Procedures
This chapter documents the step-by-step execution procedures for the AACSB international comparison study (PROC-001), based on IDEA-001.
PROC-001: AACSB International Comparison Study — Step-by-Step Execution Procedure (2026-04-08)
Phase 1: Data Collection & Coding
1-1. Sample Confirmation
Step 1: Collect AACSB-Accredited School Lists
- US: Obtain Hwang et al. (2025) existing 40-school list (from Tables 2a, 2b, 3a, 3b of the paper)
- China: Collect full list of Chinese accredited schools from AACSB official website
- Korea: Collect full list of Korean accredited schools from AACSB official website
- Source: https://www.aacsb.edu/accredited
Step 2: Determine Sample Selection Criteria
- US: Use existing 40 as-is (Top 20 UG + Top 20 MBA)
- China: Full census or top-ranked by global ranking (Financial Times, QS, etc.)
- Korea: Full census (manageable accredited count)
Step 3: Collect Sample Metadata
- Per school: accreditation years, program level (UG/MBA), city, student enrollment
1-2. Data Collection
Step 4: Collect Public Data by Country
| Data Type | US | China | Korea |
|---|---|---|---|
| Course Catalogs | University website | University website + registrar office | University website + course list |
| Syllabi | Open faculty repositories | Public course outlines (limited) | Department website + direct requests |
| Innovation Reports | School websites | Teaching reform reports | Educational innovation reports |
| Program Descriptions | Official program pages | Major introduction pages | Major introduction pages |
| AI Policy Documents | School AI guidelines | School AI usage guidelines | School AI policy documents |
Step 5: Multilingual Processing Protocol
- Preserve originals + parallel English translation
- Translation method: researcher translation + AI-assisted + cross-validation
- Coding performed in original language (assign native-proficient coders per language)
1-3. Coding
Step 6: Apply Coding Framework
Apply Hwang et al.’s (2025) 4-dimension coding framework identically:
- Dimension 1: Curricular Themes (S, E, D, En, L)
- Dimension 2: Pedagogical Approaches (C, S, B, P, L)
- Dimension 3: Assessment Modes (A, R, F, Q)
- Dimension 4: AI–CT Relationship (CT Level: H/M/L, CT Linkage: E/I/A)
Step 7: Coder Training and Pilot
- Coder composition: Minimum 2 coders per language (English/Chinese/Korean)
- Pilot coding: Code 5 schools per country first
- Inter-coder reliability: Target Cohen’s Kappa >= 0.80
- Disagreement resolution: Discussion -> coding guideline refinement -> re-coding
Step 8: Full Coding Execution
- Two independent coders -> discuss disagreements -> finalize coding
- Code additional contextual variables (national AI policy, accreditation years, etc.)
Step 9: Organize Coding Data
- Compile school-level coding results into structured spreadsheet/CSV
- File location:
docs/05_분석/Data_Raw/
Phase 2: Qualitative Comparative Mapping
Step 10: Derive Country-Specific Clusters
- Place schools on CT Level x CT Linkage matrix (identical to Hwang et al.)
- Derive clusters independently for US / China / Korea
- Verify applicability of 3 clusters (Explicit/Implicit/Nominal) + check for new cluster emergence
Step 11: Cross-National Comparison
- Compare cluster distributions (e.g., does Korea have more Nominal?)
- Compare within-cluster detail patterns (same Explicit cluster but different themes/pedagogy?)
- Visualization: Alluvial diagram or Sankey diagram for cross-national flow representation
Step 12: Qualitative Interpretation
- Explain cross-national differences using contextual variables (AI policy, educational tradition, accreditation experience)
- Construct cross-case comparison tables
Phase 3: Quantitative Analysis
3-A. Topic Modeling (STM)
Step 13: Corpus Preparation
- Collect course description/syllabus text by country
- Preprocessing: tokenization, stopword removal, stemming (language-specific)
- Construct English unified corpus (post-translation) or country-specific individual analysis
Step 14: STM Execution
- Use R
stmpackage - Determine number of topics:
searchK()function, k = 5-25 range - Model fitting:
prevalence ~ country + program_level - Output: Top keywords per topic, representative documents, topic proportions
Step 15: STM Results Analysis
- Statistical test of topic distribution by country (
estimateEffect) - Match against 5 pre-defined themes
- Interpret newly discovered themes
- File location:
docs/05_분석/Quan/
3-B. Network Analysis
Step 16: Construct Network
- Convert coding data to adjacency matrix
- Co-occurrence network: create edges between coding categories co-occurring in the same school
- Weight = co-occurrence frequency
Step 17: Execute Network Analysis
- R
igraphortidygraph+ggraphpackages - Whole network: degree centrality, betweenness centrality, community detection (Louvain)
- Country-specific sub-networks: repeat same analysis
- Cross-national structural comparison: QAP (Quadratic Assignment Procedure) via
snapackage
Step 18: Visualization
- Force-directed layout for whole and country-specific network visualization
- Node size = centrality; edge thickness = co-occurrence frequency; color = coding dimension
- File location:
docs/05_분석/Quan/
3-C. QCA
Step 19: Calibration (for fsQCA)
- Transform each condition variable to 0-1 fuzzy set
- Set calibration anchors (fully in, crossover, fully out)
- Example: CT Level Higher = 1.0, Mid = 0.5, Lower = 0.0
Step 20: Necessity Analysis
- Test whether each condition is a necessary condition for the outcome
- Consistency >= 0.90 -> necessary condition
Step 21: Sufficiency Analysis (Truth Table)
- Construct truth table -> set frequency threshold (minimum 2 cases)
- Consistency threshold >= 0.80
- Boolean minimization -> derive parsimonious / intermediate / complex solutions
Step 22: Results Interpretation
- Identify sufficient condition paths (what condition combinations lead to Explicit CT)
- Confirm equifinality (do different paths exist by country?)
- R packages:
QCAorSetMethods - File location:
docs/05_분석/Quan/
Phase 4: Integration & Triangulation
Step 23: Integrate Results
- Confirm alignment: qualitative mapping clusters vs. network communities
- Evaluate: pre-defined 5 themes vs. STM topics matching
- Map: QCA paths vs. sub-patterns within mapping clusters
Step 24: Interpret Divergences
- If divergence between methods: analyze cause and provide theoretical interpretation
- The divergence itself can be a research contribution (reporting methodological sensitivity differences)
Step 25: Write Final Report
- File location:
docs/06_라이팅 프로세스/
Phase 5: Writing & Submission
Step 26: Draft Paper
- IMRaD structure (Introduction, Method, Results, Discussion)
- Determine target journal (top AACSB/education-related journals)
Step 27: Internal Review
- All co-authors review -> incorporate feedback
Step 28: Submit
- Submit to target journal; respond to revisions
Tools and Software
| Purpose | Tool |
|---|---|
| Coding data management | Excel / Google Sheets |
| Topic Modeling | R (stm, tidytext, quanteda) |
| Network Analysis | R (igraph, tidygraph, ggraph, sna) |
| QCA | R (QCA, SetMethods) |
| Visualization | R (ggplot2, ggraph) + Quarto for reports |
| Reference management | Zotero / BibTeX |
| Collaboration | Google Drive + GitHub Pages (public/) |
Timeline (Draft)
| Period | Task |
|---|---|
| 2026-04 to 05 | Phase 1: Data collection and coding |
| 2026-05 to 06 | Phase 2: Comparative Mapping |
| 2026-06 to 08 | Phase 3: Quantitative Analysis |
| 2026-08 to 09 | Phase 4: Integration |
| 2026-09 to 10 | Phase 5: Writing & Submission |
Created: 2026-04-08 Related: IDEA-001, Hwang et al. (2025) JBMC Analysis outputs: docs/05_분석/Qual/ (mapping), docs/05_분석/Quan/ (STM, Network, QCA)