1 Introduction
During the past two decades, building information modeling (BIM) has emerged as an innovative transformation, overcoming traditional inefficiency, high cost, and risks through its advanced integration of information in the AEC industry (
Caglayan and Ozorhon, 2023). Although there is a general appreciation for its potential to bring radical change in the AEC sector with BIM, actual BIM usage compared to what is expected remains extremely low and constitutes a great deal of wasted resources. In the United States alone, an estimated $11 billion annually is lost due to inefficiency in facility operations (
Durdyev et al., 2022). Therefore, the broad diffusion of BIM has become urgent to develop resource-efficient practices within the AEC industry. In light of this, a profound analysis of the influencing factors of BIM adoption and its operational mechanism is of significant importance to BIM’s successful adoption and policy formulation.
Several researchers have investigated the motivations behind BIM adoption from different theoretical perspectives. Many studies refer to Rogers’ (
1995) Diffusion of Innovation Theory (DIT) to seek an understanding of how technical characteristics influence BIM adoption (
Olawumi et al., 2018;
Cao et al., 2016;
Shirowzhan et al., 2020). Given that BIM impacts transcend mere technical issues, some scholars seek to apply Institutional Theory (INT) (
DiMaggio and Powell, 1983), to understand the part played by social motivations in the adoption of BIM within organizations, by considering the impact of the outside environment (
Cao et al., 2014;
Cao et al., 2016). Furthermore, scholars have employed the technology-organization-environment (TOE) framework (
Tornatzky and Fleischer, 1990), to integrate factors influencing BIM adoption and test the interdependencies of technology, organizational context, and environmental background on BIM adoption (e.g.,
Qin et al., 2020;
Belay et al., 2021; Saka et al., 2024). On an individual level, scholars have utilized the Technology Acceptance Model (TAM) by Davis (
1989), to examine how the behavioral beliefs of individual users influence BIM adoption (
Chung et al., 2009;
Park et al., 2012;
Aladağ et al., 2023).
Although these four theoretical frameworks produce several insights into the drivers and mechanisms that underpin BIM adoption, no single theoretical perspective can guarantee a full explication of the entire BIM adoption process. Currently, there is still a lack of research that integrates these four perspectives to chart precisely the overall path of BIM adoption. Moreover, with unceasing research development, contradictions in certain findings have emerged (e.g.,
Xu et al., 2014;
Son et al., 2015;
Chen et al., 2019;
Won et al., 2013;
Yuan et al., 2019), further contribute to the difficulty in comprehensively grasping all factors that influence BIM adoption and their inner mechanisms.
In recent years, several meta-analyses have been conducted on the topic of BIM adoption. Oesterreich and Teuteberg (
2019) investigated barriers from the perspective of information systems, categorizing socio-technical causes into four dimensions: structure, people, task, and technology. Shahruddin et al. (
2021) reviewed and defined the main framework of BIM capabilities required for building practitioners, emphasizing the user perspective. More specifically, Abideen et al. (
2022) focused on the operational phase of facility management and explored the drivers of BIM usage, dividing them into three categories: technical, organizational, and legal/contractual. Despite these meta-analyses encompassing a broad array of perspectives, they have not integrated internal and external environmental factors along with individual-level psychological factors. Moreover, the implications of the complementarities or interactions among the multiple theoretical frameworks under study remain unexplored.
Given the theoretical and practical imperatives outlined above, it’s time to reassess the nature of what we currently know and don’t know about “promoting” and “hindering” BIM adoption, using the results of empirical research to inform future model applications. Consequently, we conducted a meta-analysis that could suggest an extended framework where intrinsic antecedents (individual variables in TAM) as mediators between extrinsic antecedents (external variables in TOE, DIT, INT) and BIM adoption itself. This study investigated what factors affect BIM adoption by mediating effects targeted via two different paths and further provided quantitative analyses respective to the degree of various factors affecting BIM adoption. We also extend prior meta-analyses by adding national conditions (BIM maturity in different countries) and contextual factors (organization type, job level, and research publication time) as moderating variables. This improvement aims to identify the key boundary conditions within which policy interventions should be drawn to promote BIM. This study contributes to the growing body of BIM adoption research by furthering a holistic understanding of the involved mechanisms. It integrates four theoretical perspectives explaining the potential pathways between BIM adoption and its antecedents, providing empirical validation for the comprehensive model through meta-analysis.
The rest of the paper is organized as follows: Section 2 provides the cognitive background and literature review. Section 3 describes the data collection and the overall research methodology. Section 4 presents the testing of the proposed model and the meta-analysis of the results. Section 5 discusses the findings, summarizes the limitations, and proposes directions for future research. Finally, Section 6 concludes the study. An overview of this paper is depicted in Fig.1.
2 Theoretical foundation and literature review
2.1 Definition of BIM adoption
During its early development phases, BIM was primarily regarded as sophisticated digital software (
Alathamneh et al., 2024). However, with ongoing advancements, BIM’s influence has expanded beyond technology to include social and organizational dimensions (
Papadonikolaki et al., 2019;
AlBalkhy et al., 2024). Accordingly, in this study, BIM is defined as the specific deployment of digital software by designated stakeholders, underpinned by relevant policies (such as national standards or industry guidelines), to facilitate particular processes (such as collaboration) throughout the lifecycle of a project (
Ahmad et al., 2016).
Rogers (1995), in the DIT, defines the “adoption” of innovative technology as “the decision of fully utilize innovation as the best available course of action.” Existing literature often conflates the terms “adoption” and “implementation” of BIM, leading to a lack of clear definitions (
Hochscheid and Halin, 2020). In alignment with the recommendations of Succar and Kassem (
2015), this study overlays the concepts of “implementation” and “diffusion” within the broader context of macro (i.e., market-wide) adoption onto “adoption.” It suggests that BIM adoption includes three consecutive stages: awareness, intention, and behavior. Thus, BIM adoption is viewed as a successful implementation process, further expanding the understanding of BIM adoption in a broader context.
Given the complex nature of BIM adoption, this study integrates the multi-dimensional theoretical frameworks of TOE, DIT, INT, and TAM. By examining external factors related to technology, organization, and environment alongside internal psychological factors, this meta-analysis systematically analyzes how these elements collectively shape the comprehensive process of BIM adoption, clarifying the varying levels of adoption behaviors.
2.2 Psychological factors as intrinsic antecedents
Perceived usefulness (PU) and perceived ease of use (PEU), as core components of the TAM, significantly influence users’ potential acceptance or rejection of technology (
Qin et al., 2020). According to TAM, a user’s cognitive beliefs regarding PEU and PU of a specific system affect their intention and behavior toward system usage, ultimately determining their actual use of the system (
Davis, 1989).
PU is defined as “the degree to which a person believes that using a particular system would enhance his or her job performance” (
Davis, 1989). In their systematic review of 30 years of TAM development, Davis et al. (
2024) confirmed that PU is a major determinant of user acceptance. They noted that potential users can form realistic and stable perceptions of usefulness even without having used the system (
Davis et al., 2024). Specifically, in the context of BIM adoption, PU, which reflects a cognitive belief regarding expected benefits, evolves through different stages (
Bhattacherjee and Lin, 2015). In the initial phase, employees’ expectations of BIM benefits are often based on indirect information, which may fluctuate and diverge from reality. This discrepancy is particularly pronounced among late adopters, who may lack the skills and resources necessary for effective BIM implementation (
Ma et al., 2020). When expectations do not align with reality, it can lead to cognitive dissonance among employees, prompting them to reassess their expectations through direct experiences with BIM (
Bhattacherjee and Lin, 2015). This iterative learning process, from initial expectations to refined perceptions, highlights the critical role of PU in facilitating BIM adoption.
PEU is defined as “the degree to which a person believes that using a particular system would be free of effort” (
Davis, 1989). It serves as a key secondary determinant in TAM, influencing user acceptance both directly and indirectly by enhancing PU (
Davis et al., 2024). Specifically, all else being equal, systems that are easier to use are more likely to be accepted by users. Although most literature positions PEU as a predictor of PU (
Davis, 1989), it is anticipated to present a positive correlation; that is, the less effort required by a system, the more beneficial it is perceived to be for users in completing tasks. However, some studies indicate that the relationship between PEU and PU is not always positive (
Zhao et al., 2023). The variability in the findings suggests that BIM adoption may be influenced by multiple factors. Consequently, to gain a better understanding of the mechanisms that underlie BIM adoption, it is essential to explore how various external factors impact this process.
2.3 Technical factors as external antecedents
Based on the Diffusion of Innovation Theory (DIT) (
Rogers, 1995), this study defines the external technical factors influencing BIM adoption as technical characteristics. We focus on relative advantage, compatibility, and complexity as the three factors most consistently correlated with innovation adoption (
Tornatzky and Klein, 1982).
The relative advantage of BIM refers to the extent to which it provides added benefits to an organization compared to traditional technologies (
Wang et al., 2021). When compared with conventional computer-aided software within the architecture, engineering, and construction (AEC) industry, BIM demonstrates superior intelligence and interoperability. These characteristics significantly enhance data sharing and project delivery processes (
Martínez-Carricondo et al., 2020).
Compatibility of BIM pertains to how well it aligns with existing values, past experiences, and the needs of potential adopters (
Shirowzhan et al., 2020). At the organizational level, compatibility manifests as alignment with collaborative management principles and business requirements. On a technical level, compatibility is expressed through suitable technical interfaces across various projects, facilitating the exchange of data and models among different users and software applications (
Cursi et al., 2022).
Complexity is defined as the negative conditions toward BIM before the decision to adopt it (
Ahmed and Kassem, 2018). DIT suggests that complexity negatively correlates with the rate of innovation adoption in social systems (
Rogers, 1995). In the context of BIM, complexity arises primarily from the challenges associated with learning the requisite technologies and tools, coupled with the fragmented nature and complex workflows typical of the AEC industry (
Chen et al., 2019). Numerous studies have validated this negative correlation, establishing that complexity is a critical factor influencing BIM adoption (
Ding et al., 2015;
Wang et al., 2021).
2.4 Organizational factors as external antecedents
In the exploration of external organizational factors influencing BIM adoption, this study identifies three primary determinants: top management support (
Wang and Song, 2017;
Wang et al., 2020b;
Siebelink et al., 2021), organizational readiness (
Liu et al., 2010), and organizational culture (
Munianday et al., 2022).
Top management support is vital for providing the essential resources—such as hardware, training, and technical assistance—necessary for the effective utilization of information systems (
Yoon and Guimaraes, 1995). As a nascent project information management system, BIM particularly benefits from the economic and policy backing of top management during the initial stages of adoption (
Boton et al., 2021). This support not only alleviates the financial burdens associated with implementation but also encourages organizational transformation, thereby increasing the likelihood of successful BIM adoption (
Song et al., 2017).
Organizational readiness refers to an organization’s capability to allocate the requisite resources for BIM adoption (
Iacovou et al., 1995). This includes providing training and education to enhance the BIM expertise of personnel, as well as making necessary software and hardware investments. Such efforts have a direct impact on the adoption of BIM (
Belay et al., 2021).
Organizational culture plays a crucial role in influencing innovation by cultivating an environment that promotes and supports innovative practices (
Martins and Terblanche, 2003). In the AEC industry, characterized by significant instability and loose coupling of organizational structures, the intricacies of organizational cultures are further exacerbated (
Alankarage et al., 2023). The temporary nature of projects in this sector causes professionals to adhere more strictly to established work habits, leading to a diminished acceptance of BIM and an overall resistance to change (
Zhang et al., 2020). In such a context, an organizational culture that is anchored in shared values and goals can help alleviate the barriers to BIM adoption, enabling professionals to assimilate new knowledge and processes more effectively (
Munianday et al., 2022).
2.5 Environmental factors as external antecedents
Institutional theory outlines the key external environmental factors that impact BIM adoption, primarily identifying coercive and mimetic pressures (
Cao et al., 2014). Coercive pressure is described as “both formal and informal pressures exerted on organizations by other organizations upon which they are dependent” (
DiMaggio and Powell, 1983). In the context of BIM adoption, coercive pressure often emanates from regulators and industry associations. On one hand, recognizing the potential benefits of BIM for enhancing project management efficiency, certain governments and their affiliated organizations have begun mandating the use of BIM in public projects (
Ahmed and Kassem, 2018). On the other hand, considering the significant investment costs associated with AEC projects, external pressures may compel organizational managers to revise their strategies to facilitate more proactive BIM adoption (
Cao et al., 2014).
Mimetic pressure refers to an organization’s choice to emulate the behaviors of successful peers when faced with uncertain circumstances (
DiMaggio and Powell, 1983). When goals are ambiguous, imitating successful cases serves as an effective strategy for risk mitigation (
March, 1963). Given that BIM is an emerging technology within the AEC industry, its adoption necessitates complex coordination and organizational changes across various departments, which increases uncertainty and complicates the adoption process (
Gao et al., 2023). As a result, organizations are often inclined to model their practices after successful projects undertaken by their peers to mitigate risks associated with BIM implementation.
2.6 The moderating effect of national BIM maturity
The readiness level for BIM in the AEC industry varies significantly across different countries (
Li et al., 2023). While some nations with low BIM adoption are actively researching and developing policies aimed at enhancing BIM implementation, their actual adoption rates still fall short of expectations. Conversely, countries with advanced BIM adoption have progressed to more sophisticated stages (
Charef et al., 2019), resulting in an increasing disparity compared to other nations. In this context, the BIM ecosystems across various countries exhibit multidimensional complexity, influencing stakeholders’ perceptions regarding the ease of use and usefulness of BIM, which in turn impacts its promotional effects.
Understanding the full range of differences within national BIM ecosystems will not only improve the targeting of national policies but also facilitate the effective adoption of BIM on a broader scale. Given that the BIM maturity model reflects the characteristics of its ecosystem, this meta-analysis will explore the potential moderating effect of national BIM maturity on both the direct and indirect pathways of BIM adoption.
In this study, we adopted Model B: Macro-Maturity Components, developed and validated by Succar and Kassem (
2015), as our national BIM maturity model. This model includes eight complementary components: objectives and milestones, champions and drivers, regulatory framework, noteworthy publications, learning and education, measurements and benchmarks, standardised parts and deliverables, and technology infrastructure (
Succar and Kassem, 2015). Each component is assessed across five levels of maturity: low, medium-low, medium, medium-high, and high maturity. Detailed criteria for maturity evaluation are outlined in Succar and Kassem (
2015).
Using the aforementioned national BIM maturity model, we classified the 62 independent studies included in the meta-analysis, which span 13 countries across six continents, into different BIM adoption maturity levels. This classification is based on the eight complementary components outlined in Appendix A1.
2.7 The moderating effect of contextual factors
The dissemination of digital innovation is a complex process influenced by various social systems, including temporal dynamics (
Shibeika and Harty, 2015). This complexity often leads to variations and inconsistencies in research findings, particularly within BIM adoption studies across different organizations and time frames (
Chen et al., 2019;
Wang et al., 2020a). To address these inconsistencies, this meta-analysis examines the moderating roles of job level, organization type, and time span in BIM adoption.
The increasing adoption of BIM within the AEC industry hinges on the collaborative efforts of all stakeholders within the BIM ecosystem. Human factors are fundamentally significant in this context, as they directly influence the decision-making processes associated with BIM adoption (
Adekunle et al., 2022). Notably, substantial variations in expertise, skills, and resource availability among stakeholders at different job levels not only shape their perspectives and attitudes toward BIM but also affect their decision-making behaviors (
Hosseini et al., 2018). Therefore, understanding how these job-level differences modulate the impact of BIM adoption is essential. This study categorizes the job levels across 62 independent studies into three groups: all employees, management, and general employees, to further explore how these differences moderate BIM adoption.
At the organizational level, various stakeholders pursue different goals, resulting in diverse motivations for adopting BIM. This diversity is also manifested in how stakeholders perceive the usefulness and ease of use of BIM, ultimately influencing adoption rates. To investigate the potential moderating effects of different organization types on BIM adoption, this study classifies the types of organizations represented in the 62 independent studies into five categories: all project stakeholders, construction, design, engineering and consulting services, and owners. This classification enables a comprehensive analysis of BIM adoption behaviors across various organizational types and provides a theoretical foundation for developing targeted promotion strategies.
Moreover, since each empirical study is cross-sectional, the findings are relevant only to BIM adoption at a specific point in time. However, the development and implementation of BIM are dynamic and time-dependent processes, with the factors influencing adoption continuously evolving as technology and policy directions change. Notably, in 2016, due to mandatory policies in advanced regions such as the United States, Europe, Singapore, and South Korea, countries at the forefront of BIM transitioned from initial application stages to a higher level of integration and collaboration, thereby widening the gap between them and countries with less developed BIM capabilities (
Charef et al., 2019;
Edirisinghe and London, 2015). To examine the specific impact of time on BIM adoption, this study will employ meta-analysis, dividing the research timeline into two phases: the first phase spanning from January 1, 2012, to December 31, 2016, and the second phase extending from January 1, 2017, to December 31, 2023. This approach aims to reveal the factors, mechanisms, and trends influencing BIM adoption at different stages over time.
2.8 Integrated framework for BIM adoption
BIM, as a disruptive and innovative technology, presents a complex and multi-layered adoption process. In addition to direct influences, the adoption of BIM is shaped indirectly by the interplay of various influencing factors. As illustrated in Fig.2, we anticipate that external factors,such as technology, organization, and environment, will indirectly affect BIM adoption through the psychological variables of PU and PEU, although these factors may also have a direct impact. Within this integrated framework, this study seeks to combine mainstream theoretical perspectives to explore the collective influence of external and internal factors on the comprehensive process of BIM adoption.
The TOE framework, proposed by Tornatzky and Fleischer (
1990), offers a structured approach for analyzing the adoption process of innovative technologies at the organizational level. Although the TOE framework is well-regarded for its simplicity and intuitive design, Dedrick and West (
2003) have noted its limitations, particularly in terms of an unclear structure and a generalized classification of variables that do not constitute a cohesive conceptual framework. To address these shortcomings, we have incorporated the DIT and the INT into the TOE framework, thereby creating a more comprehensive framework for studying BIM adoption.
Specifically, within the technological dimension, Rogers’ (
1995) DIT elucidates the ongoing spread, diffusion, and adoption process of new ideas, practices, or technologies within a social system, enhancing our understanding of how technological characteristics can either promote or inhibit innovation adoption. In the environmental dimension, the INT, as articulated by DiMaggio and Powell (
1983), provides a theoretical basis for examining how external environments—comprising norms and values—impact organizational innovation adoption (
Bhakoo and Choi, 2013), thereby enriching the analysis of environmental factors within the TOE framework.
In addition, building on the Theory of Reasoned Action (TRA) developed by Ajzen and Fishbein (
1980), Davis (
1989) proposed the TAM. TAM offers a structured framework for comprehending individual-level adoption of BIM by drawing theoretical connections between individual beliefs related to PU and PEU, as well as intentions and actions (
Davis, 1989). According to TAM, external variables can trigger individual cognitive responses, such as PU and PEU, which in turn influence behavioral responses concerning BIM adoption (
Gao et al., 2023). Nevertheless, the absence of clear definitions for these external variables often leads to inconsistent empirical findings and limited predictive power within TAM research (
Gangwar et al., 2014). To overcome this limitation, the present study integrates TAM with the TOE framework, DIT, and INT to formulate a comprehensive framework for BIM adoption. In this context, TAM acts as an intermediary, connecting external antecedents (the variables derived from TOE, DIT, and INT) to BIM adoption. This integrated framework facilitates the establishment of a causal chain labeled as “external variable – internal psychology – behavioral decision making,” captures BIM adoption behaviors at various levels, and presents a novel theoretical perspective on the interaction of multi-level influencing factors within the BIM adoption process.
Furthermore, current research presents mixed empirical results regarding the relationship between different types of motivation and BIM adoption outcomes (e.g.,
Cao et al., 2022;
Saka et al., 2020;
Cao et al., 2014;
Wang et al., 2020a). The mechanisms behind these discrepancies remain unclear, highlighting the inconsistencies and gaps in existing research regarding how these factors affect BIM adoption. Consequently, this study employs meta-analysis, utilizing the integrated framework to investigate the specific impact of various factors on BIM adoption and their mechanisms of action, while exploring the moderating effects of national BIM maturity (comprising eight components) and contextual factors (including job level, organization type, and time span). The aim is to elucidate the contributions of various factors to BIM adoption, thereby fostering a more comprehensive and systematic understanding of BIM adoption research.
3 Research method
3.1 Literature search and inclusion criteria
Since 2012, scholars have maintained a keen interest in exploring methods to promote BIM adoption, with the first article addressing factors influencing BIM adoption published in that year. Consequently, this meta-analysis establishes its keyword search timeframe from January 1, 2012, to December 31, 2023. To secure a sufficient sample size for the meta-analysis, we conducted searches across several databases, including Web of Science, Scopus, Elsevier, ProQuest, and CNKI. The search was guided by the following terms: “(BIM OR Building Information Model OR Building Information Modeling) AND (Adoption OR Implement OR Diffusion OR Barriers OR Drivers OR Factors OR Affecting) AND (AEC OR Architecture OR Engineering OR Construction OR Owner OR Project OR Organization).” Our initial search yielded a total of 23,698 articles. After excluding articles that did not meet our criteria based on their titles and abstracts, we retained 333 relevant papers.
In alignment with the research topic and the literature selection criteria outlined for the meta-analysis, we further screened the 333 papers according to the following four standards:
(1) The literature must comprise empirical studies that quantitatively analyze the factors influencing BIM adoption, excluding theoretical, review, and interview-based literature.
(2) The focus of the research should be specifically on BIM adoption within enterprises.
(3) The literature must provide specific data detailing the relationship between BIM adoption and influencing factors, such as correlation coefficients, means, standard deviations, sample sizes, and corresponding statistical tests like
F,
t, and χ
2, regression coefficients, and path coefficients (
Wolf, 1986;
Hunter and Schmidt, 2004;
Wu and Lederer, 2009;
Liao et al., 2024).
(4) The data set must be independent. In instances of duplicate data sets, the study presenting the most comprehensive report will be prioritized. Conference papers will not be duplicated with journal papers, and preference will be given to formally published journal articles.
Throughout the multiple rounds of literature screening, we also reviewed the reference and citation lists of relevant review papers to enhance our literature list and ensure thoroughness. Ultimately, we included 58 eligible papers for the meta-analysis. Among these studies, one paper contained three independent studies, while two included two independent studies each. As a result, our analysis included a total of 62 independent studies from 13 countries conducted between 2012 and 2023, representing a total of 11,228 study subjects. The specific screening process is illustrated in Fig.3.
3.2 Literature code
When conducting a meta-analysis, the primary objective is to extract relevant information and data from existing research. This process involves extracting and coding features of the literature as well as reporting data. In this study, the extracted data from each research article includes the title, authors, publication year, journal name, country of origin, organization type, job level, and other study characteristics. Additionally, statistical measures such as sample size, reliability coefficient (Cronbach’s Alpha), and correlation coefficients are incorporated. Furthermore, relevant theories associated with BIM adoption and the factors influencing its adoption are also included in the analysis.
Each factor that facilitates BIM adoption in the studies included in our meta-analysis must be classified by researchers into the corresponding categories within the BIM adoption integration framework. For example, the factor “Integration and accuracy of models (IAM)” identified in the study by Liao and Teo (
2017) has been classified under “compatibility” within the technical characteristics dimension. This classification is based on the premise that the compatibility of BIM entails the integration of BIM software to facilitate the sharing of model data (
Shirowzhan et al., 2020). The factors influencing BIM adoption identified in the 62 independent studies are outlined in the Supplementary Material A2. Fig.4 depicts the frequency of independent studies along with their sample sizes. These metrics are correlated with factors across various dimensions of the BIM adoption integration framework.
To ensure accuracy, scientific validity, and reliability in the coding process, an independent research team composed of two individuals undertook the entire literature coding process. During this phase, the researchers initially selected a small sample for trial coding. Following continuous discussions and revisions, a suitable coding scheme was developed for the sample data of this study. Using this scheme, the remaining studies were independently coded, achieving a consistency rate exceeding 90% in the coding results. Any discrepancies primarily stemmed from the subjective judgments of the coders. The researchers then reached a consensus through discussions, referencing the original texts and other relevant materials. The key data from the 62 independent studies can be found in the Supplementary Material A3, while the geographical distribution of BIM adoption studies is illustrated in Fig.5.
3.3 Meta-analysis process
The data in this study were processed using the professional software Comprehensive Meta-Analysis 3, which is a tool specifically designed for conducting meta-analyses. The specific data processing procedures can be found in Fig.6.
According to the meta-analysis procedure established by Glass et al. (1981), a meta-analysis involves synthesizing the results from multiple empirical studies that share a common theme and are independent of one another. Subsequently, individual effects are quantitatively consolidated into a unified effect size. Given that empirical studies on BIM adoption typically utilize correlation coefficients as outcome data, our study will adopt correlation coefficients (r) as the unified effect size.
Moreover, a prevalent challenge in meta-analytic research is the issue of publication bias. This bias arises from the tendency of academic journals to favor the publication of studies showcasing positive results, while those reporting negative or inconclusive findings are frequently rejected. To ensure the robustness of the conclusions and mitigate bias stemming from publication practices, it is essential for a meta-analysis to evaluate the publication bias present in the literature. In this study, we employed two established evaluation metrics to assess publication bias: Egger’s regression intercept and the Fail-Safe Number (FSN). A p-value for Egger’s regression intercept greater than 0.05 indicates a lack of significant publication bias (
Egger et al., 1997). The Fail-Safe Number represents the quantity of negative results that would need to be added to overturn the current conclusions of the meta-analysis (
Rosenthal, 1979). Specifically, when FSN exceeds 5
K+10 (where
K is the count of primary studies providing correlation coefficients), the likelihood of publication bias is minimal. A larger FSN suggests more stable results, thereby reducing the potential for reversal (
Rosenthal, 1979).
Following the correction for measurement and sampling errors, a heterogeneity test must be conducted to determine whether to apply a fixed effects model or a random effects model for significance assessment. The commonly used indicators for heterogeneity test are the
Q statistic and
I2. The
Q statistic can only assess the presence of heterogeneity among studies, while
I2 can also evaluate the degree of heterogeneity (
Huedo-Medina et al., 2006). Therefore, in this study,
I2 was selected as the evaluation criterion for testing heterogeneity. When
I2>75%, it indicates a highly significant heterogeneity test and a random effects model should be used; otherwise, a fixed effects model should be used (
Huedo-Medina et al., 2006).
Upon determining the appropriate effect model, final hypothesis testing will be conducted. This process includes main effect testing and moderating effect testing, aimed at assessing whether the aggregated correlation coefficients from multiple empirical studies demonstrate statistical significance.
4 Results
4.1 Determination of the effect size
The 62 independent studies included in this meta-analysis comprise a total of 11,228 valid samples. Prior to conducting the main effect test, it is essential to calculate the effect sizes reported in each study. To mitigate the attenuation bias resulting from the reliability deficiencies of the scale, a reliability correction must be applied to the correlation coefficient when it is utilized as the effect size in an empirical study (
Borenstein et al., 2021). The formula for reliability correction is as follows:
Note: is the correlation coefficient between the two variables, and represents the scale reliability coefficients of the independent variable and dependent variable respectively (Cronbach’s Alpha value).
Single empirical studies employing path coefficients as effect sizes do not necessitate reliability adjustment. This is due to the fact that the estimation of path coefficients in Structural Equation Modeling (SEM) involves measuring latent variables using multiple observed variables, akin to the instrumental variable method (
Wright, 1960). Measurement errors are already factored in, allowing the path coefficients to be directly considered as corrected correlation coefficients.
4.2 Publication bias analysis
The results of the publication bias analysis are presented in Tab.1. The coefficient of Egger’s regression intercept exceeds 0.05, indicating that bias cannot be detected in the meta-analysis (
Egger et al., 1997). When the (FSN is greater than 5
K+10 (where
K represents the number of primary studies providing correlation coefficients), it suggests a low likelihood of publication bias and indicates that the results of the meta-analysis are relatively stable (
Rosenthal, 1979). For instance, the effect size of BIM’s relative advantage on PU for stakeholders, based on Egger’s regression intercept, is −2.25094 with a
P of 0.63863 (exceeding 0.05), indicating that no bias can be established. The Fail-Safe Number is 1,824, implying that 1,824 non-significant studies would be necessary to overturn the results of the meta-analysis.
Regarding the FSN, all paths meet the criteria except for those concerning the complexity of BIM to PU and PEU. Additionally, in terms of the Egger test, with the exceptions of the two paths related to organizational culture and PU as well as BIM adoption, all other Egger tests lacked statistical significance. This suggests that the sample included in this study does not exhibit significant publication bias.
4.3 Heterogeneity test
The purpose of the heterogeneity test is to assess whether there is variance among the effect sizes. The fixed-effects model assumes that the variability observed in effect sizes results from within-study variation (sampling error), while the random-effects model attributes the observed variability to both within-study and between-study variation (
Lipsey and Wilson, 2001). If there is no heterogeneity among the effect sizes, the fixed-effects model should be utilized in the main effects analysis; otherwise, the random-effects model is recommended. The results of the heterogeneity test are presented in Tab.2. All
P of the
Q statistic are less than 0.05, with the exception of the coercive pressures of BIM on PEU, which is 0.875. The remaining values
I2 range from 0.79890 to 0.97685, all exceeding the critical value of 0.75. This indicates considerable total heterogeneity among the effect sizes, attributed to between-study variation that cannot be solely explained by sampling error. Furthermore, because the selected samples in this study originate from various countries, involve different types of organizations, and have different measurement methods and research focuses. Therefore, the random model is more appropriate for this meta-analysis.
4.4 Main effect test
Next, the study examined the effects of the technical characteristics of BIM, specifically relative advantage, compatibility, and complexity, on promoting BIM adoption. The findings indicate that both compatibility (PU: r = 0.648, PEU: r = 0.425, BIM Adoption [BA]: r = 0.398; P < 0.001) and relative advantage (PU: r = 0.527, PEU: r = 0.3364, BA: r = 0.488; P < 0.001) significantly promote BIM adoption. In contrast, the complexity of BIM (PU: r = −0.066, P = 0.570; PEU: r = −0.205, P = 0.207; BA: r = −0.198, P = 0.01) does not have a significant effect on fostering its adoption. Notably, the correlation coefficient between BIM compatibility and PU from stakeholders is the most substantial (r = 0.634). The forest plots illustrating the effects of BIM compatibility on PU and PEU among stakeholders are presented in Fig.7 and Fig.8.
Regarding the dimension of organizational support for BIM, it was found that top management support, organizational readiness, and organizational culture all contribute positively to BIM adoption. Among these factors, organizational culture exhibits the most significant effect on BIM adoption (PU: r = 0.578, PEU: r = 0.619, BA: r = 0.489), exceeding the influence of top management support (PU: r = 0.490, PEU: r = 0.399, BA: r = 0.446) and organizational readiness (PU: r = 0.263, PEU: r = 0.555, BA: r = 0.396), with all effects being statistically significant (P < 0.001).
In the analysis of the external environment surrounding BIM, both coercive and mimetic pressures were found to facilitate BIM adoption. Notably, the effect of mimetic pressure on BIM adoption (PU: r = 0.323, PEU: r = 0.299, BA: r = 0.301) is greater than that of coercive pressure (PU: r = 0.274, PEU: r = 0.208, BA: r = 0.298), and all effects are statistically significant (P < 0.001).
Path testing demonstrated that the indirect effects of external factors on BIM adoption via internal cognitive beliefs (PU and PEU) are more pronounced than their direct impacts, thereby affirming the mediating role of these cognitive beliefs. The specific data are presented in Tab.2.
4.5 Moderating effect test
4.5.1 National BIM maturity model
In this section, we conduct an in-depth analysis of the national BIM maturity model, examining the influence of eight complementary components on both the direct and indirect pathways of BIM adoption through intrinsic antecedents. The results of these moderating effects indicate: Component I (Objectives, Stages, and Milestones) and Component IV (Noteworthy Publications Level) demonstrate identical moderating effects. Our analysis indicates that in countries with a maturity level of 2 for both Components I and IV, PEU (BA: r = 0.935), top management support (PEU: r = 0.666, PU: r = 0.734), organizational culture (PEU: r = 0.952, PU: r = 0.907), and organizational readiness (r = 0.753) have a significant positive correlation with BIM adoption (P < 0.05). In contrast, the relative advantage positively influences BIM adoption only in countries at maturity levels 3 and 4, with the most pronounced effect observed at maturity level 3 (P < 0.05). Conversely, this impact is not significant in countries classified at lower maturity levels 1 and 2 (P > 0.05), suggesting that the role of top management support may not be as evident in the early stages of BIM adoption.
Component II (Champions and Drivers) reveals that in countries with a maturity level of 3, PEU (BA: r = 0.935), top management support (BA: r = 0.696, PEU: r = 0.674, PU: r = 0.729), organizational culture (PEU: r = 0.952, PU: r = 0.907), and organizational readiness (BA: r = 0.641) exert a stronger influence on BIM adoption. The differences between these groups are statistically significant (P < 0.05). However, in countries with a maturity level of 2, these effects do not reach statistical significance (P > 0.05).
Component III (Regulatory Framework) indicates that in countries with a maturity level of 1, top management support (PEU: r = 0.666, PU: r = 0.734), organizational culture (BA: r = 0.614, PEU: r = 0.952), and organizational readiness (BA: r = 0.695) show a stronger positive correlation with BIM adoption. This highlights that Organizational factors play a crucial role in driving BIM adoption within less developed regulatory frameworks. Meanwhile, in countries with a maturity level of 2, a stronger positive correlation is observed between relative advantage (r = 0.422) and the mediating pathway of PEU, with differences between groups being statistically significant (P < 0.05).
Component V (Learning and Education) illustrates that in countries at maturity level 2, relative advantage (BA: r = 0.422) and top management support (PEU: r = 0.666, PU: r = 0.734) have a stronger positive correlation with BIM adoption. Conversely, in countries at maturity level 1, organizational culture (PEU: r = 0.952; PU: r = 0.907) and organizational readiness (BA: r = 0.659) exhibit a stronger positive correlation. The difference between these groups is significant (P < 0.05).
Component VI (Measurements and Benchmarks) indicates that in countries with the lowest maturity (level 1), PEU (BA: r = 0.719), top management support (PEU: r = 0.64, PU: r = 0.729), organizational culture (PEU: r = 0.952, PU: r = 0.907), and organizational readiness (BA: r = 0.695) are more strongly positively correlated with BIM adoption. The difference between these groups is significant (P < 0.05).
Component VII (Standardised parts and deliverables). The results indicate that Component VII exclusively moderates the mediating pathway between top management support and PEU. Countries with a maturity level of 2 (r = 0.463) demonstrate a stronger positive correlation between top management support and PEU compared to countries at maturity level 3 (r = 0.095). The difference between these groups is significant (P < 0.05).
Component VIII (Technology infrastructure). In countries with a maturity level of 1, the positive influence of organizational culture (PEU: r = 0.952, PU: r = 0.907) and organizational readiness (BA: r = 0.695) on BIM adoption is stronger. Meanwhile, in the direct pathway of top management support for BIM adoption, countries with a maturity level of 2 (r = 0.724) exhibit a higher promotional effect. However, in the mediation path adopted by top management support for BIM, countries with a maturity of 1 (PEU: r = 0.674, PU: r = 0.729) have a higher promoting effect. The difference between these groups is significant (P < 0.05).
Despite these findings, we did not observe a significant moderating effect of the national BIM maturity model (comprising 8 components) on the relationship between compatibility, complexity, coercive pressures, and mimetic pressures related to BIM adoption. The statistical analysis did not indicate any significant differences in the Q statistic between groups (P > 0.05). For further details, please refer to Appendix A4.
4.5.2 Contextual factors
This meta-analysis also investigates how three contextual moderators—individual job level, organizational type, and time span—moderate both the direct and indirect pathways through which external antecedents influence BIM adoption.
First, job promotion significantly enhances compatibility (PU: r = 0.489) and top management support (PEU: r = 0.822, PU: r = 0.811), as well as coercive pressures (BA: r = 0.278) and PEU (BA: r = 0.828) regarding BIM. The differences between groups are significant (P < 0.05), suggesting that the benefits and compatibility of BIM are more pronounced and utilized by individuals in higher management positions.
Second, the type of organization significantly moderates the effect of BIM adoption. Specifically, in response to coercive pressure, owners (BA: r = 0.560, PEU: r = 0.415) exhibit a stronger influence on BIM adoption compared to construction organizations (BA: r = 0.303, PEU: r = 0.158), while engineering consulting service organizations demonstrate a negative influencing effect (BA: r = −0.144). In regards to the intermediary pathway supported by top management for BIM, the promotion effect within construction organizations (PEU: r = 0.321, PU: r = 0.352) is lower than the overall level observed in the AEC industry (PEU: r = 0.846, PU: r = 0.740). The difference between these groups is significant (P < 0.05).
Third, our findings indicate that, over time and especially after 2016, the promotional effects of BIM adoption related to relative advantage (BA: r = 0.499) and top management support (PEU: r = 0.792, PU: r = 0.724) have gradually diminished. Moreover, the differences among these groups are statistically significant (P < 0.05).
However, we did not observe significant moderating effects from the three background factors of job level, organization type, and time span on the relationship between BIM adoption and complexity, organizational readiness, organizational culture, mimetic pressures, and PU. Additionally, no statistically significant differences in the Q statistic were found between these groups (P > 0.05). For further details, please refer to the Supplementary Material A5.
5 Discussion
5.1 Theoretical implications
This study is the first to develop and validate a comprehensive, theory-driven BIM adoption meta-analysis framework based on the TOE framework. It aims to identify which factors within the TOE framework are more effective in promoting BIM adoption, through which pathways, and in what contexts. Given the global AEC industry’s urgent shift toward intelligent systems, advancing BIM adoption is crucial. Based on this, researchers have examined factors influencing BIM adoption through external perspectives such as TOE, INT, and DIT (
Shirowzhan et al., 2020;
Cao et al., 2016; Saka et al., 2024), as well as psychological perspectives like TAM (
Aladağ et al., 2023). Although these theories help elucidate the motivations behind BIM adoption, they remain fragmented and lack a unified framework, leading to inconsistent and contradictory research findings (
Wang et al., 2020b;
Hong et al., 2019). Therefore, to examine the nature of “promoting” and “hindering” BIM adoption, a systematic quantitative analysis of the relevant literature has become necessary. In this context, meta-analysis is regarded as an effective way to achieve this goal. In this study, we respond to Oesterreich and Teuteberg’s (
2019) call for a deeper exploration of the interaction between these factors, focusing on how external and internal factors jointly influence BIM adoption.
First, the integrated theoretical framework developed here refines the external factors of technology, organization, and environment by incorporating DIT and INT into the TOE framework. When analyzing how external antecedents influence BIM adoption, we also incorporate TAM, using PU and PEU as internal mediating factors. By establishing a clear causal chain, the framework outlines the pathway of “external variables—internal psychology—behavioral decision making,” providing a more comprehensive understanding of BIM adoption. Our results indicate that external factors influence BIM adoption more effectively through the mediation of internal psychological factors. This highlights the important role of intrinsic psychological factors in explaining how external factors affect BIM adoption. Therefore, we encourage future research to thoroughly consider the mediating role of intrinsic psychological factors, such as PU and PEU, to deepen the understanding of the relationship between BIM adoption and its antecedents.
Second, this meta-analysis quantitatively assesses which factors are most effective in promoting BIM adoption. In the technical dimension, our study finds that compatibility has a more significant effect on BIM adoption than comparative advantage. This reshapes our understanding of the technical drivers of BIM. Ensuring compatibility between technological interfaces and organizational needs is essential when adopting BIM as an inter-organizational information system (
Shirowzhan et al., 2020). However, unlike previous research (
Kim et al., 2016;
Wang et al., 2021), our results show that the mediating effect of complexity on PU and PEU is not significant. This suggests that BIM’s complexity may not directly affect stakeholders’ perceptions of its ease of use and usefulness. A possible explanation is that BIM’s complexity primarily stems from the lack of uniform standards in the AEC industry. However, this challenge is not inherent to the BIM system itself but is attributed to stakeholders’ insufficient proficiency and limited awareness of BIM (
Koutamanis, 2020).
Third, our findings offer new insights into the organizational dimension of BIM adoption. Contrary to previous studies (
Wang et al., 2020b;
Cao et al., 2022), our meta-analysis shows that organizational culture is more likely to promote BIM adoption than top management support and organizational readiness. In the AEC industry, long-established practices and resistance to new technologies make change difficult (
Zhang et al., 2020). In this context, an organizational culture with shared values and goals becomes critical, as it can help overcome barriers to learning and adapting to BIM, thus promoting its adoption and implementation (
Munianday et al., 2022).
Fourth, in the environmental dimension, our research shows that mimetic pressure is more effective in promoting BIM adoption than coercive pressure. This aligns with previous findings (
Cao et al., 2014;
Wang et al., 2020a). When BIM is applied to engineering projects, organizations often face coordination challenges and changes, which increase the difficulty of BIM adoption (
Eastman et al., 2011). To mitigate these uncertainties, stakeholders often imitate the practices of peer projects to reduce the risks associated with BIM adoption (
Saka et al., 2024).
Fifth, our findings provide a more comprehensive and updated understanding of national BIM maturity and contextual factors in the varied results of BIM adoption. So far, few studies have explored how BIM maturity impacts adoption across multiple countries, limiting the understanding of national BIM ecosystems. To address this limitation, this meta-analysis employs an eight-dimensional national BIM maturity model to analyze how it moderates the relationship between external factors and BIM adoption. Our analysis shows that organizational culture significantly drives BIM adoption across different maturity stages. For example, in Components III (Regulatory framework) and V (Learning and education), the positive relationship between organizational culture and BIM adoption is stronger in countries with lower maturity. Even in a high-maturity context, such as Component II (Champions and drivers), organizational culture remains a key driver. This implies that, whether formulating relevant policies, learning to adopt BIM, or establishing industry leadership, it is essential to prioritize cultivating and maintaining a positive organizational culture. However, Component VII does not show a moderating effect of organizational culture, suggesting that in standardized environments, the influence of culture may be overshadowed by top management support.
Lastly, beyond national BIM maturity, this meta-analysis is the first to systematically examine contextual factors such as job level, organization type, and time span in shaping external factors on BIM adoption. Once again, our findings reveal an interesting phenomenon: as job positions advance, managers’ perceptions of BIM compatibility, top management support, and external policy pressure increase. This enhanced perception improves the PU and PEU of BIM, strengthening the role of PU in BIM adoption, suggesting that managers may play a pivotal role in integrating technology, organization, and environment. For organization type, our results show that owners perceive coercive pressure more strongly than construction organizations. As primary decision-makers, owners seek to maximize economic benefits and are sensitive to external pressures like policies (
Sun et al., 2023). Additionally, over time, the relative advantage of BIM and top management support has declined in its influence on adoption. A potential explanation is that early BIM adoption focused on data sharing and project delivery, making comparative advantage and management support more significant. However, as BIM has become more widespread, its initial advantages may diminish due to habitual use.
In summary, this series of findings not only expands our understanding of the factors influencing BIM adoption but also offers new directions for future research. It underscores the importance of promoting BIM adoption more effectively in a dynamically evolving technological and managerial environment.
5.2 Practical implications
This research has important practical implications for professionals in the field. Our results show that compatibility, organizational culture, and mimetic pressure are key drivers of more effective BIM adoption in the technical, organizational, and environmental dimensions, respectively.
When implementing BIM in the AEC industry, managers should focus on ensuring compatibility to facilitate smooth integration and data exchange between various systems and tools (
Li et al., 2021). This approach helps optimize resource allocation and project processes, ensuring seamless information transfer and consistency throughout the project lifecycle. Additionally, it is essential for managers to cultivate a strong organizational culture. By fostering a motivating and supportive culture, they can encourage employees to adopt a positive attitude toward new technologies. This strategy can provide a competitive edge by accelerating the acquisition of BIM skills. Lastly, managers should enhance learning from peer organizations that use advanced BIM technologies. As BIM evolves rapidly with the rise of artificial intelligence, staying informed about the latest advancements adopted by peers is crucial for reducing risk and promoting BIM adoption.
Furthermore, our study highlights the importance of national BIM maturity and contextual factors as critical boundary conditions influencing BIM adoption. Although managers may not be able to alter their country’s BIM ecosystem, they can leverage international experiences to improve their organization’s culture and strategy. Organizations should also focus on engaging and training management teams to lead BIM adoption and cultural transformation effectively. Strong leadership is essential for integrating technology and management, which can optimize BIM adoption and promote innovation in business processes. Additionally, BIM policy development and implementation should consider the specific needs and responses of different types of organizations. For instance, owners, due to their financial interests, are particularly sensitive to mandatory policies. Finally, as BIM moves from initial benefits to more complex applications, it is vital for organizations to continually foster technological innovation and education.
5.3 Limitations and future research
Although this meta-analysis has provided valuable insights, it has certain limitations that warrant further investigation in future research. First, as with other meta-analyses, there is a potential risk of conflating different variables when combining them into higher-order structures (
Cortina, 2003;
Oh, 2020). To cautiously address this issue, we ensured that all variables in the model were clearly defined to minimize measurement errors (
Liao et al., 2024).
Second, this meta-analysis synthesizes findings from studies with diverse theoretical backgrounds and measurement methods. Given the diversity of the factors studied and the limited sample size, it was difficult to precisely quantify the differences in effects among the factors. Thus, future research could focus on conducting detailed meta-analyses of specific factors and explore longitudinal studies to increase the depth and applicability of the findings.
Third, although this meta-analysis examines several key moderators, the limited sample size of empirical studies restricts the range of moderators that can be included. Future research could further investigate the moderating role of some compelling factors such as project size, organization size, and industry context. Additionally, the influence of external factors on intrinsic psychological factors may differ across contexts. Future research could explore the moderating effects of individual characteristics and organizational resources. Moreover, using emerging technologies like machine learning and artificial intelligence to process larger, more complex data sets could help uncover hidden patterns and dynamic relationships in BIM adoption.
Fourth, while this meta-analysis quantifies the relationship between BIM adoption and its antecedents, and explores how national BIM maturity and background factors moderate theoretical pathways in the integration framework, it lacks specific case studies illustrating the practical application of the BIM integration framework. We encourage future research to incorporate specific engineering examples to thoroughly investigate the dynamic process of BIM adoption, including its psychological and external mechanisms.
6 Conclusions
The framework of this meta-analysis highlights how various antecedents are affecting the adoption of BIM. We reviewed the literature about BIM adoption from 2012 to 2023 and synthesized 62 empirical studies with 11,228 subjects in this meta-analysis. This meta-analysis has represented the complete journey of BIM adoption by applying the TOE framework along with other supportive theories like DIT, INT, and TAM to identify how various factors influence BIM adoption. The findings show that compatibility is the most influential factor that promotes BIM adoption in the technical dimension, organizational culture is of the essence in the organizational dimension, and mimetic pressure is the most important driver in the environmental dimension. We also confirm perceived usefulness and perceived ease of use acting to mediate the relationship between the external factors and BIM adoption.
This study further investigates the moderating role of national BIM maturity and contextual factors in the relationship between external factors and BIM adoption. The findings support that national BIM maturity (i.e., the eight components) and contextual factors (job level, organization type, and time span) modulate some of the theoretical pathways in the BIM adoption framework. These findings have significant theoretical and practical implications and provide new avenues for future research.
PDF
(3547KB)
