Data Collection Strategy

Data collection is organized as a 5-layer architecture. The AACSB Excel database (1,077 schools) determines where to look (sampling frame + metadata). The actual coding content is collected from school websites, syllabi, and Phase 2 participants. A dual-criteria sampling strategy (global ranking + data availability) ensures cross-national comparability while maintaining coding feasibility.

The 5-Layer Data Architecture

Layer 1: AACSB Excel (already secured)
  = WHO + WHERE + HOW BIG
  = Sampling frame + QCA contextual variables
  │
Layer 2: Course catalogs / program pages  (additional collection)
  = WHAT (how AI is included thematically)
  = Curricular Themes coding
  │
Layer 3: Syllabi originals  (Open Syllabus / direct collection)
  = HOW + EVIDENCE + WHY
  = Pedagogical Approaches + Assessment + AI–CT Relationship coding
  │
Layer 4: Instructor interviews  (Phase 2)
  = ENACTED (actual teaching practice)
  │
Layer 5: Student surveys  (Phase 2)
  = ACHIEVED (student CT experience)

Layer-by-Layer Details

Layer 1 — AACSB Excel (Secured)

Variable	Use
School list (1,077 schools)	Confirm population
Country (69 countries)	Filter: US 556, China 54, Korea 19
Public/Private	QCA condition variable
Student enrollment	School size variable
Program level (UG/MBA/PhD)	QCA condition variable
Program list	Pre-screening by “AI” keyword
Delivery mode	Descriptive statistics
Website URL	Entry point for Layer 2 collection

Layer 2 — Course Catalogs (Additional Collection Required)

Target	Source	Codeable Dimensions
AI-related course titles/descriptions	University website course catalog	Curricular Themes (S/E/D/En/L)
Program learning outcomes	Program pages	CT Linkage (explicit or not)

Layer 3 — Syllabi (Additional Collection Required)

Target	Source	Codeable Dimensions
Learning objective verbs	Syllabi originals	CT Level (Bloom’s mapping)
Class activity descriptions	Syllabi originals	Pedagogical Approaches (C/S/B/P/L)
Assessment/assignment items	Syllabi originals	Assessment Modes (A/R/F/Q)
Explicit CT mentions	Syllabi originals	CT Linkage (Explicit/Implicit/Absent)

Country-specific sources:

Country	Primary Source	Expected Coverage
US	Open Syllabus	~80% match expected
China	University website + direct requests	Limited public availability
Korea	University website + direct requests	Moderate availability

Layers 4–5 — Phase 2 (Interviews + Surveys)

Conducted after Phase 1 QCA results are available. Purposive sub-sample (~15–30 schools) selected from Phase 1 QCA pathways and deviant cases.

Sampling Strategy: Dual-Criteria Approach (Strategy D)

Available Sampling Strategies

Strategy	Description	Pros	Cons
A: Single ranking (Hwang et al. method)	US News Top 20 UG + 20 MBA	Simple; replicates prior study	US-centric; no cross-national comparability
B: Country-specific local rankings	US = US News, China = ARWU, Korea = JoongAng	Reflects local context	Different criteria; “Top 20” means different things per country
C: Single global ranking	QS or FT, same criteria for all three countries	Comparable across countries	Limited entries from China/Korea
D: Dual criteria — Ranking + Data availability	Global ranking controls “quality level”; data availability ensures “coding feasibility”	Best of both worlds	Requires two-step filtering

Recommended: Strategy D — Dual-Criteria Filtering

Filter 1: AACSB Accreditation  (population)
  US: 556  |  China: 54  |  Korea: 19
        ↓
Filter 2: Global Ranking Entry  (QS/FT)
  US: ~100  |  China: ~30  |  Korea: ~10–15
        ↓
Filter 3: Data Availability  (Open Syllabus / website syllabi)
  US: ~60–80  |  China: ~20–25  |  Korea: ~10–15
        ↓
Final Sample
  US: 40  (maintains Hwang et al. comparability)
  China: 20–25  (maximum available)
  Korea: 15–19  (near-census or full census)

Stratification within Each Country

Tier	Criterion	Role
Tier 1	Global Top 100	Leading schools
Tier 2	Global 101–300	Mid-tier schools
Tier 3	AACSB-accredited, not ranked	Offsets “elite bias”

Expected Country × Tier Distribution

Country	Tier 1	Tier 2	Tier 3	Total
US	15–20	10–15	5–10	~40
China	5–10	10–15	5	~20–25
Korea	3–5	5–7	5–7	~15–19
Total				~75–84

Advantages of Strategy D

Benefit	Explanation
Comparability	Same global ranking controls “quality level” across three countries
Feasibility	Data availability filter removes schools where coding is impossible
QCA use	Ranking Tier usable as QCA condition variable: “Do higher-ranked schools show Explicit CT?”
Hwang et al. continuity	US 40 maintained → direct comparison with prior study
Korea census	19 schools allows near-census → minimal sampling bias
Bias mitigation	Tier 3 inclusion reduces “elite school only” bias

QCA Adequacy Check

Country	Expected Sample	Minimum for fsQCA	Judgment
US	40	15–20	Sufficient
China	20–25	15–20	Sufficient
Korea	15–19	15–20	Borderline (full census compensates)
Total	75–84	—	Sufficient for combined analysis

COUNTRY Variable Calibration (fsQCA)

The COUNTRY variable is decomposed into structural proxies rather than used as a nominal condition:

Proxy Condition	Operationalization	Source
National AI Policy Strength	Government AI Readiness Index score	Oxford Insights
AACSB Maturity	Mean AACSB accreditation years per country	AACSB data

Ranking Tier calibration:

Score	Criterion
1.0	Tier 1 (Global Top 100)
0.67	Tier 2 (Global 101–300)
0.33	Tier 3 (AACSB-accredited, unranked)

Comparison with Hwang et al. (2025)

Item	Hwang et al.	This Study
Sampling frame	US News Top 20 UG + 20 MBA	AACSB Excel 1,077 → country-specific selection
Layer 1 data	None (direct selection)	AACSB metadata (size, type, etc.)
Layer 2 data	Course catalogs, program websites	Same + systematic collection
Layer 3 data	Publicly available syllabi	Open Syllabus (US) + direct (China/Korea)
QCA condition variables	None	AACSB Excel metadata
Countries	US only	US, China, Korea

Improvement: Using AACSB Excel metadata (Public/Private, size, programs) as QCA conditions enables institution-characteristic–based analysis that was not possible in Hwang et al.

Data Collection Execution Flow

Step 1: AACSB Excel country-specific filtering  (Layer 1)
  US 556 → Top 40–50 selected
  China 54 → Full list or top-ranked
  Korea 19 → Full census

Step 2: Access course catalogs via website URLs  (Layer 2)
  Identify AI-related courses
  Collect course descriptions

Step 3: Obtain syllabi  (Layer 3)
  US → Open Syllabus matching
  China/Korea → Website + direct requests

Step 4: Apply 4-dimension coding framework
  Layer 2–3 data coded using Hwang et al. framework

Step 5: QCA + Comparative Mapping  (Phase 1 analysis)

Step 6: Purposive case interviews/surveys  (Layers 4–5, Phase 2)

Data Requirements to Collect

Data	Source	Notes
QS Global MBA Ranking	topuniversities.com	Published annually, free to view
FT Global MBA Ranking	rankings.ft.com	Published annually, free to view
Shanghai ARWU Business	shanghairanking.com	Research-based; includes Chinese schools
Government AI Readiness Index	Oxford Insights	For `COUNTRY` variable calibration
OECD AI Policy Index	oecd.ai	Supplementary AI policy data