Drucker-Simon AI Literacy Behaviors Dataset

Version 2.0 Update Notes (August 2026)To enhance analytical accuracy, methodological transparency, and cross-disciplinary generalizability, the dataset has been upgraded to Version 2.0.Crucially, to preserve the original audit trail and timestamps, all original v1.0 raw data files and deprecated scripts remain strictly unmodified and are co-located within the 02_Main_RCT_Data_and_Code/ folder alongside the new v2.0 materials.The v2.0 update includes the following three major changes:2.1.

Correction and Replacement of Analytical CodeIssue Identified: A computational logic error was identified in the original statistical scripts (ICC_analysis.py / Data_analysis.py v1.0) during a routine re-evaluation.Action Taken: The flawed scripts have been deprecated but retained for transparency. Corrected, fully documented Python scripts (Data_analysis_v2_corrected_Aug2026.py, EQ_ICC_analysis_v2_corrected_Aug2026.py, and BS_ICC_analysisv2_corrected_Aug2026.py) are now provided in 02_Main_RCT_Data_and_Code/.

These new scripts ensure accurate calculation of ICC, correlation matrices, and effect sizes. All derived results in v2.0 are generated exclusively from this corrected code.2.2. Addition of Pedagogical Instrument FilesTo improve reproducibility and allow other researchers to adapt the intervention protocol, foundational instructional materials have been organized into 01_Supplementary_Materials/:Suppl_Material1.2 Graded AI Glossary.xlsx: The tiered vocabulary list used to scaffold AI concept learning.Suppl_Material1.1_Graded Terminology Definition Table.xlsx: The standardized rubric mapping used by instructors and students to evaluate explanation quality.Additional instruments include the analytical rubric (1.3), score rubric (1.4), and the 16-item AI interaction questionnaire (1.5).2.3.

Integration of Cross-Disciplinary Transferability DataTo assess whether the intervention's effects extend beyond the original Public Administration context, we conducted a concurrent student-led ecological evaluation across 15 diverse non-STEM majors (N = 155) in August 2026.These data are stored in 03_Student_Led_CrossDisciplinary_ExplorationData/:CrossDisciplinary_PreTest_Data_Aug2026.xlsxCrossDisciplinary_PostTest_Data_Aug2026.xlsxKey Finding: These new pre/post-test datasets indicate that self-reported AI literacy gains were sustained over three months of unsupervised use (Cohen’s d = 0.85), suggesting the protocol’s core moves are highly adaptable across varied non-STEM instructional contexts.3.

Directory Structure & File Manifest├── 01_Supplementary_Materials/ # Methodological instruments & tools (v2.0)│ ├── Suppl_Material1.1_Graded Terminology Definition Table.xlsx│ ├── Suppl_Material1.2 Graded AI Glossary.xlsx│ ├── Suppl_Material1.3_Rubric.pdf│ ├── Suppl_Material1.4_Score_rubric.pdf│ └── Suppl_Material1.5_Questionnaire.pdf│├── 02_Main_RCT_Data_and_Code/ # Main experiment data, code, and qualitative evidence│ │ # (Contains both v1.0 original files and v2.0 corrections)│ ├── Suppl_Material2.1_Stu4_Original Record/ # [Folder] Actual assignment outputs & process records for 4 focal students│ ├── Behavioral Control Rating Scale.xlsx│ ├── Behavior-Quality_correlation.py # [v1.0 Deprecated - retained for audit]│ ├── BS_ICC_analysis.py # [v1.0 Deprecated - retained for audit]│ ├── BS_ICC_analysisv2_corrected_Aug2026.py # [v2.0 Corrected Code]│ ├── Data_analysis.py # [v1.0 Deprecated - retained for audit]│ ├── Data_analysis_v2_corrected_Aug2026.py # [v2.0 Corrected Code]│ ├── EQ_ICC_analysis.py # [v1.0 Deprecated - retained for audit]│ ├── EQ_ICC_analysis_v2_corrected_Aug2026.py # [v2.0 Corrected Code]│ ├── Grading_Rubric.xlsx│ ├── Random selection for Teacher C.py│ ├── Random selection for Teacher E.py│ ├── README_en.txt # Original v1.0 readme│ ├── Student_GroupInfo.csv│ ├── Suppl_Material2.2_4stu_BehavioralDetails.xlsx│ ├── Suppl_Material2.3_Follow-Up Survey Data.xlsx│ ├── Suppl_Material2.4_StudentVoices_Follow-Up.pdf # Anonymized interview transcripts│ ├── Suppl_Material2.5_Student Co-Design Protocol.pdf│ ├── Teacher_A_Scores.csv # [v1.0 Raw Data]│ ├── Teacher_B_Scores.csv # [v1.0 Raw Data]│ ├── Teacher_C_Scores_Final.csv # [v1.0 Raw Data]│ ├── Teacher_D_I_group_behavior_scores.csv # [v1.0 Raw Data]│ └── Teacher_E_I_group_behavior_scores.csv # [v1.0 Raw Data]│└── 03_Student_Led_CrossDisciplinary_ExplorationData/ # Exploratory transferability data (Aug 2026) ├── CrossDisciplinary_PreTest_Data_Aug2026.xlsx └── CrossDisciplinary_PostTest_Data_Aug2026.xlsx4.

Environment RequirementsFile Formats: Universal CSV, XLSX, PDF, and PY formats.Software: To run the updated analysis code (*_v2_corrected_Aug2026.py), a Python 3.8+ environment is recommended with the following libraries:pandasscipynumpy5. Ethics and Anonymization StatementAll participant names in the interview transcripts (Suppl_Material2.4) and behavioral records (Suppl_Material2.1, 2.2) have been anonymized or replaced with participant IDs prior to public release.The cross-disciplinary data (03_...) contains only aggregated, non-identifiable survey responses.This study was conducted under institutional ethical guidelines, and informed consent was obtained from all participants.