"Graduate Students' ChatGPT Use in Educational Research: Action Research Project Data"

Dataset Description: ChatGPT Use in Graduate Action Research Projects
Research Hypothesis: This study hypothesized that ChatGPT significantly influences graduate students' research competencies, improving speed, writing quality, reflection, and time management during action research projects, with positive correlations between perceived usefulness and research outcomes.
Data Overview: Dataset contains responses from 19 graduate education students (82.6% response rate) from Universidad del Desarrollo, Chile, during 2023-2024. Data captures ChatGPT usage experiences across two critical action research phases: design and implementation.
Data Collection
Multi-moment approach using three online questionnaires:
Moment 1: Design phase completion - project planning usage
Moment 2: Implementation phase end - execution utilization
Moment 3: Project completion - satisfaction survey (expert-validated)

Questionnaires examined usage frequency, purposes, perceived benefits, challenges, and efficiency perceptions.
Key Findings
Usage Patterns: 61% used ChatGPT in design phase, 83% in implementation. Primary applications included objective formulation, feasibility analysis, monitoring instruments, and decision evaluation.
Benefits: 94.4% reported improved text quality, 88.9% enhanced reflection capacity, 88.9% faster task completion. Idea generation was most valued support (86.67% design, 57.89% implementation).
Challenges: 72% found prompt design required substantial information input; 44% considered it time-consuming.
Statistical Correlations: Strong positive relationships between speed-writing (r=0.969), speed-reflection (r=0.937), time investment-result quality (r=0.745), and information usefulness-utility (r=0.976).
Data Interpretation: No significant differences between research phases (p=0.798) indicate consistent tool value. Strong correlations suggest students perceiving speed improvements also report enhanced writing and reflection. Positive correlation between prompt design effort and result quality indicates initial investment yields better outcomes.
Applications: Data supports AI integration strategies in graduate education, training protocol development, research competency assessment, and investigation of AI assistance versus independent skill development balance.
Limitations: Non-probabilistic convenience sampling limits generalizability. Findings specific to education graduate students using ChatGPT 3.5 during 2023-2024. Self-reported perceptions may not reflect actual performance improvements.
Data Structure contains questionnaire responses from both design and implementation stages, statistical calculations for correlation coefficients, and accompanying graphs visualizing usage patterns and perception data.