AI evidence extraction

Adoption of artificial intelligence tools for academic writing

Nguyen Thu Hoai, Lai Thi Thu Thuy · International Journal of Evaluation and Research in Education (IJERE) · 2026

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

9/10

Relevance

2/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.11591/ijere.v15i2.37993

Methodology & findings

Study design

Cross-sectional survey with structural equation modeling (SEM).

Sample

N = 404, 3 groups

Primary method

Structural equation modeling (SEM) using IBM SPSS Statistics 26 and AMOS 24. Two-stage process: (1) Confirmatory factor analysis (CFA) to evaluate measurement model psychometric properties including convergent validity (AVE>0.5), discriminant validity (Fornell-Larcker criterion), and composite reliability (CR>0.7); (2) Structural model analysis testing hypothesized relationships. Goodness-of-fit evaluated using χ²/df<3, CFI>0.9, TLI>0.9, RMSEA<0.08. Internal consistency assessed via Cronbach's alpha. Pearson correlation analysis for bivariate relationships.

Main result

The study found that "PE emerged as the strongest driver, exerting a substantial positive impact on BI (β=0.585, p<0.001)" and that "the model was highly effective, accounting for 75.3% of the variance in UB (R²=0.753)," demonstrating that "all seven hypotheses were supported, and the high R² values confirm the model's robust ability to explain the factors driving the adoption of AI writing tools among the surveyed academic population."

Reports effect sizes and confidence intervals.

Research paradigm

positivist

Author conclusions

The authors conclude: "The adoption of AI tools for academic research by Vietnamese lecturers is a complex process driven by a rational evaluation of benefits and usability, but significantly tempered by perceptions of risk. This study successfully integrated the UTAUT model and PRT to demonstrate that PE and EE are key drivers of intention, while perceived risk is a formidable barrier. To translate intention into actual use, both individual motivation and strong institutional support in the form of FC are indispensable."

Risk of bias

Selection bias: Non-probability sampling (convenience and snowball sampling) limits representativeness; Social desirability bias: Self-reported survey data; Common method bias: All data collected via single survey method; Cross-sectional design: Cannot establish causality or capture temporal dynamics; Selection bias from non-probability sampling (convenience and snowball sampling); Self-report bias - reliance on self-reported measures; Common method bias - all data collected via single survey method; Social desirability bias - academics may overstate intentions to adopt AI; Generalizability limitation - sample may not be representative of all Vietnamese lecturers; Selection bias: non-probability sampling (convenience and snowball) limits representativeness; Common method bias: self-report data only; Social desirability bias: lecturers may respond based on perceived institutional expectations; Cross-sectional design prevents causal inference

Limitations

The authors state: "First, this study utilized non-probability sampling techniques, including convenience and snowball sampling
Therefore, the sample may not be fully representative of the entire population of Vietnamese lecturers, which limits the generalizability of the findings
Second, the cross-sectional design provides only a static snapshot of AI adoption
a longitudinal study would be beneficial for capturing the dynamic nature of technology acceptance over time
Third, the data is based on self-reports, which may be subject to common method bias or social desirability bias."

Open questions raised

Limited research on determinants of AI adoption by faculty in developing countries (most studies in Western contexts)
Need for longitudinal studies to capture dynamic nature of technology acceptance over time
Qualitative studies needed to provide richer insights into specific nature of lecturers' perceived risks and mitigation strategies
Comparative studies exploring how adoption factors differ across academic disciplines
Incorporation of other variables such as personal innovativeness, trust in AI, or moderating effect of prior AI experience
Lack of research on AI adoption determinants in developing countries, particularly Vietnam

Data: No datasets mentioned as publicly available.; Not mentioned as available in the paper.Code: No code repositories mentioned.; Not mentioned in the paper.Extracted from: pdfAgreement 50%

Explore related topics

AI Manuscript Writing Academic Writing Empirical Study