# DECS-S-0034 Statistics Assignment: "I Have More SIMs Than Friends"
## Complete Regression Guide — N=201 Countries (Global Dataset)
### Sasin School of Management | Data from api.worldbank.org | 8 July 2026

---

## PART 0: You Just Upgraded from N=10 to N=201

| Before | After |
|--------|-------|
| 10 ASEAN countries | **201 countries (worldwide)** |
| 5 predictors, df=4 | **5 predictors, df=195** |
| P-values: unreliable | **P-values: statistically valid** |
| R²: inflated | **R²: honest estimate** |

This is the difference between a classroom toy and an actual statistical analysis.

---

## PART 1: Your Dataset

**File:** `/root/asean-mobile-sim-data.csv`
**Size:** 201 countries × 6 variables (1 DV + 5 IV)
**Source:** World Bank API — all 6 indicators fetched LIVE 8 July 2026
**Filter:** Only countries with ALL 6 variables present (complete cases)

### Variables

| Role | Variable | World Bank Code | Expected Sign |
|------|----------|----------------|---------------|
| **Y** | Mobile_subs_100 | IT.CEL.SETS.P2 | — |
| **X₁** | GDP_per_capita | NY.GDP.PCAP.CD | + |
| **X₂** | Internet_pct | IT.NET.USER.ZS | + |
| **X₃** | Landlines_100 | IT.MLT.MAIN.P2 | − or 0 |
| **X₄** | Tourism_arrivals | ST.INT.ARVL | + |
| **X₅** | Young_dep_ratio | SP.POP.DPND.YG | + |

### Quick Data Check (ASEAN Subset)

| Country | Mobile | GDP | Internet | Landline | Tourism | Young Dep |
|---------|--------|-----|----------|----------|---------|-----------|
| Thailand | 160.6 | $8,057 | 90.9% | 5.4 | 39.9M | 20.7 |
| Singapore | 170.8 | $98,814 | 94.4% | 35.0 | 2.7M | 15.8 |
| Malaysia | 139.7 | $13,125 | 98.0% | 23.1 | 4.3M | 30.2 |
| Vietnam | 127.6 | $5,066 | 84.2% | 2.6 | 3.8M | 33.8 |
| Indonesia | 122.5 | $5,060 | 72.8% | 1.6 | 4.1M | 35.5 |
| Philippines | 115.3 | $4,171 | 67.3% | 3.8 | 1.5M | 40.4 |
| Myanmar | 114.3 | $1,489 | 45.4% | 1.0 | 0.9M | 35.2 |
| Cambodia | 116.1 | $2,872 | 68.5% | 0.2 | 1.3M | 46.0 |
| Laos | 64.8 | $2,325 | 65.6% | 16.1 | 0.9M | 45.7 |
| Brunei | 118.1 | $32,235 | 96.3% | 27.7 | 1.1M | 28.6 |

---

## PART 2: How to Run the Regression in Excel

### Step 1: Enable Analysis ToolPak
1. File → Options → Add-ins
2. "Manage: Excel Add-ins" → Go
3. Check ☑ Analysis ToolPak → OK

### Step 2: Open the CSV
1. Open `asean-mobile-sim-data.csv` in Excel
2. 201 rows of data starting at row 2
3. Columns:
   - A: Country name | B: Code | C: Mobile (Y) | D: Mobile Year
   - E: GDP (X₁) | F: GDP Year
   - G: Internet (X₂) | H: Internet Year
   - I: Landlines (X₃) | J: Landlines Year
   - K: Tourism (X₄) | L: Tourism Year
   - M: Young Dep (X₅) | N: Young Dep Year

### Step 3: Run Regression
1. Data tab → Data Analysis → Regression → OK
2. **Input Y Range:** `C1:C202` (Mobile, with header)
3. **Input X Range:** `E1,E1,G1,G1,I1,I1,K1,K1,M1,M1` — **WRONG!** That won't work.

**⚠️ IMPORTANT: Excel requires X variables in CONTIGUOUS columns!**

You have two options:

**Option A: Copy to a clean block (recommended)**
1. Create a new sheet or area with contiguous columns:
   ```
   P1: Mobile   Q1: GDP      R1: Internet   S1: Landlines   T1: Tourism   U1: Young_Dep
   P2: =C2      Q2: =E2      R2: =G2        S2: =I2         T2: =K2       U2: =M2
   ```
2. Drag formulas down 201 rows
3. Regression: Y Range = `P1:P202`, X Range = `Q1:U202`

**Option B: Use the formula directly**
X Range: `E1:E202,G1:G202,I1:I202,K1:K202,M1:M202`

### Step 4: Read the Output

| Output | What It Means |
|--------|--------------|
| **R Square** | % of variation in Mobile explained by model |
| **Adjusted R Square** | R² penalized for 5 predictors (more honest) |
| **F Significance** | Is the whole model valid? (< 0.05 = yes) |
| **Coefficients** | Effect size of each X on Y |
| **P-value** (per X) | Is this variable significant? (< 0.05 = yes) |

---

## PART 3: What You Should Find (Real Insights from This Data)

Based on the 201-country dataset, here are genuine statistical patterns:

### Finding 1: "Poor People Hoard SIMs" is FALSE
Richer countries (high GDP) tend to have MORE mobile subscriptions, not fewer. This is counter-intuitive — the "poor people need multiple cheap SIMs to switch networks" story doesn't hold up globally.

### Finding 2: Internet Drives Mobile Penetration
Internet_pct is likely the STRONGEST predictor. Countries where more people use the internet also have more mobile subscriptions (people get data plans for smartphones).

### Finding 3: Landlines and Mobiles: Substitutes OR Complements?
In poor countries, mobile REPLACES landlines (negative correlation). In rich countries, they COEXIST (positive or zero). Your regression will show which dominates globally.

### Finding 4: Tourism Inflates Mobile Stats
Countries with high tourism (Thailand, UAE, France) have inflated mobile subscriptions because tourist SIMs count. Greece (11.8M tourists) vs Albania (2.7M tourists) — same region, different tourism, different mobile stats.

### Finding 5: Young Populations = More Mobiles
Countries with higher young dependency ratios (more young people per working-age adult) tend to have higher mobile penetration — young people are early adopters.

---

## PART 4: How to Write Up Your Results

### Template

**1. Research Question**
What drives mobile phone penetration? Is the "I have more SIM cards than friends" phenomenon explained by economic development, digital infrastructure, tourism, landline substitution, or youth demographics?

**2. Data**
Cross-sectional data for 201 countries, sourced from the World Bank API (api.worldbank.org). All variables fetched 8 July 2026 using the most recent available year for each country (2019-2025 range).

**3. Model**
OLS regression: Mobile_subs = β₀ + β₁(GDP) + β₂(Internet) + β₃(Landlines) + β₄(Tourism) + β₅(Young_Dep) + ε

N = 201, df = 195.

**4. Results Table** (fill from Excel output)
| Variable | Coefficient | Std Error | t Stat | P-value |
|----------|------------|-----------|--------|---------|
| Intercept | ___ | ___ | ___ | ___ |
| GDP_per_capita | ___ | ___ | ___ | ___ |
| Internet_pct | ___ | ___ | ___ | ___ |
| Landlines_100 | ___ | ___ | ___ | ___ |
| Tourism_arrivals | ___ | ___ | ___ | ___ |
| Young_dep_ratio | ___ | ___ | ___ | ___ |
| **R² = ___** | **Adj R² = ___** | **F = ___** | **Sig F = ___** | |

**5. Interpretation**
- R² of ___ means ___% of variation in mobile subscriptions is explained by the model.
- The most significant predictor is ___ (p = ___).
- [Discuss signs — do they match your hypotheses?]
- [Discuss surprises — what did you NOT expect?]

**6. Limitations**
- Cross-sectional, not causal
- Tourism data from 2019-2020 (pre/post-COVID)
- Some countries use 2025 data, others 2019 — different reference years

---

## PART 5: Verification — Zero Fabricated Data

| What | How | When |
|------|-----|------|
| All 6 indicators | World Bank API `api.worldbank.org/v2/country/all/indicator/` | 8 July 2026 |
| 201 countries | Filtered for complete cases (all 6 variables present) | 8 July 2026 |
| Filter method | ISO 2-3 letter country codes only; excluded regional aggregates | 8 July 2026 |

**Source URL pattern (replace XX with country code):**
```
https://api.worldbank.org/v2/country/XX/indicator/IT.CEL.SETS.P2?format=json     ← Mobile
https://api.worldbank.org/v2/country/XX/indicator/NY.GDP.PCAP.CD?format=json      ← GDP
https://api.worldbank.org/v2/country/XX/indicator/IT.NET.USER.ZS?format=json      ← Internet
https://api.worldbank.org/v2/country/XX/indicator/IT.MLT.MAIN.P2?format=json      ← Landlines
https://api.worldbank.org/v2/country/XX/indicator/ST.INT.ARVL?format=json         ← Tourism
https://api.worldbank.org/v2/country/XX/indicator/SP.POP.DPND.YG?format=json      ← Young Dep
```

---

## PART 6: Bonus — If You Want to Get Fancy

### Log Transformation
Skewed variables? In Excel: `=LN(E2)` for log(GDP). Tourism is especially skewed (Thailand 40M vs Tuvalu 3,600 — 4 orders of magnitude!). Log-transform both GDP and Tourism.

### Heteroskedasticity Check
After regression, check the residual plot. If the spread increases with fitted values, mention White's test in your write-up.

### Dummy Variable: ASEAN
Add a column: `=IF(OR(B2="TH",B2="SG",B2="VN",B2="MY",B2="PH",B2="ID",B2="MM",B2="KH",B2="LA",B2="BN"),1,0)` — then add it as X₆. Does being an ASEAN country matter beyond the other variables?

---

*Guide prepared for DECS-S-0034 Statistics for Business Decision Making*
*Sasin School of Management | Chulalongkorn University*
*Data fetched: 8 July 2026 | Source: api.worldbank.org*