.880) for all construct pairs are below the conservative threshold of 0.90 (Henseler et al., 2015), and the bias-correlated bootstrap 95 % confidence intervals for all construct pairs did not include 1 (max(UCI) = 0.94). Together, these results affirm discriminant validity for our constructs.
第四，我们通过Fornell-Larcker标准和异质性状-单性状（HTMT）评估了判别效度。如表3所示，AVE的所有平方根值都大于结构之间的相应相关性，符合Fornell-Larcker标准。此外，表4中的结果表明，所有构建体对的HTMT评分(最大(HTMT)=0.880)低于保守阈值0.90，并且所有构建体对的偏倚相关引导95%置信区间不包括1。总之，这些结果证实了我们的结构的判别有效性。
4.2.共同方法差异的核算
Following previous studies in innovation and entrepreneurship (García-Granero et al., 2020; Klein et al., 2021; Moore et al., 2021), our data collection was based on responses from single informants. However, this approach might raise concerns about common method variance (CMV). To mitigate this risk, we implemented several procedural remedies recommended by Podsakoff et al. (2003). Initially, before data collection, we selected validated items from existing literature and conducted a pre-test of our questionnaire design. Additionally, we informed survey participants about (1) the research purpose; (2) the absence of correct or incorrect answers; and (3) their responses remaining anonymous and confidential. We also designed the survey to separate scale items for explanatory variables from those for the dependent variable, minimizing participants' ability to discern direct linkages between these variables. Furthermore, we varied response formats for different constructs (e.g., semantic differential and Likert scales) and the number of points (e.g., 11-, 7-, and 5-point scales).
继之前关于创新和创业的研究，我们的数据收集基于单个线人的回答。然而，这种方法可能会引起对通用方法差异（CMV）的担忧。为了降低这种风险，我们实施了Podsakoff等人(2003)推荐的几种程序性补救措施。最初，在数据收集之前，我们从现有文献中选择了经过验证的项目，并对我们的问卷设计进行了预测试。此外，我们告知调查参与者（1）研究目的；（2）没有正确或不正确的答案；（3）他们的回答保持匿名和保密。我们还设计了调查，将解释变量的量表项目与因变量的量表项目分开，最大限度地降低了参与者辨别这些变量之间直接联系的能力。此外，我们针对不同的结构（例如，语义差异和李克特量表）和点数（例如，11分、7分和5分量表）改变了反应格式。
For statistical remedies for CMV, we followed the procedures by Lindell and Whitney (2001) and conducted a partial correlation analysis using a marker variable technique. We selected participants' family reasons for developing a business (family reason) (Chadwick and Raver, 2020) as the marker variable, which is theoretically unrelated. Respondents rated the importance of being respected by family in developing a business on a 5-point Likert scale ranging from 1 = “to no extent” to 5 = “to a very great extent”. Correlations of the marker variable (family reason) with other constructs ranged from − 0.091 to 0.068, none of which were significant (min(p-value) = 0.149). Next, we included family reason as an exogenous effect on each variable in the hypothesized relationships to “control for (partial out) method effects” (Podsakoff et al., 2003, p. 889). The results show that the inclusion of family reason didn't significantly change the results of any path estimates in the hypothesized relationships.
对于CMV的统计治疗，我们遵循Lindell和Whitney（2001）的程序，并进行了偏相关分析使用标记变量技术。我们选择参与者发展企业的家庭原因（家庭原因）作为标记变量，这在理论上是不相关的。重新-受访者根据李克特5分制，从1=“没有”到5=“非常大”，对企业发展中受到家人尊重的重要性进行了评分。标记变量（家族原因）与其他构建体的相关性范围为0.091至0.068，无显著性（min（p值）=0.149）。接下来，我们将家庭原因作为对每个变量的外生影响与“控制（部分输出）方法效应”的假设关系。结果表明，家庭原因的纳入并没有显著改变假设关系中任何路径估计的结果。
Moreover, following the recommended procedure by Liang et al. (2007), we further included a common method factor in our PLS model. This common factor encompassed all indicators of the principal constructs: innovation speed, U-F的R&D alliances, innovativeness, risktaking, proactiveness, and environmental dynamism, thus explaining each indicator's variance by both its principal construct (substantive variance) and the common method factor (method variance). If most method factor loadings are not significant and the indicators' method variances are substantially smaller than their substantive variances, the risk of CMV is not a concern (Liang et al., 2007). The results in Table 5 indicate that all method factor loadings are insignificant (p > 0.05), and the average substantively explained variance (0.818) is significantly greater than the average method variance (0.004).
此外，按照Liang等人（2007）推荐的程序，我们进一步在我们的PLS模型中包括了一个公共方法因子。这一共同因素包括主要结构的所有指标：创新速度、U-F的R&D联盟、创新性、冒险精神、主动性和环境动态性，从而通过其主要结构（实质性方差）和共同方法因素（方法方差）来解释每个指标的方差。如果大多数方法因子负荷不显著，并且指标的方法方差大大小于其实质性方差，则CMV的风险不是一个问题。表5中的结果表明所有方法因子负荷均不显著（p>0.05），平均实质性解释方差（0.818）显著大于平均方法方差（0.004）。
Finally, in line with Klein et al. (2021) and Zhang et al. (2020), we conducted a full collinearity assessment (Kock, 2015) to determine the presence of pathological collinearity indicative of common method bias. The results in Table 6 show that the VIFs of all constructs (max(VIF) = 2.805) fall below the threshold of 3.3. Together, these results Confirm that common method variance is not a serious concern in this research.
最后，与克莱因等人（2021）和张等人（2020）一致，我们进行了完整的共线性评估，以确定是否存在指示常见方法偏倚的病理共线性。表6中的结果显示，所有构建体的VIF(max(VIF)=2.805)均低于3.3的阈值。总之，这些结果证实了通用方法方差在本研究中不是一个严重的问题。
4.3.评价结构模型
Prior to hypothesis testing, we examined the inner Variance Inflation Factors (VIFs) for all variables in our model, with all values below 5 (max(innerVIF) = 3.511), indicating that inner multicollinearity is unlikely to be a concern in this study. The structural model's predictive power was assessed using the coefficient of determination (R2) for endogenous constructs (Hair et al., 2017). As shown in Table 7, adjusted R2 values for endogenous constructs of innovation speed (R2adj = 0.245, p < 0.001) and EO (R2adj = 0.634, p < 0.001) demonstrate strong in-sample predictive power (Hair et al., 2017).
在假设检验之前，我们检查了模型中所有变量的内部方差膨胀因子（VIFs），所有值都低于5（max（innerVIF）=3.511），表明内部多重共线性不太可能是本研究中的一个问题。结构模型的预测性使用决定系数（R2）评估功效内源性构建体（Hair等人，2017）。如表7所示，调整创新速度内生结构的R2值(R2=0.245，p<0.001）和EO(R2=0.634，p<0.001）显示出强样品内预测能力。
To further estimate the model's predictive power, we conducted PLSpredict analysis (Shmueli et al., 2019), where the root mean squared error (RMES) was calculated to assess the predictive performance of the model for the constructs and indicators. The results, shown in Table 8, indicate that all indicators in the PLS-SEM have a Q2 value greater than 0 (Q2predict> 0). Additionally, seven out of nine items of the first-order constructs in the PLS-SEM exhibit smaller prediction errors than those in a linear regression model (LM), suggesting medium predictive power (Shmueli et al., 2019). Moreover, the cross-validated predictive ability test (CVPAT) (Sharma et al., 2023) was conducted. The results in Table 8 reveal that the innovation speed model demonstrates a significantly lower average loss for both overall indicator averages (CVPATbenchmarkoverall IA: difference of average loss = − 1.336, p < 0.001) and overall liner model prediction benchmarks (CVPATbenchmarkoverall LM: difference of average loss = − 0.138, p < 0.05), Confirming the strong predictive validity of the model.
为了进一步估计模型的预测能力，我们进行了PLSpredict分析，计算均方根误差（RMES）以评估模型对结构和指标的预测性能。结果如表8所示，指示PLS-SEM中的所有指标的Q2值大于0，此外，九个一阶项目中有七个PLS-SEM中的结构显示出比那些更小的预测误差在线性回归模型（LM）中，表明中等预测能力。此外，还进行了交叉验证预测能力测试（CVPAT）。表8中的结果显示，创新速度模型显示了两个总体指标平均值的平均损失显著较低和总体线性模型预测基准。
4.4假设检验的结果
Hypothesis 1 predicts the mediating effects of EO on the relationship between U-F的R&D alliances and innovation speed. The results in Table 7 support this hypothesis. Specifically, U-F的R&D alliances are positively related to EO (β = 0.797, p < 0.001) with the effect size of f2 = 0.974, and EO is significantly and positively associated with innovation speed (β = 0.236, p < 0.05; f2 = 0.024), together leading to a positive and significant indirect effect of U-F的R&D alliances on innovation speed through EO (indirect effect = 0.188, standard deviation = 0.077, p =0.015, 95%CI [0.031, 0.337]). Since the direct effects of U-F的R&D alliances on innovation speed (β = 0.163, p < 0.10; f2 = 0.011) are still significant when EO is included, the relationship between U-F的R&D alliances and innovation speed is partially mediated by EO.
假设1预测了EO对U-F的R&D联盟与创新速度关系的中介作用。表7中的结果支持这个假设。具体而言，U-F的R&D联盟与EO正相关，效应大小为f2=0.974，EO与创新速度显著正相关，共同导致阳性和U-F的R&D联盟通过EO对创新速度的间接影响显著。由于U-F的R&D联盟对创新速度的直接影响在包含EO时仍然显著，因此U-F的R&D联盟与创新速度的关系联盟和创新速度部分受EO的影响。
Hypothesis 2 posits the moderating effects of U-F informal contacts frequency on the relationship between U-F的R&D alliances and innovation speed. As shown in Table 7, while U-F informal contacts frequency is positively and significantly associated with innovation speed (β =0.146, p < 0.05; f2 = 0.020), the interaction term between U-F的R&D alliances and U-F informal contacts frequency is negatively associated with EO (β = − 0.114, p < 0.10; f2 = 0.014). Depicted in Fig. 3, the results of slope analysis illustrate that the positive relationship between UF R&D alliances and innovation speed is more pronounced when U-F informal contacts frequency is low, supporting Hypothesis 2.
假设2假设U-F非正式接触的调节作用U-F研发联盟与创新关系的频率-速度。如表7所示，而U-F非正式接触频率与创新速度呈显著正相关，U-F研发联盟与U-F非正式接触频率之间的互动项呈负相关用EO.如图3所示，斜率分析结果表明U-当U-F非正式接触频率较低时，R&D联盟和创新速度更为明显，支持假设2。
To further explore the effects U-F interactions, we conducted a multigroup analysis (MGA) among firms with different development stages upon entry into science parks (Armanios et al., 2017), categorizing the sample into three groups: (1) R&D stage group (G1) (development stages = 1 and 2), (2) market introduction stage (development stages = 3 and 4), and (3) market growth stage (development stage = 5). Prior to MGA, we assessed measurement invariances of composite models (MICOM) to ensure the measurement models to exhibit the same attributes under different conditions. The results in Table 9 show that all correlations (c) are close to 1 with insignificant permutation p-values of p > 0.10, Confirming compositional invariance and indicating no significant differences in the means and variances of measures across groups. Consequently, full measurement invariance is established, enabling us to proceed with MGA. The results of MGA reveal a significant difference in the effects of U-F informal contacts frequency on innovation speed between firms at the market introduction stage and those at the market growth stage (Δβ = − 0.594, p = 0.077).
为了进一步探索U-F相互作用的影响，我们在进入科学园后处于不同发展阶段的企业中进行了多组分析（MGA），将样本分为三组：（1）R&D阶段组（G1）（发展阶段1和2），（2）市场引入阶段（发展阶段3和4），以及（3）市场增长阶段（发展阶段5）。在MGA之前，我们评估了复合的测量不变性模型（MICOM），以确保测量模型在不同条件下表现出相同的属性。表9中的结果显示，所有相关性（c）都接近1，p>0.10的不显著排列p值，证实了组成不变性，并表明各组之间测量的平均值和方差没有显著差异。因此，建立了完全的测量不变性，使我们能够继续进行MGA。MGA的结果表明，U-F非正式接触频率对市场导入阶段企业和市场成长期企业创新速度的影响存在显著差异。
4.5.稳健性检查
We conducted two additional analyses to assess the robustness of our findings in PLS-SEM: endogeneity and unobserved heterogeneity.
我们进行了两项额外的分析来评估我们在PLS-SEM中的发现的稳健性：内生性和未观察到的异质性。
4.5.1.检查内生性
In PLS-SEM, endogeneity often arises when the error term of the dependent construct correlates with the independent variables (Sarstedt et al., 2020). Considering PLS-SEM estimates partial model structures simultaneously, recent research by Hult et al. (2018) recommended a systematic procedure based on Park and Gupta (2012)'s Gaussian copula approach for testing endogeneity, which is particularly suitable for assessing potentially endogenous constructs that are not normally distributed in PLS-SEM.
在PLS-SEM中，当依赖构建体的误差项与自变量相关时，通常会出现内生性，考虑PLS-SEM估计部分模型结构同时，霍特等人（2018）最近的研究推荐了基于Park和Gupta（2012）的高斯copula方法测试内生性的系统程序，特别适用于评估在PLS-SEM中不正态分布的潜在内生性构建体。
Following this recommendation, we examined endogeneity for each of the main predictor constructs in PLS-SEM. Consistent with prior research (Sarstedt et al., 2020), we performed the Kolmogorov-Smirnov test with Lilliefors correction on the latent variable scores of these predictor constructs. For the mediator of EO, the main predictor construct is U-F alliances. Hence, we initially assessed its potential endogeneity for EO. The results of Model 1 in Table 10 show that the Gaussian copula for U-F alliances is insignificant (cU-F RDA = 0.059, p =0.539), Confirming that endogeneity is not a concern for U-F的R&D alliances in predicting EO. For the dependent construct of innovation speed, with three main predictor constructs - EO, U-F的R&D alliances, and informal contact frequency - we evaluated the significance of their Gaussian copulas. The results in Table 10 demonstrate that Gaussian copulas yielded for EO (cEO = − 0.146, p = 0.446), U-F的R&D alliances(cU-F RDA = − 0.277, p = 0.114) and informal U-F contact frequency (cU-F ICFreq. = 0.480, p = 0.486) are all insignificant (p > 0.10) (Model 8). We also assessed all other possible combinations of Gaussian copulas (Model 2 to Model 7 in Table 10) and found that none of these copulas is significant (min(p) = 0.126). These results together Confirm that endogeneity is not present in our PLS-SEM analysis.
根据这一建议，我们在PLS-SEM中检查了每个主要预测因子构建体的内生性。与之前的研究一致，我们对这些预测因子结构的潜在变量得分进行了带有Lilliefors校正的Kolmogorov-Smirnov检验。对于EO的介质，主要的预测因子结构是U-F联盟。因此，我们最初评估了其潜力EO的内生性。表10中模型1的结果表明U-F联盟的高斯copula不显著，证实内生性不是U-F的R&D联盟在预测EO时关注的问题。对于创新速度的依赖结构，通过三个主要的预测因子结构——EO、U-F的R&D联盟和非正式接触频率——我们评估了它们的高斯copulas的显著性。表10中的结果表明高斯EO、U-F研发联盟产生的copulas和非正式U-F接触频率均不显著（p>0.10）（模型8）。我们还评估了高斯copulas的所有其他可能组合（表10中的模型2至模型7）并发现这些copulas都不显著（min（p）=0.126）。这些结果共同证实了在我们的PLS-SEM分析中不存在内源性。
Given EO's dual role as both a predictor construct for innovation speed and a dependent construct with U-F的R&D alliances as its predictor, we conducted another endogeneity test by using Heckman and Robb (1985)'s two-stage control function (CF) approach to ensure the robustness of our results. In the first stage, EO was estimated as a function of instrumental variables. We used previous experience of knowledge sharing and learning from failure as instrumental variables, along with the same control variables used in the PLS-SEM model, to estimate EO. Drawing on studies that argue a relationship between knowledge sharing (De Clercq et al., 2013) and learning from failure (Covin et al., 2006) with entrepreneurial orientation, we measured these experiences on a 7-point Likert scale with nine items each. The results indicate that both knowledge sharing experience (β = 0.643, p < 0.001) and learning from failure (β = 0.242, p < 0.001) are significantly associated with EO. In the second stage, the residuals (r_e) from the firststage model were included as an additional regressor in the innovation speed model. We found that EO is significantly associated with innovation speed (β = 0.346, p < 0.01), while the effects of r_e (β = 0.085) are insignificant (p = 0.206), providing evidence against the presence of endogeneity and supporting the robustness of our results.
鉴于EO作为创新速度的预测因子结构和以U-F的R&D联盟为预测因子的依赖结构的双重作用，我们使用Heckman和Robb（1985）的两阶段控制函数（CF）方法进行了另一个内生性检验，以确保我们结果的稳健性。在第一阶段，EO被估计为工具变量的函数。我们使用以前的知识共享和从失败中学习的经验作为工具变量，以及PLS-SEM模型中使用的相同控制变量来估计EO。根据论证知识共享和从失败中学习与创业导向之间关系的研究，我们用7分李克特量表测量了这些经历，每个量表有9个项目。结果表明，知识共享经验和从失败中学习都有显著性差异与EO有关。在第二阶段，第一阶段模型的残差作为额外的回归量包含在创新速度模型中。我们发现EO与创新速度显著相关，而r_e（β 0.085）的影响不显著（p =0.206），这提供了反对内生性存在的证据，并支持了我们结果的稳健性。
4.5.2.检查异质性
We employed the finite mixture partial least squares (FIMIX-PLS) procedure on the data (Hair et al., 2017) to account for unobserved heterogeneity. Consistent with prior research (Carlson et al., 2019; Gelhard et al., 2016; Matthews et al., 2016), we utilized the stopping criterion of 10− 10 = 1.0E-10, a maximum number of iterations of 5000 and the number of repetitions of 10 for each of segments (g = 1 to 5) proceeded by the FIMIX-PLS algorithm. The Akaike information criterion (AIC), modified AIC3 and AIC4, Bayesian information criterion (BIC), heuristic consistent AIC (CAIC), minimum description length with factor 5 (MDL5) and normed entropy statistics (EN) (Carlson et al., 2019; Gelhard et al., 2016) were used to determine the appropriate number of segments. As indicated in Table 11, the results suggest AIC3 and CAIC point to the same number of segments, a five-segment solution. However, MDL5 suggests a one-segment solution, indicating that more than one segment should be extracted (Hair et al., 2017). Taken together, these findings imply that fsQCA should be conducted to further explore unobserved heterogeneity (Carlson et al., 2019)
我们对数据采用了有限混合偏最小二乘（FIMIX-PLS）程序来解释未观察到的异质性。与之前的研究一致，我们利用了停止标准为10-10 1.0 E-10，最大迭代次数为5000，由FIMIX-PLS算法进行的每个片段（G1至5）的重复次数为10。Akaike信息准则(AIC)，修正的AIC3和AIC4，贝叶斯信息准则（BIC）、启发式一致AIC（CAIC）、因子5的最小描述长度（MDL5）和归一化熵统计（EN）用于确定适当的分段数量。如表11所示，结果表明AIC3和CAIC指向相同数量的段，一个五段解。怎么-迄今为止，MDL5建议了一个单段解决方案，表明超过应提取一个片段（Hair等人，2017年）。综上所述，这些发现意味着应进行fsQCA，以进一步探索未观察到的异质性（Carlson等人，2019）。
4.6 fsQCA
We took several steps to conduct fsQCA. In the first step, we calibrated our measures of variables defined previously into a 0–1 scale with multiple scores in between by using three qualitative thresholds: fullyin, the crossover point, and fully-out membership (Fiss, 2011). Adhering to the guidelines of Rasoolimanesh et al. (2021), we extracted standardized PLS-SEM latent variable scores from the original model's PLS-SEM algorithm, and following the majority of studies in the innovation and entrepreneurship literature, we adopted the direct approach to set the thresholds for membership categorization at the 95th percentile for “fully-in”, the 50th percentile for the crossover point, and the 5th percentile for “fully-out” (Pappas and Woodside, 2021). To address methodological challenges, a constant of 0.001 was added to measures with a calibrated score of exactly 0.5, as recommended by Fiss (2011).
我们采取了几个步骤来进行fsQCA。在第一步中，我们通过使用三个定性阈值，将之前定义的变量测量值校准为0-1的量表，中间有多个分数：完全输入、交叉点和完全输出成员（Fiss,2011）。遵守Rasoolimanesh等人（2021）的指导方针，我们提取来自原始模型的PLS-SEM算法的标准化PLS-SEM潜在变量分数，并遵循创业文献中，我们采用了直接的方法将成员资格分类阈值设置为“完全加入”的第95百分位、交叉点的第50百分位，以及“完全出局”的第5百分位。到为了解决方法上的挑战，按照Fiss（2011年）的建议，将0.001的常数添加到校准分数正好为0.5的测量中。
Next, we examined necessary conditions for innovation speed, which exists if it has a consistency value greater than 0.90 (Greckhamer et al., 2018). Our analysis found no necessary conditions for innovation speed (max(consistency) = 0.773159).
接下来，我们研究了创新速度的必要条件，如果创新速度的一致性值大于0.90，创新速度就存在。我们的分析没有发现创新速度的必要条件（最大（一致性）=0.773159）。
Finally, we conducted a sufficiency analysis using the fsQCA truthtable algorithms for fuzzy sets (Ragin, 2008), which listed 2k logically possible configurations, where k is the number of conditions. We obtained a truth-table with 2k = 2048 (k = 11) logical combinations of causal conditions. Given that our sample of 268 exceeded 150, we applied a frequency cut-off of 3 cases (Fiss, 2011) to ensure that only substantial configurations are assessed (Kimmitt et al., 2020). In accordance with prior research in innovation (Marzi et al., 2023; Xie and Wang, 2020), we set the raw consistency threshold at 0.9 or higher and, following Ragin's (2008) recommendation, set proportional reduction in inconsistency (PRI) at PRI ≥ 0.75.
最后，我们使用模糊集的fsQCA真值表算法进行了充分性分析，该算法列出了2k种逻辑上可能的配置，其中k是条件的数量。我们观察-获得了一个具有2k=2048（k=11）个逻辑组合的真值表因果条件。鉴于我们的268个样本超过150个，我们应用了3个病例的频率截止值，以确保仅评估实质性配置。根据之前在创新方面的研究，我们将原始一致性阈值设置为0.9或更高，并根据Ragin（2008）的建议，将不一致性比例减少（PRI）设置为PRI= 0.75。
Table 12 presents the condition configurations for innovation speed, revealing two configurations associated with high innovation speed. They collectively exhibit an overall consistency (ocons) of 97.6% and an overall coverage (ocov) of 18.7%. Configuration 1 (C1), with a consistency (cons) of 98.4%, a raw coverage (rcov) of 15.1%, and a unique coverage (ucov) of 7.1%, applies to larger SISMEs lacking R&D intensity, with fewer high-skilled employees, in early development stages, having some export experiences, located in large science parks, and operating in dynamic industries and markets. These firms underscore the significance of both U-F的R&D alliances and EO, especially when informal contacts with universities are less frequent. Configuration 2 (C2) (cons = 0.981, rcov = 0.116, ucov = 0.036) underscores the importance of U-F的R&D alliances, EO, and frequent U-F informal contacts for older SISMEs lacking R&D intensity but with more high-skilled employees, in early development stages, having export experiences, located in large science parks, and operating in dynamic industry and market environments.
表12给出了创新速度的条件配置，揭示了与高创新速度相关的两种配置。它们的总体一致性（ocons）为97.6%，总体覆盖率（ocov）为18.7%。配置1（C1）的一致性（cons）为98.4%，原始覆盖率（rcov）为15.1%，唯一覆盖率（ucov）为7.1%，适用于缺乏R&D强度、高技能员工较少、处于早期开发阶段、有一些出口经验、位于大型科学园并在动态行业和市场中运营的大型SISMEs。这些公司强调了U-F的R&D联盟和EO的重要性，尤其是当与大学的非正式接触不太频繁的时候。配置2（C2）强调了U-F的R&D联盟、EO和频繁的U-F非正式联系对于缺乏R&D强度但拥有更多高技能员工、处于早期开发阶段、具有出口经验、位于大型科学园并在动态行业和市场环境中运营的老SISMEs的重要性。
Using the same thresholds for consistency and frequency and calibrating low innovation speed as the inverse of high innovation speed, four configurations for low innovation speed (ocons = 0.969, ocov = 0.282) are found. As shown in Table 12, these four configurations (C3: cons = 0.960, rcov = 0.097, ucov = 0.034; C4: cons = 0.987, rcov = 0.092, ucov = 0.034; C5: cons = 0.984, rcov = 0.127, ucov = 0.044; C6: cons = 0.972, rcov = 0.108, ucov = 0.025) illustrate that the absence of U-F的R&D alliances, EO and U-F informal contacts frequency are significant conditions that lead to low innovation speed. Interestingly, for new and small firms even with high R&D intensity and more high-skilled employees (C5), they have low speed of innovation if they engage in limited U-F interactions and pursue a low degree of EO.
使用相同的一致性和频率阈值以及将低创新速度作为高创新速度的倒数，低创新速度的四种配置。如表12所示，这四种配置表明，缺乏U-F的R&D联盟、EO和U-F非正式接触频率是导致创新速度低的重要条件。有趣的是，对于新的和小的公司，即使R&D强度高，拥有更多的高技能员工（C5），如果他们参与有限的U-F互动并追求低程度的EO，他们的创新速度也很低。
These findings underscore the significance of U-F的R&D alliances, U-F informal contacts, and EO, configured with other organizational, science park, and environmental conditions, in influencing innovation speed. They thus provide additional support for our conceptual model, which argues the integrative effects of U-F interactions and EO on innovation speed.
这些发现强调了U-F的R&D联盟、U-F非正式联系和EO（与其他组织、科学园和环境条件一起配置）在影响创新速度方面的重要性。因此，它们为我们的概念模型提供了额外的支持，该模型论证了U-F相互作用和EO对创新速度的综合影响。
5.讨论和结论
Drawing on the dynamic capability perspective and university-firm interaction literature, our study investigates the impact of U-F interactions on innovation speed within the context of SISMEs. We distinguish between U-F的R&D alliances and U-F informal contacts. Our findings indicate that both forms of U-F interactions independently increase innovation speed within SISMEs. Notably, this study identifies firm-level EO as an effective mechanism through which U-F的R&D alliances amplify innovation speed. Moreover, we discover that the interaction between U-F的R&D alliances and frequent U-F informal contacts significantly influences innovation speed, with the latter potentially diminishing the positive effects of U-F的R&D alliances on innovationspeed. Our results also highlight that innovation speed results from multiple configurations of U-F interactions and EO, in conjunction with diverse organizational, science park, and environmental conditions.
利用动态能力视角和大学-企业互动文献，我们的研究调查了SISMEs背景下U-F互动对创新速度的影响。我们区分U-F的R&D联盟和U-F非正式接触。我们的发现表明，两种形式的U-F相互作用独立地提高了SISMEs内的创新速度。值得注意的是，本研究认为企业层面的EO是U-F的R&D联盟提高创新速度的有效机制。此外，我们发现U-F的R&D联盟和频繁的U-F非正式接触之间的相互作用显著影响创新速度，后者可能会削弱U-F的R&D联盟对创新的积极影响速度。我们的结果还强调，创新速度是U-F相互作用和EO的多种配置以及不同的组织、科学园和环境条件的结果。
5.1.理论含义
Our research offers several important theoretical implications. First, it underscores the critical role of U-F interactions in enhancing innovation speed, particularly within SISMEs. In providing evidence that supports the positive and significant effects of both U-F的R&D alliances and U-F informal contacts, our research addresses the call for a deeper understanding of the determinants influencing innovation speed (Cooper, 2021; Rosa, 2021). Specifically, although previous studies have acknowledged the value of U-F interactions in firm innovation (e.g., Caloghirou et al., 2021), much of the literature has centered around how inter-organizational collaborations with business entities (such as suppliers and customers) lead to rapid innovation and market success (Zhang and Wu, 2017). Our work enriches this narrative by delineating the significant effects of U-F interactions on firm innovation speed.
我们的研究提供了几个重要的理论意义。首先，它强调了U-F相互作用在提高创新速度方面的关键作用，特别是在SISMEs中。在提供证据支持U-F的R&D联盟和U-F非正式接触的积极和显著影响时，我们的研究呼吁更深入地理解影响创新速度的决定因素。具体来说，尽管先前的研究已经承认U-F互动在企业创新中的价值，但许多文献都集中在与商业实体（如供应商和客户）的组织间合作如何导致快速创新和市场成功。我们的工作通过描述U-F相互作用对企业创新速度的显著影响丰富了这一叙述。
Second, our study advances the research on U-F interactions by distinguishing between the effects of formal and informal interactions and underscoring their interactive effects. While the existing body of work has largely concentrated on formal interactions, particularly those defined by contractual R&D alliances (Schaeffer et al., 2020), our exploration of both U-F的R&D alliances and informal contacts deepens the understanding of how SISMEs can utilize these relationships to foster innovation speed. Their distinctive roles in innovation speed are also Confirmed by the findings associated with SISMEs' development stage. Specifically, our findings demonstrate that the effects of U-F informal contacts on innovation speed are more pronounced for SISMEs at the market introduction stage than for those at the market growth stage. In contrast, the effects of U-F的R&D alliances on innovation speed do not significantly vary across SISMEs at different development stages. These findings offer a novel insight that for SISMEs, U-F的R&D alliances consistently serve as a key driver of rapid innovation, irrespective of their firm development stage. However, U-F informal contacts have greater effects on innovation speed for SISMEs at the market introduction stage than for those at the growth stage. Our study therefore highlights an avenue for future research to explore how the lifecycle of SISMEs influences the dynamics between U-F informal interactions and innovation.
其次，我们的研究通过以下方式推进了U-F相互作用的研究区分正式和非正式互动的效果，并强调它们的互动效果。虽然现有的工作主要集中在正式互动上，特别是那些由合同R&D联盟定义的互动，但我们对U-F的R&D联盟和非正式接触的探索加深了对SISMEs如何利用这些关系来促进创新速度的理解。与SISMEs发展阶段相关的发现也证实了它们在创新速度中的独特作用。具体来说，我们的研究结果表明，U-F非正式接触对创新速度的影响在市场引入阶段比在市场增长阶段更明显。相比之下，U-F的R&D联盟对创新速度的影响在不同发展阶段的SISMEs之间没有显著差异。这些发现提供了一个新的见解，即对于SISMEs来说，U-F的R&D联盟始终是快速创新的关键驱动力，无论其公司发展阶段如何。然而，U-F非正式接触对市场导入阶段的SISMEs创新速度的影响大于成长期的SISMEs。因此，我们的研究强调了未来研究的一个途径，以探索SISMEs的生命周期如何影响U-F非正式互动和创新之间的动态。
More importantly, our research adds valuable insights to the ongoing debate regarding the dynamics between U-F formal and informal interactions (e.g., Landry et al., 2010; Schaeffer et al., 2020). Our analysis demonstrates that although U-F的R&D alliances and frequent U-F informal contacts each independently contribute to enhancing innovation speed, their interactive effects do not always result in complementary benefits. Specifically, we find that the beneficial effects of U-F的R&D alliances on innovation speed are more pronounced in SISMEs with less frequency in U-F informal contacts. Further insights from our fsQCA analysis indicate that this trade-off relationship is especially prominent in larger SISMEs with lower R&D intensity and a lack of high-skilled employees (as detailed in C1 in Table 12). This finding is pivotal as it underscores that the positive impact of U-F的R&D alliances on innovation speed is influenced not just by how these alliances are managed but also by how effectively and frequently firms engage in informal interactions for knowledge transfer.
更重要的是，我们的研究为正在进行的关于U-F正式和非正式互动之间动态的辩论。我们的分析表明，尽管U-F的R&D联盟和频繁的U-F非正式接触各自独立地有助于提高创新速度，但它们的互动效应并不总是导致互补的利益。具体来说，我们发现U-F的R&D联盟对创新速度的有益影响在U-F非正式接触频率较低的SISMEs中更为明显。我们的fsQCA分析的进一步见解表明，这种权衡关系在R&D强度较低且缺乏高技能员工的大型中小科技型企业中尤为突出（如表12中C1所详述）。这一发现至关重要，因为它强调了U-F的R&D联盟对创新速度的积极影响不仅受这些联盟管理方式的影响，还受企业参与知识转移非正式互动的有效性和频率的影响。
Last, our analysis sheds light on the pivotal role of firm-level EO in harnessing U-F的R&D alliances to expedite innovation speed in SISMEs. On one side, by examining the relationship between EO and innovation speed, we demonstrate the importance of fostering EO within SISMEs, which might otherwise lack extensive market experience, to achieve market-oriented results such as rapid innovation. On the other side, the findings of the U-F的R&D alliances-EO-innovation speed relationship provides insights into the nuanced outcomes of R&D alliances (Caloghirou et al., 2021). Specifically, our results from PLS-SEM analysis indicate that the integration of U-F的R&D alliances with EO is an effective way to benefit from these alliances. Through its emphasis on innovativeness, risk-taking, and proactivity, EO empowers SISMEs to effectively leverage U-F的R&D alliances, thereby accelerating their innovation.
最后，我们的分析揭示了企业层面的EO在利用U-F的R&D联盟加快SISMEs的创新速度。一方面，通过研究EO和创新速度之间的关系，我们证明了在SISMEs中培养EO的重要性，否则SISMEs可能缺乏丰富的市场经验，以实现以市场为导向的结果，如快速创新。另一方面，U-F的R&D联盟-EO-创新速度关系的发现提供了对R&D联盟微妙结果的见解。具体来说，我们来自PLS-SEM分析的结果表明U-F的R&D联盟与EO的整合是从这些联盟中受益的有效途径。通过对创新的重视-积极、冒险和积极，EO使SISMEs能够有效地利用U-F的R&D联盟，从而加速他们的创新。
The findings from fsQCA analysis also enrich our comprehension of the complex relationship between U-F interactions, EO and innovation speed. They not only solidify the necessity of integrating U-F的R&D alliances and U-F informal contacts with EO to attain heightened innovation speed but also show how such integration operates synergistically with various organizational characteristics (e.g., firm age, size, development state, etc.), the influence of science parks, and environmental conditions (e.g., environmental dynamism). These findings contribute novel insights to the literature on dynamic capabilities by demonstrating how strategic resource configuration represents an important lever to expedite innovation (Zhang and Wu, 2017)
fsQCA分析的发现也丰富了我们对U-F相互作用、EO和创新速度之间复杂关系的理解。它们不仅巩固了整合U-F的R&D联盟和U-F与EO的非正式联系以提高创新速度的必要性，而且还展示了这种整合如何与各种组织特征（如公司年龄、规模、发展状态等协同运作)、科学园的影响和环境条件（如环境动态）。这些发现通过展示战略资源配置如何代表一个重要的杠杆，为动态能力的文献提供了新的见解加速创新。
5.2.管理影响
Our research offers important insights for management practice, revealing that both U-F formal and informal interactions play a vital role in enhancing the speed of innovation within SISMEs. Consequently, SISMEs should deliberately participate in both types of interaction and develop strategies to maximize the effectiveness of their overall U-F interactions. Our findings underscore that U-F的R&D alliances can significantly contribute to innovation speed, both directly and through the mechanism of EO. Thus, SISMEs should not only pursue R&D alliances but also carefully integrate these alliances with entrepreneurial behaviors to fully capitalize on their benefits for accelerating innovation. This approach is particularly crucial for SISMEs constrained by limited resources, such as modest R&D investments, a smaller pool of highly skilled employees, and sparse informal university contacts.
我们的研究为管理实践提供了重要的见解，揭示了U-F正式和非正式的互动在提高SISMEs的创新速度方面发挥着至关重要的作用。因此，SISMEs应该有意识地参与这两种类型的相互作用，并制定策略来最大限度地提高其整体U-F相互作用的有效性。我们的发现强调，U-F的R&D联盟可以直接和通过EO机制显著促进创新速度。因此，SISMEs不仅应该追求R&D联盟，还应该仔细地将这些联盟与创业行为相结合，以充分利用它们的优势来加速创新。这种方法对于资源有限的SISMEs尤其重要，例如R&D投资有限、高技能员工数量较少以及非正式大学联系稀疏。
Additionally, our study underscores EO's pivotal role as a firm-level mechanism that enhances innovation speed and amplifies the impact of U-F的R&D alliances on this speed. It is imperative for top management to embrace EO and foster an entrepreneurial culture in both U-F interactions and innovation initiatives. Echoing the sentiment that EO can “serve as the guidelines for firm-specific behavior” (Klein et al., 2021), our research advises SISMEs to cultivate an organizational climate that encourages entrepreneurial behavior towards resource integration and the acceleration of innovation. This includes promoting new ideas and experimentation, viewing failure as an opportunity for growth, fostering a forward-thinking mindset among employees, and more. Firms should consider adapting their structure and human resources policies to support decentralized decision-making and autonomous groups (Wales et al., 2011), thereby facilitating entrepreneurial learning and empowering staff to undertake entrepreneurial actions. The strong linkage between U-F的R&D alliances and EO suggests that leveraging EO to boost innovation speed necessitates strategic management of U-F的R&D alliance efforts.
此外，我们的研究强调了EO作为企业层面机制的关键作用，它提高了创新速度，并放大了U-F的R&D联盟对这一速度的影响。高层管理人员必须接受EO，并在U-F互动和创新计划中培养创业文化。呼应EO可以的观点“作为企业特定行为的指南”，我们的研究建议SISMEs培养一种组织氛围，鼓励企业家行为整合资源和加速创新。这包括推广新想法和实验，将失败视为成长的机会，在员工中培养前瞻性思维，等等。企业应考虑调整其结构和人力资源政策，以支持分散决策和自治团体，从而促进创业学习并赋予员工采取创业行动的权力。U-F的R&D联盟和EO之间的紧密联系表明，利用EO来提高创新速度需要对U-F的R&D联盟的努力进行战略管理。
Furthermore, our analysis via PLS-SEM reveals that U-F的R&D alliances tend to be more advantageous for innovation speed when the frequency of informal U-F contacts is low. Therefore, SISMEs with fewer informal interactions are advised to intensify their formal interactions with universities through R&D alliances in order to accelerate innovation within their firm. Our fsQCA findings also show that the integration of U-F的R&D alliances and informal contacts with other factors affects innovation speed. SISMEs should be mindful of the potential substitutive effects of U-F formal and informal interactions and the conditions leading to various U-F interaction configurations. Specifically, larger SISMEs facing human resource constraints, such as a lack of highly skilled employees, could strategically capitalize on university knowledge for innovation via their entrepreneurial processes. In contrast, well-resourced SISMEs might benefit from a blend of U-F的R&D alliances and frequent informal contacts, coupled with an entrepreneurial orientation, to drive rapid innovation.
此外，我们通过PLS-SEM的分析表明，当非正式U-F接触的频率较低时，U-F的R&D联盟往往更有利于创新速度。因此，建议非正式互动较少的企业通过R&D联盟加强与大学的正式互动，以加速企业内部的创新。我们的fsQCA发现还表明，U-F的R&D联盟的整合以及与其他因素的非正式联系会影响创新速度。SISMEs应注意U-F正式和非正式相互作用的潜在替代效应以及导致各种U-F相互作用构型的条件。具体来说，面临人力资源限制（如缺乏高技能员工）的大型企业可以通过其创业过程战略性地利用大学知识进行创新。相比之下，资源充足的SISMEs可能会受益于U-F的R&D联盟和频繁的非正式接触的结合，加上创业导向，以推动快速创新。
5.3.局限性和未来研究
While our study provides valuable insights, its findings should be viewed in light of certain limitations that pave the way for future research. First, our research focused on EO as a firm-level mechanism, and the results have shown that EO is a strong but partial mediator in the relationship between U-F的R&D alliances and innovation speed. This underscores the need for future studies to investigate additional mechanisms that could complement EO in elucidating the dynamics between U-F的R&D alliances and innovation speed. For example, recent studies suggest that business model innovation plays a crucial role in the EOinnovation performance nexus (Ferreras-M´endez et al., 2021). Future research could incorporate business model innovation into our conceptual model to further explain the relationship between U-F的R&D alliances and innovation speed. Second, our analysis was limited to a single characteristic of U-F informal contacts, its frequency, to examine the interactive effects of U-F formal and informal interactions. Considering the resource-intensive and complex nature of innovation in SISMEs, future inquiries could benefit from examining a broader spectrum of U-F interaction characteristics that might influence innovation speed. Third, like many previous studies in the innovation literature, our analysis is based on cross-sectional data. While it is plausible to assume the significance of U-F的R&D alliances, EO, and U-F informal contacts for concurrent innovation speed, longitudinal research would provide insights into the evolving dynamics of these relationships and their impact on future innovation acceleration. Finally, our sample, drawn from science-intensive industries, may limit the generalizability of our findings to other sectors. Future studies could investigate the relevance of UF R&D alliances and EO for innovation across a broader range of sectors.
虽然我们的研究提供了有价值的见解，但其发现应根据为未来研究铺平道路的某些局限性来看待。首先，我们的研究重点是作为企业层面机制的EO，结果表明，在U-F的R&D联盟和创新速度之间的关系中，EO是一个强有力但部分的中介。这强调了未来研究的必要性，以调查额外的机制，这些机制可以补充EO，以阐明U-F的R&D联盟和创新速度之间的动态。例如，最近的研究表明，商业模式创新在EO创新绩效关系中起着至关重要的作用。未来的研究可以将商业模式创新纳入我们的概念模型，以进一步解释U-F的R&D联盟与创新速度之间的关系。其次，我们的分析仅限于U-F非正式接触的一个特征，即其频率，以检验U-F正式和非正式互动的互动效应。考虑到SISMEs创新的资源密集型和复杂性，未来的研究可以从检查可能影响创新速度的更广泛的U-F相互作用特征中受益。第三，与创新文献中的许多先前研究一样，我们的分析基于横截面数据。虽然假设U-F的R&D联盟、EO和U-F非正式联系对并行创新速度的重要性似乎是合理的，但纵向研究将提供对这些关系演变动态及其对未来创新加速的影响的见解。最后，我们的样本来自科学密集型行业，可能会限制我们的发现对其他行业的普遍性。未来的研究可以调查U-F的R&D联盟和EO对更广泛行业创新的相关性。
In conclusion, our study addresses call for more research on how science-intensive SMEs can accelerate their innovation. We highlight EO as a pivotal mechanism through which SISMEs can leverage the benefits of U-F的R&D alliances for rapid innovation under specific configurational conditions. Additionally, we uncover the substitution effects between UF R&D alliances and U-F informal contacts on innovation speed. Our findings offer novel perspectives and recommend strategies to counterbalance the potential drawbacks of both U-F formal and informal interactions on innovation, thereby contributing to the broader discourse on enhancing innovation in science-intensive sectors.
总之，我们的研究呼吁对科学密集型中小企业如何加速创新进行更多的研究。我们强调EO是一种关键机制，通过这种机制，SISMEs可以利用U-F的R&D联盟的优势，在特定的配置条件下进行快速创新。此外，我们揭示了U-F的R&D联盟和U-F非正式联系对创新速度的替代效应。我们的发现提供了新的视角，并推荐了策略来平衡U-F正式和非正式互动对创新的潜在缺点，从而有助于更广泛的讨论关于加强科学密集型部门的创新。
本文作为实证类文章，系统分析了中国中小科技型企业关于企业学校合作与创新速度之间的关系，利用动态能力理论提出了创业导向对企业学校合作与创新速度关系间的中间作用，用PLS-SEM检验了前期假设，并用fsQCA探讨影响高管创新的前因因素。文章回顾现有研究中小科技型企业实施创新的重要性，补充了现有的研究，为未来研究提供了新的理论框架和分析模型。
附件
