Statistics Basics: Confidence Intervals, Z-Tests, and Hypothesis Testing

📊 Confidence Intervals — Intuition + Math (Simple Guide)

📚 Table of Contents

• Confidence Interval
• Alpha
• Hypothesis

---

1. Confidence Interval (What it REALLY Means)

💡 Imagine measuring average income of a city using just 25 people. You won’t get the exact true value — but you can estimate a **range** where it likely lies.

📖 Step-by-Step + Intuition

Step 1: Standard Error

SEM = 100 / sqrt(25) = 20

🧠 Simple Meaning:
- 100 = how spread out people are (population variability)
- √25 = how much data we collected

👉 More data → less uncertainty 👉 Less data → more uncertainty

💡 Think of SEM as “average mistake” in your estimate.

Step 2: Confidence Multiplier

1.96 comes from statistics tables.

🧠 Meaning:
To be 95% sure, we go about **2 steps away** from the average.

Step 3: Final Interval

CI = 520 ± (1.96 × 20) CI = (480.8, 559.2)

🧠 Meaning:
We are saying:
👉 “The real average is probably somewhere between 480 and 559.”

💻 CLI Simulation

$ ci --mean 520 --sigma 100 --n 25 Output: Range: 480.8 to 559.2

---

2. Alpha (Error Risk)

📖 Simple Explanation

alpha = 1 - 0.95 = 0.05

🧠 Meaning:
Out of 100 times: - 95 times → correct decision - 5 times → wrong decision

💡 Alpha = Risk you are willing to take.

---

3. Hypothesis (Decision Making)

📖 Simple Explanation

H0: mu = 500 H1: mu ≠ 500

🧠 Meaning:
- H₀ = default belief (nothing changed) - H₁ = something different is happening

💡 Statistics tries to **challenge assumptions**, not prove them directly.

---

4. Why Bigger Sample is Better

💡 Imagine guessing average height: - Ask 5 people → unreliable - Ask 100 people → accurate

SE = 100 / sqrt(40) ≈ 15.81

🧠 More data → smaller error → tighter range

---

5. Z-Test (How We Decide)

📖 Intuition

Z = (520 - 500) / 20 = 1

🧠 Meaning:
Your result is only 1 step away from expected value.

👉 Not far enough → nothing unusual

💡 Big Z (like 2 or 3) = strong evidence 💡 Small Z (like 1) = weak evidence

---

6. 70% Confidence (Less Strict)

CI = 520 ± 20.8

🧠 Lower confidence = narrower range = more risk

---

7. Proportion Test (Real-Life Example)

💡 Example: 55% people like a product vs expected 50%

Z = 1

🧠 Difference is small → not meaningful

---

🔗 Related Articles

• Confidence Interval Deep Dive
• Z vs T Test Guide
• Statistical Thinking Basics

---

🎯 Final Understanding:

• Confidence interval = smart guessing range
• Alpha = risk level
• Z-score = how surprising your result is
• More data = better decisions
• Statistics = decision tool, not truth machine