MODULE 04
Statistics and Inference
From descriptive summaries to hypothesis tests, estimation theory, experiment design and causal reasoning.
23 lessons~10h reading
- 0124 min
Descriptive Statistics
BeginnerComing soonMean, median, mode, range, variance, standard deviation, quartiles and IQR, computed by hand.
- 0220 min
Skewness and Kurtosis
IntermediateComing soonThird and fourth standardised moments, what they say about shape, and how to read them off a histogram.
Assumes: Descriptive Statistics
- 0322 min
Populations, Samples and Sampling Methods
BeginnerComing soonSimple random, stratified, cluster and systematic sampling, and the biases each introduces.
- 0426 min
Sampling Distributions and Standard Error
IntermediateComing soonThe distribution of a statistic, standard error of the mean, and the finite population correction.
Assumes: The Central Limit Theorem
- 0528 min
Point Estimation and Estimator Properties
AdvancedComing soonBias, variance, mean squared error, consistency, efficiency and sufficiency.
Assumes: Sampling Distributions and Standard Error
- 0630 min
Maximum Likelihood Estimation
IntermediateComing soonWriting a likelihood, log-likelihood tricks, and deriving MLEs for Bernoulli, normal and Poisson by hand.
Assumes: Point Estimation and Estimator Properties
- 0720 min
Method of Moments
IntermediateComing soonMatching sample to population moments, and comparing the resulting estimators with MLE.
Assumes: Maximum Likelihood Estimation
- 0830 min
Bayesian Estimation and Conjugate Priors
AdvancedComing soonPrior to posterior updates, beta-binomial and normal-normal conjugacy, MAP vs posterior mean.
Assumes: Maximum Likelihood Estimation · Gamma and Beta Distributions
- 0928 min
Confidence Intervals
BeginnerComing soonConstructing intervals for means and proportions, what 95% actually means, and the width/confidence trade-off.
Assumes: Sampling Distributions and Standard Error
- 1030 min
The Hypothesis Testing Framework
BeginnerComing soonNull and alternative hypotheses, test statistics, rejection regions, p-values and how to interpret them honestly.
Assumes: Confidence Intervals
- 1128 min
Type I/II Errors, Power and Sample Size
IntermediateComing soonThe error trade-off, computing power, and determining the sample size a study needs.
Assumes: The Hypothesis Testing Framework
- 1224 min
The z-Test
BeginnerComing soonOne- and two-sample z-tests for means and proportions, with full numeric walkthroughs.
Assumes: The Hypothesis Testing Framework
- 1330 min
t-Tests
BeginnerComing soonOne-sample, two-sample pooled, Welch and paired t-tests, and choosing between them.
Assumes: The z-Test · t, Chi-Squared and F Distributions
- 1428 min
Chi-Squared Tests
IntermediateComing soonGoodness-of-fit and tests of independence on contingency tables, with expected-frequency calculations.
Assumes: The Hypothesis Testing Framework
- 1532 min
F-Test and ANOVA
IntermediateComing soonComparing variances, one-way and two-way ANOVA, sum-of-squares decomposition and post-hoc tests.
Assumes: t-Tests
- 1628 min
Non-Parametric Tests
IntermediateComing soonSign, Wilcoxon, Mann–Whitney, Kruskal–Wallis and Kolmogorov–Smirnov tests, and when to prefer them.
Assumes: t-Tests
- 1724 min
Multiple Testing Correction
AdvancedComing soonFamily-wise error rate, Bonferroni, Holm, and Benjamini–Hochberg FDR control.
Assumes: Type I/II Errors, Power and Sample Size
- 1822 min
Inference for Correlation
IntermediateComing soonTesting Pearson correlation, Spearman and Kendall alternatives, and the dangers of correlation mining.
Assumes: Covariance and Correlation · t-Tests
- 1932 min
Regression Inference and Diagnostics
AdvancedComing soonStandard errors of coefficients, t and F tests in regression, R², residual plots and influence measures.
Assumes: F-Test and ANOVA · Least Squares and the Normal Equations
- 2028 min
Resampling: Bootstrap and Permutation
AdvancedComing soonBootstrap confidence intervals, the jackknife, and permutation tests that need no distributional assumption.
Assumes: Confidence Intervals
- 2126 min
Design of Experiments
IntermediateComing soonRandomisation, blocking, factorial designs, confounding and replication.
Assumes: F-Test and ANOVA
- 2230 min
A/B Testing in Practice
IntermediateComing soonDesigning an online experiment end to end: metrics, guardrails, peeking, novelty effects and sequential tests.
Assumes: Type I/II Errors, Power and Sample Size
- 2332 min
Causal Inference Basics
AdvancedComing soonConfounding, Simpson's paradox, causal DAGs, randomised vs observational studies, and simple adjustment strategies.
Assumes: Design of Experiments