Unified Theory of Acceptance & Use Tech 2

• UTAUT2 is a comprehensive framework that extends existing technology acceptance models by incorporating constructs like hedonic motivation, price value, and habit.
• Recent studies employ advanced SEM techniques to operationalize UTAUT2 with high psychometric standards and reliable reflective measurement models.
• Empirical evidence from domains such as software engineering, mobile banking, and pro-social technologies shows that UTAUT2 extensions significantly boost predictive power for behavioral intentions and usage.

The Unified Theory of Acceptance and Use of Technology 2 (UTAUT2) is a comprehensive theoretical and empirical framework for modeling and explaining technology acceptance and use, developed to extend the explanatory power of its predecessor, UTAUT. UTAUT2 consolidates determinants from major technology adoption traditions and incorporates constructs specifically adapted for consumer contexts. Across contemporary empirical studies, UTAUT2 is operationalized with high psychometric standards, and its extensions have demonstrated significant explanatory power across contexts such as software engineering, text-to-video models, banking, and technologies with pro-social valence.

1. Theoretical Origins and Core Constructs

UTAUT2 was proposed by Venkatesh, Thong, and Xu (2012) as an extension of the original UTAUT. The foundational UTAUT model synthesized eight pre-existing technology acceptance theories to identify four core determinants of Behavioral Intention (BI) and Use Behavior (UB): Performance Expectancy (PE), Effort Expectancy (EE), Social Influence (SI), and Facilitating Conditions (FC). UTAUT2 augments this with three constructs—Hedonic Motivation (HM), Price Value (PV), and Habit (HB)—to better capture consumer context behaviors, increasing explained variance in BI from 56% to 74% and in UB from 40% to 52% (Lambiase et al., 2024).

The canonical UTAUT2 framework employs the following constructs:

• Performance Expectancy (PE): Belief in performance improvement gained from technology use.
• Effort Expectancy (EE): Perceived ease of using the technology.
• Social Influence (SI): Perceived pressure from important others to use the technology.
• Facilitating Conditions (FC): Availability of organizational or technical infrastructure to support use.
• Hedonic Motivation (HM): Pleasure or enjoyment derived from use.
• Price Value (PV): Cognitive trade-off between benefits and monetary cost.
• Habit (HB): Automaticity of behavior due to prior experience.

These constructs are measured reflectively, usually with multiple validated items on a Likert scale (typically 7-point), and loadings above 0.7 are required for indicator reliability.

2. Structural Model and Variants

The standard UTAUT2 structural model links the seven predictors to BI, and links BI (and sometimes FC/HB) to UB. The canonical equations expressed in the literature are:

$$\mathrm{BI} = \beta_{1}\,\mathrm{PE} + \beta_{2}\,\mathrm{EE} + \beta_{3}\,\mathrm{SI} + \beta_{4}\,\mathrm{FC} + \beta_{5}\,\mathrm{HM} + \beta_{6}\,\mathrm{PV} + \beta_{7}\,\mathrm{HB} + \varepsilon_{\mathrm{BI}}$$

$$\mathrm{UB} = \beta_{8}\,\mathrm{BI} + \beta_{9}\,\mathrm{FC} + \beta_{10}\,\mathrm{HB} + \varepsilon_{\mathrm{UB}}$$

Empirical implementations occasionally exclude PV if direct costs are not user-borne or are negligible (e.g., LLMs with organizational licenses) (Lambiase et al., 2024, Lambiase et al., 3 Apr 2025). Extensions may incorporate additional context-dependent constructs, e.g., Perceived Realism and Novelty Value for text-to-video models (Mvondo et al., 2024), security and trust for mobile banking (Apaua et al., 2022), or pro-social “warm-glow” in social technology adoption (Saravanos et al., 2022).

3. Empirical Methodologies and Operationalization

Recent studies apply UTAUT2 using advanced structural equation modeling approaches (PLS-SEM or CB-SEM). Constructs are operationalized with reflective measurement models, ensuring:

• Indicator reliability (loadings >0.7)
• Internal consistency (Cronbach’s α, composite reliability ρ_c >0.7)
• Convergent validity (Average Variance Extracted, AVE ≥ 0.5)
• Discriminant validity (HTMT < 0.85)
• No problematic collinearity (VIF < 3 or 5 depending on context)
• Model-level fit indices (e.g., CFI, TLI, RMSEA) show excellent fit where reported (Mvondo et al., 2024, Apaua et al., 2022).

The dependent variables are BI (multi-item) and UB (single- or multi-item frequency scales), and, in certain contexts, task-specific UB variants distinguish behavioral patterns with greater granularity (Lambiase et al., 3 Apr 2025).

4. Key Findings Across Domains and Contexts

UTAUT2 has been validated in multiple advanced technology contexts, with empirical findings summarized below:

• LLM Adoption in Software Engineering: PE is the dominant predictor of BI (β = 0.463, p < 0.001, f² = 0.17); HB is next (β = 0.274, p < 0.001, f² = 0.09). Actual use is primarily explained by HB (β = 0.378, p < 0.001, f² = 0.12), with significant but smaller effects from FC and BI. EE, SI, HM, and FC have negligible or non-significant effects on intention in this domain (Lambiase et al., 2024).
• Task-Specific LLM Adoption: HB and general usage frequency strongly drive all task-specific usage measures. EE and SI are influential on select tasks, while PE strongly impacts BI and general UB but is weak or negatively mediated at the task level (Lambiase et al., 3 Apr 2025).
• Text-to-Video Modeling (SORA): Perceived Realism and Novelty Value are substantively stronger determinants of willingness to use than canonical UTAUT2 constructs. Solution pathways for high adoption involve additive and conjunctive combinations of PR, NV, PE, EE, SI, and HM (Mvondo et al., 2024).
• Mobile Banking: Extensions with Perceived Security, Institutional Trust, and Technological Trust significantly improve variance explained (R²(BI) rises from 0.74 to 0.79, R²(UB) from 0.52 to 0.55 over baseline UTAUT2) (Apaua et al., 2022).
• Pro-Social ("Warm-Glow") Technology: Intrinsic and extrinsic "warm-glow" effects (affective utility of pro-social action) can outperform PE and HM in predicting BI, signaling that moral or social utility constructs substantially enhance UTAUT2’s explanatory power for such technologies (Saravanos et al., 2022).

Context	Strongest Predictor(s)	R² (BI)	R² (UB)	Reference
LLMs in SE (general)	PE, HB	0.64	0.41	(Lambiase et al., 2024)
LLMs (task-specific)	HB, UB	0.29–0.46 (UBₖ)	—	(Lambiase et al., 3 Apr 2025)
Text-to-video (SORA)	PR, NV, PE	0.589	—	(Mvondo et al., 2024)
Mobile Banking	Perceived Security, Trust	0.79	0.547	(Apaua et al., 2022)
Pro-social tech ("WG")	Intrinsic Warm-Glow	0.666	—	(Saravanos et al., 2022)

Behavioral intention is frequently more strongly predicted than actual usage, in line with technology acceptance theory.

5. Moderators and Extensions

UTAUT2 accommodates context-specific moderators including demographic variables (age, gender, experience, income, education), cultural values (Hofstede dimensions), and technology-specific antecedents such as perceived security or pro-social affect. For example, in LLM adoption by software engineers, hypothesized moderation by Hofstede cultural dimensions (PDI, UAI, COL, MAS, LTO) was empirically unsupported (all β_interaction ns), suggesting that domain-specific practicalities can dominate over classic socio-cultural moderators (Lambiase et al., 2024).

Mobile banking research indicates complex moderation patterns, e.g., perceived security effects on BI and UB are strongest in young males with low experience, institutional trust more salient for older, experienced women, and technological trust for highly educated, experienced users (Apaua et al., 2022).

Extensions such as perceived realism, novelty value, and warm-glow are context-driven and required for full explanatory fidelity in emerging or domain-specific technologies (Mvondo et al., 2024, Saravanos et al., 2022).

6. Methodological Rigor and Limitations

UTAUT2 research exhibits methodological rigor: large samples (n > 180, often n > 300), multi-item reflective measurement, systematic reliability/validity analysis, calibrated SEM approaches, and—where relevant—multi-path or qualitative comparative analysis (e.g., fsQCA in (Mvondo et al., 2024)).

Notable limitations arise from context: exclusion of constructs (e.g., PV when direct payment is rare), reliance on self-reported use frequency, and potential response biases due to online survey administration. Certain models, such as those for pro-social or warm-glow technologies, rely on hypothetical scenarios and do not measure actual usage (Saravanos et al., 2022).

7. Implications, Practical Guidance, and Research Directions

UTAUT2 consistently shows high explanatory and predictive power when its operationalization is fit to domain. Key practical recommendations include:

• Highlighting performance benefits and embedding technologies into daily workflow to support habit formation (Lambiase et al., 2024, Lambiase et al., 3 Apr 2025).
• Ensuring seamless integration and usability for task-specific applications, focusing on effort expectancy where relevant (Lambiase et al., 3 Apr 2025).
• Addressing domain-specific concerns such as perceived security and trust for sensitive domains (Apaua et al., 2022).
• Where affective or pro-social outcomes are salient, extending UTAUT2 to capture these is necessary for accurate modeling (Saravanos et al., 2022).

Future research is directed towards further contextual extension, advanced moderation modeling, and longitudinal studies to track the evolution of adoption determinants over technology life cycles (e.g., as LLMs pass through the Gartner Hype Cycle).

---

In summary, UTAUT2 is the state-of-the-art, empirically validated framework for modeling, explaining, and predicting technology acceptance and use for both general consumer and specialized professional technologies. Its core constructs, robust measurement structure, and extensibility render it a foundational theory in technology adoption research (Lambiase et al., 2024, Lambiase et al., 3 Apr 2025, Mvondo et al., 2024, Saravanos et al., 2022, Apaua et al., 2022).