🎉 75% of content is free forever — Unlock Premium from $10/mo →
CW
💼 Servicesℹ️ About✉️ ContactView Pricing Plansfrom $10

Statistical Testing: Hypothesis, t-tests and Chi-square

Module 4: Statistics and Probability🟢 Free Lesson

Advertisement

Hypothesis Testing FlowState H₀, H₁Collect DataCompute Statisticp-valuep < α?Reject H₀Fail to Reject

Why This Matters

Statistical testing is how we make decisions under uncertainty. Instead of guessing whether a drug works, a feature helps, or a difference is real — we quantify the evidence and let data guide our conclusions.

The Hypothesis Testing Framework

Hypothesis Testing Framework1. State H₀, H₁2. Collect Data3. Compute Statistic4. p-valuep < α?Reject H₀Fail to Reject

Visual: Rejection Regions

Rejection Regions (Two-Tailed, α = 0.05)Reject H₀Reject H₀Fail to Reject H₀−1.96+1.96z = 0α/2 = 0.025α/2 = 0.025Distribution under H₀ (null is true)

Type I and Type II Errors

Type I and Type II ErrorsActual RealityH₀ True (No Effect)H₀ False (Effect Exists)Reject H₀Fail to RejectTYPE I ERRORFalse Positive (α)CORRECT ✓True Positive (Power)CORRECT ✓True NegativeTYPE II ERRORFalse Negative (β)

The p-value: What It Actually Means

❌ WRONG
"p = 0.03 means there's a 3% chance H₀ is true"
✅ RIGHT
"If H₀ were true, there's a 3% chance of seeing data this extreme"
❌ WRONG
"p < 0.05 means the effect is large"
✅ RIGHT
"p < 0.05 means the effect is unlikely under H₀" (A tiny effect can be significant with large n)
❌ WRONG
"p > 0.05 means no effect exists"
✅ RIGHT
"p > 0.05 means we don't have enough evidence to reject H₀"

One-Sample t-test

Effect Size: Cohen's d

| |d| Value | Interpretation | |-----------|---------------| | 0.2 | Small effect | | 0.5 | Medium effect | | 0.8 | Large effect |

Complete Example

Assumptions

AssumptionDescriptionHow to Check
IndependenceObservations are independentStudy design, random sampling
NormalityData is approximately normalShapiro-Wilk test, Q-Q plot
ContinuousDependent variable is continuousData type inspection

Note: The t-test is robust to mild non-normality for n > 30 due to the Central Limit Theorem.

Two-Sample t-test

Welch's t-test (Unequal Variances)

Paired Samples

Chi-Square Test

Effect Size: Cramér's V

| |V| Value | Interpretation | |----------|---------------| | 0.1 | Small association | | 0.3 | Medium association | | 0.5 | Large association |

ANOVA (Analysis of Variance)

Why Not Use Multiple t-tests?

Why ANOVA Instead of Multiple t-tests?Multiple t-testsA vs B → test 1A vs C → test 2B vs C → test 3P(≥1 false+) = 14.3%ANOVASingle test: Are ANYgroups different?Then post-hoc: WHICH?FWER = α = 0.05

One-Way ANOVA

Post-Hoc: Tukey's HSD

Non-Parametric Tests

ParametricNon-ParametricWhen to Use
One-sample tWilcoxon signed-rankSmall sample, non-normal
Independent tMann-Whitney UUnequal variances, ordinal data
Paired tWilcoxon signed-rank (paired)Paired, non-normal differences
One-way ANOVAKruskal-WallisNon-normal, 3+ groups
Pearson rSpearman rhoNon-linear monotonic relationship

Multiple Comparisons Problem

Solutions

Power Analysis

Quick Reference: Which Test to Use

Which Statistical Test Should I Use?What are you comparing?1 group vs known value2 groups3+ groupsNormal?Non-normal?One-sample tWilcoxon signed-rankIndep.?Paired?Normal?Non-normal?Welch's t-testMann-Whitney UNormal?Non-normal?Paired t-testWilcoxon paired1 factor?2+ factors?Normal?Non-normal?One-way ANOVAKruskal-WallisNormal?NoTwo-way ANOVAFriedmanQuick Reference• Categorical data? → Chi-square test• Correlation? → Pearson (normal) or Spearman (non-linear) • Always check: Independence, Normality, Homoscedasticity

Key Takeaways

Practice Exercises

  1. Drug Trial: Blood pressure in 40 patients after a new drug. Historical mean 120 mmHg. Sample mean 115, std=12. Is the drug effective?
  2. A/B Test: Website A conversion 12.3% (n=5000), Website B 13.1% (n=5000). Is B significantly better?
  3. Survey Analysis: Association between education level and preferred news source (n=200). Chi-square test?
  4. Experiment Design: Detect medium effect (d=0.5) with 90% power. How many subjects per group?

Need Expert Data Science Help?

Get personalized tutoring, project support, or professional consulting.

Advertisement