Behavioral Intention to Use (BI) Overview

• Behavioral Intention to Use (BI) is defined as an individual’s self-reported likelihood to adopt or continue using technology, integrating cognitive, affective, social, and contextual factors.
• Measurement of BI employs reflective Likert scales with reliability ensured by metrics like Cronbach’s α (>0.70) and composite reliability, validating key constructs across TAM, UTAUT, and TPB models.
• Structural models such as PLS-SEM incorporate predictors like attitude, performance expectancy, and social influence, with empirical studies confirming their critical roles in shaping BI.

Behavioral Intention to Use (BI) refers to the latent psychological construct capturing an individual’s self-reported likelihood or conscious plan to adopt, continue, or recommend a specific technology or system. In empirical research, BI is operationalized and analyzed as a function of multiple cognitive, affective, and contextual factors, serving as the principal proximal antecedent of actual usage in technology acceptance frameworks. The construct is central in structural equation modeling of acceptance, adoption, and diffusion processes across domains, including education, e-learning, AI systems, and immersive platforms.

1. Theoretical Foundations and Constructs

Behavioral Intention to Use is formally rooted in the Technology Acceptance Model (TAM), Unified Theory of Acceptance and Use of Technology (UTAUT/UTAUT2), Theory of Planned Behavior (TPB), and related frameworks. In these models, BI is positioned as the endogenous variable most immediately predictive of actual system use, with key antecedents partitioned as follows:

• Cognitive Predictors: Perceived Usefulness (PU), Perceived Ease of Use (PEOU)/Effort Expectancy (EE), Performance Expectancy, Self-Efficacy (SE), System Quality, Facilitating Conditions (FC), Trust, and Perceived Cost (Misirlis et al., 2023, Revythi et al., 2017, Thiel et al., 3 Apr 2026, Nugraha et al., 8 Dec 2025, Hizam et al., 2021).
• Affective Predictors: Attitude toward Use (ATT), Hedonic Motivation (HM), Satisfaction (SAT), Intrinsic/Extrinsic Warm-Glow, and Enjoyment (Misirlis et al., 2023, Saravanos et al., 2022, Pitts et al., 24 Feb 2026, Damar et al., 19 Oct 2025).
• Social Predictors: Subjective Norm (SN), Social Influence (SI), Peer/Faculty Endorsement (Misirlis et al., 2023, Revythi et al., 2017, Damar et al., 19 Oct 2025, Nguyen et al., 2023).
• Contextual/Normative Predictors: Moral Justification, Privacy Risk, Personal Norms, and Habit (Miranda et al., 20 Mar 2026, Zharova et al., 2022, Diao et al., 2024).

In extended or domain-specific models, additional constructs such as Relative Advantage, Compatibility, Complexity, Presence, and Technology Readiness (TR) are incorporated, especially in emerging contexts like metaverse or healthcare applications (Damar et al., 19 Oct 2025).

2. Measurement Models and Operationalization

BI is operationalized as a reflective latent variable measured by multiple Likert-type items (typically 2–5 per construct), rating statements such as “I intend to use…” or “I plan to continue using…” target technology. Common scale ranges are 1–5 or 1–7, with anchors from “strongly disagree” to “strongly agree.” Internal consistency and convergent validity are assessed using Cronbach’s α (>0.70), composite reliability (CR > 0.70), and average variance extracted (AVE > 0.50). Example BI items include:

• “I intend to use MetaEducation technologies in my coursework.” (Misirlis et al., 2023)
• “I plan to continue using ChatGPT even for major requirements.” (Miranda et al., 20 Mar 2026)
• “I would use an AI chatbot while I am learning.” (Pitts et al., 24 Feb 2026)
• “I will definitely use healthcare metaverse in my education.” (Damar et al., 19 Oct 2025)

Most models report psychometric properties such as standardized factor loadings (commonly λ > 0.80), with scales reaching CR > 0.90 and AVE > 0.70 in high-quality instruments (Nugraha et al., 8 Dec 2025, Miranda et al., 20 Mar 2026, Damar et al., 19 Oct 2025, Sibug et al., 30 Apr 2026).

3. Structural Modeling Approaches

Behavioral Intention is situated as an endogenous variable in path models, typically estimated via Partial Least Squares Structural Equation Modeling (PLS-SEM), covariance-based SEM, or, less frequently, via regularized regression for ordinal outcomes. The canonical TAM path specification is:

$$\text{BI} = \beta_1\text{ATT} + \beta_2\text{PU} + \beta_3\text{PE} + \beta_4\text{SE} + \beta_5\text{SN} + \varepsilon$$

Variations include additional or alternative predictors (e.g., Trust, FC, SAT, HM, System Quality), mediating paths, and indirect effects. For instance:

• Mediation Models: FC does not directly drive BI but acts via cognitive mediators (PE, EE) (Nugraha et al., 8 Dec 2025).
• Affective Bridge Models: Both favorable attitude and negative distrust mediate effects of cognitive beliefs on BI (Du, 12 Mar 2026).
• TPB Augmentation: Attitude (A), subjective norms (SN), and perceived behavioral control (PBC) serve as discrete, proximal antecedents, with moral disengagement mechanisms as higher-level predictors (Miranda et al., 20 Mar 2026).

PLS-SEM remains the dominant estimation technique due to its robustness for complex, multi-construct models and capacity for estimating latent variables in samples with moderate to large N.

4. Empirical Results and Key Predictors Across Contexts

Empirical results consistently identify several robust predictors of Behavioral Intention. A meta-analytic summary (Diao et al., 2024) across 27 studies (N=33,833) yields the following average effect sizes:

Predictor	Mean r (BI)	Interpretation
Attitude (ATT)	0.576	Strongest direct predictor
Performance Expect.	0.389	Moderate driver
Effort Expectancy	0.259	Modest effect, contextually variable
Social Influence	0.284	Consistent moderate effect
Facilitating Cond.	0.265	Indirect/enabling effect
Habit	0.296	Moderates BI, more potent in developed regions
Hedonic Motivation	0.190	Weak to moderate
Perceived Cost	–0.166	Weak negative effect

Findings in specific domains or for novel technologies often diverge from canonical TAM/UTAUT expectations:

• MetaEducation Adoption: Attitude (β=0.42–0.70) and perceived ease of use (β≈0.33–0.35) consistently predict BI, but perceived usefulness, self-efficacy, and subjective norm show attenuated or non-significant direct effects (Misirlis et al., 2023, Misirlis et al., 2023).
• E-learning Systems: Self-efficacy (β=0.29), perceived usefulness (β=0.25), and perceived ease of use (β=0.22) are dominant predictors; social norm (β=0.18) and system access (β=0.20) contribute meaningfully (Revythi et al., 2017).
• AI Method Adoption: For statistical regularization, effort expectancy (β=0.565), performance expectancy (β=0.380), and social influence (β=0.304) outstrip trust and experience in predicting BI (Thiel et al., 3 Apr 2026).
• Healthcare Metaverse: Model explains 71.8% variance in BI; presence (β=0.230), perceived ease of use (β=0.226), satisfaction (β=0.164), and perceived usefulness (β=0.169) are key drivers, with negative effects from complexity (β=–0.145) (Damar et al., 19 Oct 2025).

Notably, hedonic and affective factors (enjoyment, satisfaction, intrinsic warm-glow) commonly equal or exceed utilitarian drivers in emerging or pro-social contexts (Saravanos et al., 2022, Damar et al., 19 Oct 2025).

5. Measurement Reliability, Validity, and Limitations

High measurement reliability for BI is routinely achieved (Cronbach’s α and CR > 0.90, AVE > 0.70), underscoring the psychometric solidity of multi-item reflective scales (Nugraha et al., 8 Dec 2025, Damar et al., 19 Oct 2025, Pitts et al., 24 Feb 2026). However, gaps remain:

• Not all studies report explicit item wordings or full reliability/validity tables, limiting reproducibility (Misirlis et al., 2023, Misirlis et al., 2023).
• Self-reported intention may not always translate into observed behavior (the “intention–behavior gap”) (Miranda et al., 20 Mar 2026).
• Cross-sectional survey designs predominate, precluding causal or longitudinal interpretation and amplifying common-method bias.
• Direct effects of infrastructure or facilitating conditions are often non-significant, with their impact better modeled as indirect via salient cognitive mediators (Nugraha et al., 8 Dec 2025).

6. Contingent and Contextual Effects

Moderator and mediation analyses reveal substantial contextual variability in BI determinants:

• Region: Effort expectancy is a significant BI predictor in developing regions but negligible in developed countries; habit is stronger in developed regions (Diao et al., 2024).
• Gender: Moderates the attitude–BI relationship, with stronger effects of positive attitude on BI in samples with higher male ratios (Diao et al., 2024).
• Past Behavior: Moderates the personal-norms effect, reducing reliance on moral obligation among habitual users (Zharova et al., 2022).
• Domain-Specific Barriers: For metaverse and AI-enabled systems, unfamiliarity and ambiguity regarding system advantage can attenuate the expected linkage between perceived usefulness and BI, underscoring the need for hands-on exposure and informed demonstration (Misirlis et al., 2023, Damar et al., 19 Oct 2025, Pitts et al., 24 Feb 2026).
• Normative Mechanisms: In ethically sensitive adoption (e.g., ChatGPT for writing), moral disengagement mechanisms influence BI via attitudes and perceived control, but situational factors and cultural context are also salient (Miranda et al., 20 Mar 2026).

7. Implications and Research Directions

The empirical landscape demonstrates that optimization of BI involves coordinated intervention across cognitive, affective, social, and infrastructural levers. Effective strategies include:

• Investing in user training and experiential exposure to increase self-efficacy, perceived usefulness, and affective engagement (Misirlis et al., 2023, Revythi et al., 2017, Damar et al., 19 Oct 2025).
• Exploiting peer and institutional endorsement to shape subjective norms (Misirlis et al., 2023, Nguyen et al., 2023, Damar et al., 19 Oct 2025).
• Elevating system quality and workflow continuity to reinforce perceived system quality and error tolerance (Garcia et al., 2020, Hizam et al., 2021).
• Designing for intrinsic and extrinsic motivational rewards, including warm-glow and enjoyment, especially in pro-social or hedonic technology use (Saravanos et al., 2022, Nugraha et al., 8 Dec 2025).
• Prioritizing transparency and user control to mitigate distrust and privacy risk in AI and algorithmic systems (Du, 12 Mar 2026, Nguyen et al., 2023).

Future research priorities include explicit reporting of measurement properties, longitudinal designs to capture dynamic intention–behavior transitions, modeling of additional mediators and context moderators, and the deployment of mixed-methods and advanced ML-augmented structural modeling to capture non-linear drivers and latent segmentations (Nugraha et al., 8 Dec 2025, Zharova et al., 2022, Du, 12 Mar 2026).

---

References

• (Misirlis et al., 2023) Should I use metaverse or not? An investigation of university students behavioral intention to use MetaEducation technology
• (Revythi et al., 2017) Extension of Technology Acceptance Model by using System Usability Scale to assess behavioral intention to use e-learning
• (Thiel et al., 3 Apr 2026) Why is Regularization Underused? An Empirical Study on Trust and Adoption of Statistical Methods
• (Du, 12 Mar 2026) Examining Users' Behavioural Intention to Use OpenClaw Through the Cognition--Affect--Conation Framework
• (Nugraha et al., 8 Dec 2025) Facilitating Conditions as an Enabler, Not a Direct Motivator: A Robustness and Mediation Analysis of E-Learning Adoption
• (Miranda et al., 20 Mar 2026) Plagiarism or Productivity? Students Moral Disengagement and Behavioral Intentions to Use ChatGPT in Academic Writing
• (Garcia et al., 2020) Usability Dimensions and Behavioral Intention to Use Markdown to Moodle in Test Construction
• (Misirlis et al., 2023) An analysis of the technology acceptance model in understanding university students behavioral intention to use metaverse technologies
• (Pitts et al., 24 Feb 2026) What Drives Students' Use of AI Chatbots? Technology Acceptance in Conversational AI
• (Diao et al., 2024) A Meta-analysis of College Students' Intention to Use Generative Artificial Intelligence
• (Nguyen et al., 2023) Factors influencing to use of Bluezone
• (Zharova et al., 2022) Understanding User Perception and Intention to Use Smart Homes for Energy Efficiency: A Survey
• (Sibug et al., 30 Apr 2026) Exploring the Adoption Intention in Using AI-Enabled Educational Tools Among Preservice Teachers in the Philippines: A Partial-Least Square Modeling
• (Saravanos et al., 2022) The Effect of Warm-Glow on User Behavioral Intention to Adopt Technology: Extending the UTAUT2 Model
• (Hizam et al., 2021) User Behavior Assessment Towards Biometric Facial Recognition System: A SEM-Neural Network Approach
• (Damar et al., 19 Oct 2025) Integrating Metaverse Technologies in Medical Education: Examining Acceptance Factors Among Current and Future Healthcare Providers