Trust-Enhanced Socialization, Externalization, Combination, and Internalization Model for Quantitative Requirements Elicitation: Longitudinal Validation in Higher Education Projects

Trust dysfunction in requirements elicitation systematically undermines software project success. However, current Requirements Engineering (RE) methodologies lack quantifiable frameworks to measure or manage trust relationships.  The Requirements Elicitation Process (REP) often produces incorrect, incomplete, and inconsistent requirements due to socio-emotional complexity [1, 2]. However, existing approaches treat trust as unmeasurable. This is problematic because projects with dysfunctional trust experience 40-60% higher defect rates and two to three times longer development cycles [3]. This void manifests across three deficiencies. Gap 1: Absence of Quantifiable Trust Measurement. No mathematically rigorous methods exist to assess trust levels. Burnay and Snoeck [3] provided conceptual definitions but lacked operational instruments. This absence creates a "measurement vacuum" that remains invisible to project management systems. Gap 2: Missing Integration Between Trust Theory and Knowledge Transformation. The SECI model [4] and trust theory [5] remain disconnected within REP contexts [6, 7]. This disconnection prevents practitioners from understanding several critical aspects [8]: dimension-phase matching, configuration effects, temporal dynamics, and bidirectional causality. Gap 3: Insufficient Empirical Evidence: Current research offers anecdotal observations [3] but lacks large-scale validation, with studies suffering from small samples (n < 10) [9, 10], cross-sectional designs, subjective metrics, and no validated predictive thresholds. This research addresses these gaps through three primary contributions. Contribution 1: Mathematical Trust Quantification Framework. We integrated the three-dimensional trust model of Mayer et al. [5] with requirements quality assessment across Socialization, Externalization, Combination, and Internalization (SECI) phases. This integration enables objective measurement on validated 100-point scales (α = 0.93), variance-based early warning indicators, and predictive modeling with empirically validated intervention thresholds. Contribution 2: Phase-Specific Trust-Knowledge Integration Theory: We establish systematic relationships between trust dimensions and SECI transformation stages, revealing how trust configurations facilitate or impede knowledge conversion, which dimensions dominate each REP phase, and how trust accumulation creates compound quality effects. Contribution 3: Large-Scale Empirical Validation: We provide comprehensive statistical evidence through longitudinal analysis of pedagogical information systems in Algerian higher education, featuring high collaborative complexity, knowledge-intensive tacit-to-explicit conversion, and trust-sensitive academic hierarchies, enabling robust validation with empirically-derived predictive thresholds for quality forecasting and intervention protocols.

The paper is organized as follows. Sections 2 and 3 lay the theoretical groundwork by establishing the REP domain context through trust theory, collaborative work, and knowledge transformation. Section 4 surveys the related literature on trust-enhanced requirements elicitation. Sections 5 and 6 present the trust-enhanced SECI model and its associated trust calculation methodology. Section 7 empirically validates the proposed approach through a MOOC case study, while Sections 8 and 9 address the key findings and directions for future research.

In this section, we present three domains of interest for our research work.

3.1 Trust

The trust could be of great significance for the RE process because it might affect the trust among stakeholders and analysts. Requirements engineers or analysts must be aware of the nature of various types of trust relationships and trust assumptions that exist during the specific activities of this process.

3.1.1 Trust in theory

Trust can be interpersonal, organizational, or institutional. Indeed, it can be horizontal (between colleagues), vertical (between managers and employees or between employees and managers), or institutional (between employees and organizations) [22].

Trust has been studied in many disciplines, including sociology [23], psychology [24], economics [25], and computer science [26] from different angles.

In general, trust is a measure of assurance that one or more entities will behave in the expected way [27]. According to [28], it is possible to classify trust perspectives into three categories: (i) that of personality theorists, who conceptualize trust as a belief, expectation or feeling deeply rooted in the personality; (ii) that of sociologists, focused on the understanding of the social bond; (iii) that of psychosociologists, who define trust in terms of the hope and consent of a third party who engages in a transaction.

The work of Marsh and Dibben [29] is considered to be the first to have introduced the notion of trust in computer systems. Today, many well-known computer systems and applications are based on this notion.

Firstly, there is agreement on the fact that trust is a dyadic relationship; it involves two people, referred to as Trustor (the one who trusts) and Trustee (the one who is trusted) [22].

Secondly, it is common to consider trust as a subjective evaluation made by the trustor over the trustee [30]. In the study [31], for example, trust is defined as "the perceived credibility and benevolence of a target of trust». The definition emphasizes the perception aspect of trust: trust is not the result of an objective process, but is the result of some sensations from the trustor about the trustee.

Thirdly, trust systematically follows from an intention. In the study [32], trust is defined as "the willingness to rely on an exchange party in whom one has confidence", thereby emphasizing the necessity of a voluntary trustor.

Fourthly, trust implies vulnerability [3, 33]. In the study [3], Burnay and Snoeck define trust as "the willingness of a party to be vulnerable to the actions of another party". The definition clearly states that a trustor is vulnerable to some risk when trusting the trustee.

Figure 2. Integrative model of trust [5]

To measure, understand, and explain trust, it is important to identify its dimensions. Research has identified several key elements of trust. It arises from certain factors of reliability that the trustor perceives in the trustee. These factors vary according to context, and are assessed by the trustor so that he or she can decide whether the trustee is trustworthy and whether the outcome of an interaction is fruitful [34]. These factors represent the dimensions of trust. Mayer et al. [5] studies have shown that trust has three main dimensions (Figure 2):

1. Benevolence: The trustor is important to the trustee, so he or she must act properly so as not to hurt him or her.
2. Competence: the trustee is capable of doing what the trustor wants (and this may be the reason why the trustor asks the trustee for help).
3. Integrity: the trustee is honest and acts in accordance with the trustee’s wishes, without malice.

3.1.2 Trust in engineering

The concept of trust has received considerable attention in engineering research communities, as trust is perceived as the basis for decision-making in many contexts and the motivation for maintaining long-term relationships based on cooperation and collaboration [8, 35]. In some environments, trust relationships are understood only as interpersonal relationships between human beings. However, in some computer environments, these trust relationships can be extended to concern different kinds of entities. Indeed, the trustee can be: any entity, such as a person, a group of persons, or an organization; a system agent representing a person or organization; a piece of system; a device; a network; a data resource like a document or a database; the development process; the business process where the system is used; or the whole system.

The trustor has to be an entity capable to assess the trustworthiness of another entity, and also capable of making decisions based on that. It is rather common that people do not trust systems that they are not familiar with, but, after experiencing their functions, trust may evolve. In this case, trust evolves in the person (trustor) even though the system has not been changed.

3.2 Collaboration work

Collaboration is a complex phenomenon whose definitions remain very vague and variable. The literature provides several definitions of collaboration, which generally translate into a joint effort toward a common goal [36]. Given the many definitions that exist, it was important to choose one that we felt was the most generic. We therefore chose the definition, which states that interprofessional collaboration is: ’made up of a set of relationships and interactions that may or may not enable professionals to pool and share their knowledge, expertise, experience and skills in order to use them concomitantly in the service of clients and for the greater good of the clients’ [37].

It is also seen by the study [38] as a process in which stakeholders with different perspectives on a problem, can constructively explore the differences and can search for solutions that go beyond their own limited visions. In a collaborative context, all participants co-construct together, even if the task can be divided into several other subtasks. In collaboration, all participants intervene in each step so that it is not possible to distinguish individual work in the final result [39]. Collaboration is often done synchronously, even if it is possible in an asynchronous way [40] to create synergy and benefit most from interaction. In fact, the quantitative prospective analysis conducted by Quinteros provides stakeholders with a strategic tool that enables them to take proactive measures to reduce employee turnover and maintain the organization’s vitality [8].

Through the above definitions, we can notice that collaboration is a complex task because it is more difficult to co-construct than to emit or receive information (communication) and more difficult than to assemble information by gathering and coordinating the work of individuals (cooperation and coordination) [41].

3.2.1 Collaboration engineering

Collaboration engineering is a structured approach to facilitating collaborative working. It involves three key roles: facilitator, practitioner, and collaboration engineer [42]:

1. A facilitator designs and conducts collaboration processes to support specific groups in achieving their goals.
2. A collaboration engineer designs and documents collaborative work practices, [36] ensuring they are transferable and reusable for practitioners to use independently [43, 44].
3. Practitioners learn to execute work practices based on the engineer’s design, reducing the skills required.

Processes are organized into six basic patterns: "Generate," "Clarify," "Reduce," "Organize," "Evaluate," and "Build Consensus." These patterns help create predictable group behavior and achieve common goals.

3.2.2 Collaboration in engineering

Collaboration in engineering involves bringing together experts from various fields to create, develop, and execute complex solutions, ensuring innovation, quality, and efficiency [45]. By implementing structured processes, adopting good tools, and promoting effective communication, teams can maximize efficiency. However, collaboration processes depend on engineering processes [41].

The Electronic Industries Alliance standard 632 (EIA-632) defines the processes required to design, develop, implement, and support complex systems throughout their lifecycle [46]. Konaté et al. [41], have used the collaborative engineering approach to define the collaborative processes within the EIA-632 standard. This led them to focus on collaborative engineering tasks.

3.3 Knowledge transformation

Knowledge is dynamic and context-specific; it is created in social interactions among individuals and organizations and depends on a particular time and space [47]. A portion of knowledge is usually recorded, but the rest may still remain in the individuals’ mind.

Knowledge transformation is the transformation of raw data into usable, understandable, and exploitable knowledge, crucial for learning, innovation, and decision-making in fields like engineering, education, and management. Given its link to organizational performance, it is essential to continue to identify the factors and contexts that facilitate employees’ knowledge creation and sharing behaviour [48].

RE is a knowledge-intensive activity [49], and a variety of tools and techniques are needed to capture and process this knowledge.

3.3.1 Knowledge management

Knowledge management (KM) refers to the management of individual, structural, and organizational knowledge [50]. Knowledge is intrinsically embedded in human cognition, and knowledge management occurs in frames of social context [51]. Thus, the consideration of the human and social factors is a necessary and crucial element of the knowledge management process. Indeed, within organizational contexts, stakeholder behavioral data constitutes a significant knowledge resource, one that increasingly shapes and informs the foundations of strategic decision-making [19]. One of the greatest challenges of knowledge organizations is to extract tacit knowledge possessed by their knowledge employees, so as to apply the right knowledge at the right place when needed and to encourage innovation. One of the main obstacles to these efforts is the lack of trust between managers and employees, and between employees themselves [52]. To effectively achieve the objectives of REP, tacit knowledge must be converted into formal knowledge that is easily accessible and can be widely used to develop quality requirements. Knowledge management processes can be used as a complement to support analyst engineers.

3.3.2 Knowledge elicitation

System engineering [53, 54] has a lot in common with knowledge management in the knowledge elicitation phase. Knowledge elicitation involves acquiring and transferring the knowledge of individuals to an abstract and effective representation to organize it, model it, and finally express it. REP involves the acquisition and transfer of knowledge from individuals, teams, or systems in order to organize, model, and formalize it in an understandable format and finally use it in various contexts.

3.3.3 Socialization, Externalization, Combination, and Internalization model

This model, developed by Ikujiro Nonaka and Hirotaka Takeuchi [4], is a framework for understanding the process of knowledge creation and transfer within organizations. The model is one of the most influential theories of knowledge creation [48] and has therefore been the one most adopted by researchers studying the relationship between knowledge creation and innovation [55]. Nonaka et al. [4] distinguishes between implicit (tacit) and explicit knowledge. Explicit knowledge is characterized by being formal and systematic knowledge that can be expressed without ambiguity through writing, schematics, databases, etc. It is stored in textbooks, system products, and documents; implicit knowledge is highly informal, personal, unverbalized, intuitive, and derived from experience. It is stored in the minds of people in the form of memory, skills, experience, education, imagination, and creativity. The model consists of four modes of knowledge conversion: SECI. It does involve a process of knowledge transformation. Stages of the SECI model (1) Socialization (Tacit to Tacit): individuals share tacit knowledge (personal and experiential) with others through direct interaction and shared experiences. This could occur through apprenticeships, mentoring, or other forms of personal interactions. (2) Externalization (Tacit to Explicit): the articulation of tacit knowledge into explicit forms. This process involves expressing personal insights and experiences in a form that can be communicated and shared with others. This often takes the form of creating documents, diagrams, or other tangible representations.

Combination (Explicit to Explicit): explicit knowledge (codified and documented) is combined and organized in a systematic way. This can include the integration of different explicit knowledge sources, such as databases, manuals, or other structured information repositories.

Internalization (Explicit to Tacit): the process of converting explicit knowledge back into tacit knowledge. Individuals acquire explicit knowledge from external sources and internalize it by applying it in their own context, thereby making it part of their personal experience and tacit knowledge.

The SECI model is often represented as a continuous cycle, emphasizing the dynamic and iterative nature of knowledge creation and transfer within a team [4]. While it is not a direct elicitation process for gathering requirements, it highlights the importance of understanding and managing the flow of knowledge within a REP.

As mentioned above, our research focuses on REP, which is a collaborative activity between stakeholders and analysts, whose success relies on the ability of both groups to work together, combining their different sets of knowledge and taking into account the mutual trust. In fact, we are focusing on three important dimensions: Collaborative Knowledge and Trust.

Firstly, REP is a collaborative activity that can be designed and defined using collaborative engineering. REP without effective knowledge conversion is determined to fail. Indeed, in the processes of knowledge internalization, because of a crisis of trust in knowledge communication, developers did not effectively communicate with users.

Secondly, one of the most cited factors affecting knowledge sharing is trust [62]. Trust is an important factor in ensuring the organizational climate and, hence, the success of REP.

To our knowledge, there is no previous work that investigates the influence of trust on the behavior of analysts and stakeholders in REP activities in individual and collaborative contexts using the SECI model. There is no coherent conceptual framework that integrates trust in REP activities. The purpose of this paper is to examine the impact of trust as a constraint for the collaboration process on knowledge creation in REP using the SECI model introduced by Nonaka and Takeuchi [63].

5.1 Socialization, Externalization, Combination, and Internalization model in Requirements Elicitation Process

REP can be viewed through the SECI model a knowledge management framework that explains how tacit and explicit knowledge are created, converted, and utilized within an organization [64]. The SECI model comprises four stages: SECI, each playing a distinct role in the elicitation process. Integrating the SECI model into REP provides a structured approach to understanding and managing the knowledge involved in the process [16]. The application of the SECI model in REP is further enhanced by considering distinct phases: Pre-elicitation, Elicitation, and Post-elicitation. Each phase corresponds to one or more specific stages of the SECI model, creating a cohesive framework for managing the flow of knowledge during RE.

5.1.1 Pre-elicitation

This phase, which is grounded in internalization, focuses on equipping the elicitation team with the necessary knowledge and context. This involves a deep dive into existing documentation, such as project charters, market research reports, and competitor analyses, to gain a comprehensive understanding of the project’s goals, target audience, and competitive landscape [65]. The team members internalize this information, transforming it into their own tacit knowledge, which forms the basis for subsequent elicitation activities. Figure 3 illustrates the systematic conversion of explicit knowledge sources into tacit understanding through structured transformation activities.

Figure 3. Pre-elicitation phase

5.1.2 Elicitation

As illustrated in Figure 4, this phase, which is grounded in socialization. emphasizes the importance of direct interaction and shared experiences among stakeholders. It involves fostering an environment where stakeholders can freely exchange ideas, perspectives, and tacit knowledge related to the system’s needs and functionalities [66]. It often involves techniques such as brainstorming sessions, interviews, and ethnographic studies to capture the implicit and unarticulated requirements that reside within the stakeholders’ minds [67]. These interactive sessions facilitate a deeper understanding of the problem domain and user needs, enabling the elicitation of requirements that might otherwise be overlooked [15].

Figure 4. Elicitation phase

5.1.3 Post-elicitation

This phase, which is grounded on externalization and combination, embodies the crucial process of articulating tacit knowledge into explicit forms that can be readily comprehended, shared, and leveraged across the organization [16]. This transformation is pivotal for ensuring that the valuable insights, experiences, and understandings held by individual stakeholders are not confined to their personal cognitive domains but are instead converted into tangible assets that can be readily disseminated and utilized by the wider development team and relevant stakeholders. As indicated in Figure 5, the combination represents the culmination of the knowledge creation process within the SECI model, involving the integration and synthesis of newly externalized knowledge with existing explicit knowledge to form more complex, comprehensive, and systematic bodies of knowledge [68].

Figure 5. Post-elicitation phase

This holistic approach ensures that knowledge is systematically created, shared, and preserved throughout REP, leading to more effective and comprehensive requirements understanding while building organizational knowledge assets for future projects. As the Figure 6 demonstrates, this framework integrates Nonaka's SECI model with the Requirements Elicitation Process, illustrating knowledge transformation across three phases: Pre-Elicitation (Internalization) converts explicit domain knowledge into analyst tacit competence; Elicitation (Socialization) enables tacit-to-tacit exchange through collaborative activities in Ba spaces; Post-Elicitation (Externalization and Combination) transforms shared understanding into formal requirements documentation, with cyclical flows illustrating continuous individual-to-collective knowledge conversion.

Figure 6. Knowledge conversion dynamics in requirements elicitation

5.2 Integrated Framework Logic: Phase-wise trust- Socialization, Externalization, Combination, and Internalization-quality pathways

This research demonstrates that knowledge conversion, team performance, and requirements quality are significantly positively correlated, with trust as a crucial facilitator operating at multiple levels to influence knowledge processes and team outcomes. We position interpersonal trust as both a precondition and outcome of effective collaboration, recognizing that horizontal and vertical trust moderate collaborative processes since trust between team members is necessary for creating and disseminating tacit knowledge [47].

Our theoretical integration operates through three interconnected phase-specific mechanisms. During pre-elicitation (Internalization), analysts convert explicit domain knowledge into tacit understanding while competence trust forms through demonstrated mastery, determining stakeholders' willingness to share complete information. During elicitation (Socialization), tacit-to-tacit knowledge sharing occurs through direct interaction as benevolence trust (perceived care) and integrity trust (consistent behavior) create psychological safety, enabling honest, unambiguous communication. During post-elicitation (Externalization-Combination), cumulative trust from previous phases enables collaborative validation as stakeholders actively participate in converting shared tacit knowledge into explicit, refined requirement artifacts. Three critical interaction points govern this process: trust accumulates across phases with compounding effects, requirements quality outcomes recursively reinforce or diminish trust for subsequent iterations, and trust instability at any phase disrupts downstream knowledge conversion regardless of mean trust levels.

5.2.1 Trust as a catalyst for knowledge creation

Given that trust among team members serves as a prerequisite for generating and sharing tacit knowledge [47], we suppose that both horizontal and vertical trust dimensions moderate collaborative processes through several mechanisms:

1. Psychological Safety: Trust creates an environment where team members feel safe to share incomplete ideas, admit uncertainties, and challenge assumptions without fear of judgment.
2. Knowledge Vulnerability: Team members are more willing to expose their knowledge gaps and seek help when trust levels are high, leading to more effective knowledge transfer.
3. Cognitive Diversity: Trust enables teams to leverage diverse perspectives and expertise, enhancing the quality of requirements through comprehensive stakeholder input.

5.2.2 Trust as a foundation of knowledge transformation

REP ensures that the collected requirements achieve optimal levels of consistency, clarity, and completeness through systematic refinement and validation [69]. Figure 7 demonstrates that trust serves as the fundamental enabler that core enabling mechanism through which three quality dimensions are sustained across all SECI model phases during REP. Without trust, knowledge transformation becomes superficial, incomplete, or entirely blocked. Trust manifests differently across each phase while maintaining its critical role throughout the entire process.

Figure 7. Trust Foundation (with Competence, Integrity, and Benevolence) as the core element around which the entire requirements elicitation cycle revolves

Trust functions as both a prerequisite and outcome of effective knowledge transfer in RE, creating a virtuous cycle where trust enables better knowledge sharing, which in turn builds stronger trust relationships among stakeholders.

This approach effectively communicates that successful RE depends not just on following the SECI process, but on cultivating trust at every phase to ensure authentic, complete, reliable knowledge transfer.

Trust serves as the fundamental cornerstone of knowledge transformation processes within RE activities. The dynamic nature of elicitation demands that varying forms of trust be cultivated and deployed strategically across the three phases of REP (see Figure 6). Each phase presents unique challenges and opportunities that require specific trust configurations to optimize knowledge flow and stakeholder engagement. REP exhibits a cyclical nature, where trust operates as both an enabler and an outcome of continuous improvement initiatives. As trust levels increase through successful collaborative experiences, teams become more willing to engage in iterative refinement, constructive feedback, and adaptive learning. This creates a positive feedback loop where enhanced trust leads to better knowledge transformation, which in turn strengthens trust relationships and enables more sophisticated elicitation practices. We adopt the dimensions of trust mentioned in the work of Mayer et al. [5]. Indeed, the entire system is supported by three foundational trust dimensions that work synergistically to create an environment conducive to effective knowledge transformation.

1. Competence Trust (CT): Stakeholder confidence in the requirements engineer's technical expertise and domain knowledge.
2. Benevolence Trust (BT): Stakeholder perception of the requirements engineer's genuine concern for stakeholder interests.
3. Integrity Trust (IT): Stakeholder belief in the requirements engineer's honesty and adherence to ethical principles.

Together, these dimensions create a robust foundation that sustains trust throughout the complexity and uncertainty inherent in REP (see Figure 6).

5.3 Conceptual framework trust-enhanced Socialization, Externalization, Combination, and Internalization model in Requirements Elicitation Process

We propose a coherent conceptual framework that integrates trust in REP activities. This model (Figure 8) explains how trust relationships become crucial during requirements gathering, where stakeholders must be willing to share knowledge, admit uncertainties, and engage honestly with analysts.

Figure 8. Conceptual framework trust-enhanced Socialization, Externalization, Combination, and Internalization (SECI) model in Requirements Elicitation Proces (REP)

This model integrates three complementary theoretical perspectives into a unified framework. The SECI model provides temporal structure through the PHASE-ACTIVITY-TASK hierarchy (Internalization, Socialization, Externalization, Combination). Collaboration Engineering operationalizes stakeholder interactions via COLLABORATION_PROCESS entities with structured facilitation PATTERNS and TRUST_ACTOR participation. Trust theory functions as a moderating mechanism through TRUST constructs.

The model depicts causal pathways where trust influences: (i) stakeholders' willingness to externalize tacit knowledge during socialization, (ii) articulation quality during externalization, and (iii) knowledge integration reliability during combination—operationalized through "participate to define" relationships linking TRUST_ACTOR to NEED specifications that produce REQUIREMENT artifacts. This represents not a static construct list, but dynamic pathways where trust enhances collaborative knowledge conversion to impact requirements quality, providing the ontological foundation for measuring trust-knowledge-quality relationships across the project's empirical validation.

This research contributes to a broader initiative aimed at adapting cross-curricular and discovery modules for diverse learners to establish a comprehensive national e-learning framework. The initiative, spearheaded by Algeria’s Ministry of Higher Education and Scientific Research (MESRS), focuses on converting traditional cross-curricular modules into Massive Open Online Courses (MOOCs).

The systematic analysis conducted during this process culminates in the creation of a Requirements Specification (RS) document. This foundational document articulates the essential requirements, learning objectives, and technical specifications necessary for developing effective large-scale online courses that meet the project’s educational goals. The literature reveals that despite the growing popularity of MOOCs, the issue of their quality remains a central concern, partly due to the lack of consensus on the criteria for establishing this quality. Engineering Practices for Building Quality System (Coursera) | MOOC List. This suggests that the relationship between measures of trust stability and results in terms of requirements quality is still being defined. It is in this context that we present our case study.

7.1 Mathematical framework

The mathematical formalization presented here is therefore problem-driven, serving to explain empirically observed trust-quality dynamics rather than introducing abstract complexity. Each measurement index, statistical model, and analytical technique directly addresses specific research questions about how trust influences requirements quality through knowledge conversion processes. Primary contributions include: (1) rigorous trust measurement framework, (2) variance analysis methodologies, (3) predictive correlation modeling for quality assessment, and (4) empirically-validated intervention thresholds.

7.1.1 Core performance indices

The framework establishes two fundamental performance indices that serve as primary indicators of REP effectiveness.

The Trust Quality Index (TQI) provides a composite measure of trust across the three SECI phases, weighted according to their differential impact on requirements quality outcomes:

$\mathrm{TQI}=\frac{\sum_i\left(w_i \times T_i\right)}{n}$           (2)

The Requirements Quality Index (RQI) quantifies quality metrics through weighted aggregation:

$\mathrm{RQI}=\frac{\sum_i\left(w_j \times Q_j\right)}{m}$              (3)

where, wᵢ represents phase weights, Tᵢ denotes trust measurements, Qⱼ indicates quality metrics, and n, m represent dimensional counts. Decision thresholds have been empirically validated: TQI < 70 triggers interventions, while RQI < 75 requires revision protocols.

7.1.2 Variance analysis methodology

Trust measurement consistency was evaluated using standard statistical parameters including mean, variance, standard deviation, and coefficient of variation (CV). A color-coded diagnostic framework was established with three reliability categories: high consistency (CV < 10%), moderate variability (10% ≤ CV ≤ 20%), and high variability (CV > 20%) requiring intervention. This methodology enables systematic identification of measurement stability and guides quality control decisions in trust assessment protocols.

7.1.3 Inter-phase correlation analysis

The inter-phase correlation analysis employs Pearson correlation coefficients to quantify trust relationship evolution across the SECI phases:

$r_{x y}=\frac{\sum\left[\left(x_i-\bar{x}\right)\left(y_i-\bar{y}\right)\right]}{\sqrt{\left[\sum\left(x_i-x\right)^2 \sum\left(y_i-\bar{y}\right)^2\right]}}$            (4)

Correlation coefficients (r) between consecutive trust measurement phases (x and y variables) were classified according to standard statistical thresholds: strong correlations (r > 0.7) indicating effective collaborative momentum and predictable inter-phase trust transfer; moderate correlations (0.3 ≤ r ≤ 0.7) suggesting natural trust evolution requiring contextual interpretation; and weak correlations (r < 0.3) signaling potential disruptions warranting targeted interventions.

This classification system provides a systematic evaluation of trust relationship strength between paired measurement phases and informs intervention strategies. As shown in Table 1, the framework incorporates advanced predictive modeling through statistical thresholds that enable proactive process management. The supervision reduction threshold (r > 0.9) indicates sufficient trust stability to reduce oversight requirements. Intervention impact measurement employs Cohen's d > 0.8 to demonstrate strong intervention effectiveness.

Table 1. Statistical threshold classifications and decision protocols

7.1.4 Reliability assessment and validation

The framework employs multiple statistical validation techniques to ensure measurement reliability. Cronbach's alpha provides an internal consistency assessment:

$\alpha=\left(\frac{k}{k-1}\right) \times\left(1-\frac{\Sigma \sigma_i^2}{\sigma_{\text {Total}}^2}\right)$             (5)

where, k represents the number of items, σ²ᵢ indicates individual item variance, and σ²_total represents total scale variance. Values α > 0.9 confirm exceptional measurement reliability.

Statistical significance of elicitation effectiveness was assessed through rigorous hypothesis testing procedures. Independent samples t-tests were conducted using the formula:

$t=\frac{\left(\bar{x}_1-\bar{x}_2\right)}{\sqrt{\frac{s_1^2}{n_1}+\frac{s_2^2}{n_2}}}$                   (6)

To evaluate mean differences in trust levels between elicitation phases or stakeholder groups. Effect sizes were quantified using Cohen's d statistic:

$d=\frac{X_1-X_2}{S_{ {pooled}}}$                 (7)

To determine the practical significance of trust improvements throughout REP. Precision of trust measurement estimates was evaluated through 95% confidence intervals calculated as

$C I=\bar{x} \pm 1.96 \times \frac{\sigma}{\sqrt{n}}$                 (8)

This comprehensive analytical framework ensures both statistical rigor and practical interpretation of trust dynamics during stakeholder engagement and requirements gathering activities.

7.1.5 Automated pattern recognition

The framework incorporates intelligent pattern recognition capabilities that automatically identify trust dynamics and generate actionable insights. Trust consistency assessment algorithms identify dimensions demonstrating exceptional stability (CV < 10%) versus those requiring enhanced attention (CV > 20%). Phase transition analysis systems detect trust evolution patterns across SECI phases, identifying optimal intervention points.

7.1.6 Predictive modeling capabilities

The framework enables 2-3 phase-ahead quality prediction through sophisticated statistical modeling. The predictive quality score incorporates multiple statistical indicators:

$\mathrm{PQS}_{t+k}=\beta_0+\beta_1 X_{1 t}+\beta_2 X_{2 t}+\ldots+\beta_n X_{n t}+\varepsilon$            (9)

where, PQSₜ₊ₖ represents predicted quality score k phases ahead, β coefficients indicate model parameters, and ε represents the error term.

7.1.7 Structural Equation Modeling

Model Specification and Estimation.

The structural equation model was estimated using maximum likelihood estimation. The model specification follows the general SEM framework where the model-implied covariance matrix is defined as: covariance matrix is defined as:

$\Sigma(\theta)=\Lambda(\Phi) \Lambda^{\prime}+\Theta$                (10)

where, Λ represents the factor loading matrix, Φ denotes the latent variable covariance matrix, and Θ indicates the error/uniqueness covariance matrix.

Model Fit Assessment.

Model fit was evaluated using multiple fit indices. The chi-square test of model fit was calculated as:

$\chi^2=(N-1) \times\left[\operatorname{tr}\left(S \Sigma^{-1}\right)-\log \left|S \Sigma^{-1}\right|-p\right]$                 (11)

where, N represents sample size, S is the sample covariance matrix, Σ is the model-implied covariance matrix, and p denotes the number of observed variables. The analysis yielded χ² = 156.240 (df = 87, p = 0.001).

Measurement Model Results.

Standardized factor loadings were calculated using:

$\lambda_{i_j}=\gamma_i \times \sqrt{\frac{\varphi_j}{\vartheta_i+\theta_i}}$                    (12)

λᵢⱼ represents the standardized factor loading, which reflects the strength of the relationship between each trust indicator i and its corresponding latent trust factor j, such as competence, benevolence, or integrity. γᵢ denotes the unstandardized path coefficient, representing the raw regression weight from the latent trust factor j to the observed trust measurement variable i. φⱼ refers to the variance of the latent trust construct, capturing how much the trust dimensions vary across stakeholders during the requirements elicitation process. ϑᵢ captures the common or true variance of the observed trust indicator i, reflecting the portion of measurement that is attributable to the underlying trust construct. Finally, θᵢ represents the unique or measurement error variance of indicator i, where the sum of ϑᵢ and θᵢ constitutes the total observed variance of the trust indicator, serving as the standardizing denominator that ensures all factor loadings are expressed on a comparable scale across trust dimensions. 

All factor loadings exceeded 0.80, confirming strong relationships between latent constructs and their indicators. Trust dimensions showed robust loadings: competence (λ = 0.89), benevolence (λ = 0.84), and integrity (λ = 0.87). Requirements quality indicators demonstrated similar strength: completeness (λ = 0.85), consistency (λ = 0.88), and clarity (λ = 0.82). Structural Model Results, standardized path coefficients were estimated as:

$\beta_{i_{\mathrm{j}}}=\gamma_{i \mathrm{j}} \times \sqrt{\left(\varphi_{\mathrm{jj}} / \psi_{i i}\right) }$                    (13)

7.2 Methodology and data collection framework

7.2.1 Study overview

This research adopts a 24-month longitudinal design to empirically validate the proposed trust-enhanced SECI model in realistic requirements elicitation settings. The study tracks trust evolution and requirements quality outcomes across complete project lifecycles, enabling robust temporal and comparative analysis.

The empirical dataset comprises 500 stakeholders involved in 45 pedagogical information systems projects, conducted within higher-education and digital learning environments. These projects include learning management systems, MOOC platforms, academic information portals, and collaborative educational tools, all characterized by intensive stakeholder interaction during requirements elicitation.

A total of 1,500 temporal observations were collected at structured intervals aligned with the Pre-elicitation (Internalization), Elicitation (Socialization), and Post-elicitation (Externalization–Combination) phases of the SECI model. This design enables both within-subject longitudinal analysis (trust evolution over time) and between-project comparative analysis, significantly strengthening external validity.

Trust was operationalized using Mayer et al.’s tripartite dimensions—Competence, Benevolence, and Integrity—while requirements quality was assessed through Completeness, Consistency, and Clarity, consistent with prior requirements engineering literature.

7.2.2 Reliability, validity, and bias mitigation

To ensure measurement reliability, internal consistency was assessed using Cronbach’s alpha, with all trust and quality constructs exceeding the recommended threshold (α > 0.80). Construct validity was confirmed through confirmatory factor analysis and structural equation modeling, with all factor loadings above 0.80.

The longitudinal design reduces common-method bias by separating measurement points across project phases. Social desirability bias was mitigated through anonymized data collection and role-separated evaluations (stakeholders assessed analysts and vice versa). The large multi-project sample further minimizes idiosyncratic effects associated with single-team dynamics.

Unlike prior exploratory studies limited to small samples, the present design supports generalizable inference across projects, roles, and organizational contexts, enabling statistically meaningful validation of the proposed model.

7.3 Variance analysis

The variance analysis examines trust measurements across the three phases of RE and provides detailed statistical insights into trust dynamics, variance patterns, and their impact on requirements quality throughout the elicitation process.

7.3.1 Intra-phase variance analysis - trust as quality predictor

The intra-phase variance analysis reveals significant differences in trust and quality metrics across SECI knowledge management phases. The elicitation phase (socialization) exhibited the lowest trust levels (M = 76.17) and highest variability (SD = 6.85, Variance = 46.37), while the post-elicitation phase (externalization-combination) demonstrated optimal performance with the highest trust (M = 78.50) and reduced variance (21.90). Despite lower baseline trust, the elicitation phase showed the strongest trust-quality correlation (r = 0.898, p < 0.01), compared to pre-elicitation (r = 0.830) and post-elicitation (r = 0.788) phases. Quality metrics remained relatively stable across phases, with means ranging from 72.17 to 78.50. These findings suggest that while the socialization phase represents the most variable and challenging stage of knowledge management, it establishes the strongest relationship between trust and quality outcomes, with subsequent phases benefiting from this foundation through improved stability and performance. The most striking finding is the inverse relationship between trust variance and quality stability. During the Elicitation phase, when trust variance peaks at 46.97, quality variance increases simultaneously to 9.37, representing a 167% increase over baseline. Table 2 shows how trust variance during collaborative sessions directly impacts requirements quality consistency. When stakeholders disagree, both trust and quality ratings fluctuate together. This finding suggests that trust instability directly undermines stakeholders' ability to produce consistent and reliable requirements.

Table 2. Intra-phase variance analysis of trust and quality metrics

Table 3. Cross-dimensional trust-quality correlation matrix by phase

Trust Impact Evidence, (1) High trust correlation (r = 0.898) during the most volatile phase indicates that despite variance, trust remains the primary driver of quality perceptions, (2) The 50% weighting assigned to Post-elicitation reflects the critical period where trust stability (variance = 21.90) enables effective knowledge externalization into quality requirements, and (3) Quality scores mirror trust patterns: both dip during Elicitation (72.17 vs 78.50) and recover during Post-elicitation, demonstrating trust's predictive power.

7.3.2 Cross-dimensional trust-quality matrix: specific trust-quality linkages

The correlation matrix (Table 3) identifies specific mechanisms linking trust dimensions to quality outcomes in REP. The matrix (Table 3) demonstrates specific trust-quality pathways, like how analyst engineers 's perception of Stakeholder's integrity (reliability) directly shapes his expectations for logical consistency in requirements specifications. The Integrity-Consistency pathway demonstrates the strongest relationship (r = 0.85-0.91), indicating that stakeholder reliability perceptions directly influence requirements coherence. Three critical pathways emerged: Competence→Completeness (r = 0.76-0.85), where perceived capability drives specification thoroughness; Integrity→Consistency (r = 0.85-0.91), where reliability trust produces logical requirement structures; and Benevolence→Clarity (r = 0.68-0.78), where interpersonal care influences communication quality, though this represents the weakest association. Progressive correlation strengthening from pre-elicitation to post-elicitation phases (p < 0.01) demonstrates trust's cumulative impact on quality outcomes, suggesting that trust-building interventions compound over the REP.

7.3.3 Stakeholder performance evolution summary

The longitudinal analysis demonstrates trust's transformative impact on requirements quality relationships (Table 2). Both stakeholders achieve substantial trust improvements (+12.3% and +13.0%) accompanied by dramatic variance reductions (-45.1% and -41.8%) and correlation strengthening (+11.0% and +15.8%), while quality scores remain stable. The analysis illustrates analyst's complete transformation journey from unstable trust assessments to mature, predictable professional collaboration with stronger trust-quality alignment. This pattern proves that trust enhancement improves quality relationship consistency and predictability without requiring quality score inflation.

Transformation Impact Patterns, (1) Trust Development Success: Substantial trust improvements across both stakeholders demonstrate the SECI framework's effectiveness, (2) Stability Enhancement: Massive variance reductions prove trust-building creates more reliable quality assessment and production processes, and (3) Relationship Strengthening: Correlation improvements show trust development enhances the precision with which trust predicts quality outcomes

7.4 Structural Equation Modeling analysis

Structural equation modeling analysis examined trust-quality relationships in REP using the trust framework of Mayer et al. [5] and SECI knowledge creation processes. The model demonstrated excellent fit (CFI = 0.952, TLI = 0.941, RMSEA = 0.058, SRMR = 0.045). Factor loadings for trust dimensions (competence λ = 0.89, benevolence λ = 0.84, integrity λ = 0.87) and requirements quality indicators (completeness λ = 0.85, consistency λ = 0.88, clarity λ = 0.82) exceeded 0.80, confirming construct validity. Structural path analysis revealed trust as the strongest predictor of requirements quality (β = 0.72, p < 0.001), with SECI knowledge processes significantly influencing both trust formation (β = 0.68, p < 0.001) and quality outcomes (β = 0.45, p < 0.001).

The structural analysis revealed three significant relationships: Trust emerged as the strongest predictor of requirements quality (β = 0.72, p < 0.001), while SECI knowledge processes demonstrated substantial influence on both trust formation (β = 0.68, p < 0.001) and direct quality outcomes (β = 0.45, p < 0.001). These results provide empirical support for the theoretical model linking collaborative knowledge creation, stakeholder trust, and REP effectiveness. The measurement model demonstrated adequate reliability and validity. All constructs exceeded recommended thresholds for internal consistency, with factor loadings above 0.80 indicating strong convergent validity. The excellent fit indices confirm that the theoretical model adequately represents the observed relationships among trust dimensions, knowledge creation processes, and requirements quality in REP.