AI evidence extraction

Artificial intelligence and academic integrity: exploring plagiarism in Ecuadorian universities

Juan Carlos Torres-Diaz, Josep M. Duart, Diana Rivera, Ana Beltran Flandoli · International Journal for Educational Integrity · 2025

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

8/10

Relevance

0/4

Quality (LMQS)

Evidence

Citations

3.21

FWCI

Top 10%

Impact

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.1007/s40979-025-00209-3

Methodology & findings

Study design

Cross-sectional survey study using a validated questionnaire (α = 0.73) administered to 4,811 students across eight universities in Ecuador.

Sample

N = 4811, 2 groups

Primary method

Binary logistic regression analysis with plagiarism level as the dependent variable. Internal consistency measured via Cronbach's alpha (α = 0.73). Model fit assessed using Nagelkerke R² (R² = 0.202).

Main result

The study found that "reporting the use of ChatGPT to instructors (OR = 1.223, p < 0.001) and trusting ChatGPT-generated information (OR = 1.176, p < 0.001) were positively associated with plagiarism, while perceiving that AI improves academic performance (OR = 0.945, p < 0.01) and possessing broader knowledge of AI tools (OR = 0.911, p < 0.01) were associated with lower plagiarism levels." The model explained "20.2% of the variance (Nagelkerke R² = 0.202)".

Reports effect sizes and confidence intervals.

Research paradigm

Positivist/quantitative empiricism

Author conclusions

The authors conclude that "findings highlight the importance of ethical awareness, digital literacy, and institutional policies for responsible GenAI integration" and that "this research contributes to understanding how GenAI use interacts with academic integrity in higher education and informs evidence-based approaches to promote ethical and transparent learning practices."

Risk of bias

Cross-sectional design prevents causal inference and may introduce reverse causality; Self-reported plagiarism is subject to social desirability bias; Selection bias possible if student participation was non-random across eight universities; Reliability moderate (α = 0.73) suggests potential measurement error; Model explains only 20.2% of variance, indicating unmeasured confounders; Cross-sectional design precludes causal inference; Self-reported plagiarism behavior subject to social desirability bias; Selection bias from participating universities not randomly selected; Potential confounding variables not controlled for in the analysis; Cross-sectional design prevents causal inference; Self-reported plagiarism may be subject to social desirability bias; Selection bias potentially present in sampling from only eight universities in Ecuador

Limitations

The authors acknowledge that "the study's cross-sectional design limits causal inference," which restricts the ability to determine cause-and-effect relationships between GenAI use and plagiarism.

Open questions raised

The authors identify the need for further investigation of causal mechanisms between GenAI use and plagiarism, and highlight the importance of institutional policy development and educational interventions for responsible AI integration in higher education.
Need for longitudinal studies to establish causal relationships between GenAI use and plagiarism
Requirement for studies examining institutional policy effectiveness
Need for research on digital literacy interventions
Investigation of context-specific factors in different educational systems
The authors identify the need for further research on ethical awareness, digital literacy development, and institutional policy frameworks for responsible generative AI integration in higher education contexts.

Data: not_statedCode: not_statedExtracted from: pdfAgreement 63%

Explore related topics

Citation Integrity Research Integrity Survey