12,445 papers · continuously updated · last export: 10 Aug 2026livingmeta.ai
← Browse all papers
AI evidence extraction

From tool to partner: generative AI usage patterns and research performance among doctoral students

Wenqin Shen · Studies in Higher Education · 2026

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

9/10
Relevance
0/4
Quality (LMQS)
E
Evidence
1
Citations
16.72
FWCI
Top 10%
Impact

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.1080/03075079.2026.2625386

Methodology & findings

Study design

Large-scale survey using latent class analysis and ordered logit regression.

Sample

> 1000, 3 groups

Primary method

Latent class analysis, ordered logit regression

Main result

The study identified three distinct GenAI usage modes among doctoral students: "Textual Tool (51.4%), Task Assistant (43.8%), and Thought Partner (4.8%)". The findings reveal that "Textual Tool use is mainly linked to gains in publication productivity, whereas the more comprehensive integration represented by Thought Partner is associated with improvements in dissertation quality", with patterns varying significantly by disciplinary context.

Reports effect sizes.

Research paradigm

Mixed methods (quantitative survey with latent class analysis)

Author conclusions

"The findings highlight that the implications of GenAI for doctoral research depend on how it is integrated into disciplinary research practices, underscoring the need for discipline-sensitive approaches to GenAI governance in doctoral education."

Risk of bias

Self-reported usage patterns (self-report bias); Cross-sectional survey design (cannot establish causality); Chinese sample (generalizability concerns to other geographic contexts); Potential selection bias in survey participation; Self-reported GenAI usage patterns (potential reporting bias); Chinese doctoral students only (geographic/cultural generalizability concerns); Confounding variables by discipline and supervisor guidance satisfaction; Self-reported survey data on GenAI usage patterns (potential social desirability bias); Cross-sectional design (cannot establish causality); Chinese doctoral population only (limited generalizability); Potential selection bias in survey response rates not mentioned; Reverse causality possible: better-performing students may adopt GenAI differently

Open questions raised

  • The authors suggest the need for discipline-sensitive approaches to GenAI governance in doctoral education, implying further research is needed on disciplinary differences in GenAI integration and governance frameworks
  • The abstract does not explicitly identify specific research gaps or future directions.
  • The paper identifies the need for discipline-sensitive approaches to GenAI governance in doctoral education and suggests that future research should examine how different disciplinary contexts condition GenAI adoption patterns and research outcomes.
Data: not_statedCode: not_statedExtracted from: pdfAgreement 63%

Explore related topics

Productivity EffectsDoctoral TrainingEmpirical Study

Related papers

← Browse all papers