Module 3
- Seminar 1 [Jan 23]: Review of Module 2.
- Seminar 2 [Jan 26]: Hypothesis Testing, One sample test for proportions, One sample test for mean, Two sample test for Proportions.
[Notebook for class]
- Seminar 3 [Feb 2]: Two-Sample Test for Means, Case I: Variances Known (Two-Sample Z-Test), Case II: Variances Unknown but Equal (Pooled t-Test), Case III: Variances Unknown and Unequal (Welch t-Test), Paired t-Test, The F-Test for Equality of Variances.
[Notebook for class] [Diagram with formulas]
- Seminar 4 [Feb 9]: Chi-Square Tests for Categorical Data. Hypothesis Testing: Goodness-of-Fit and Test of Independence (No Association).
[Notebook for seminar][Notebook with solutions]
- Seminar 5 [Feb 16]:One-Way ANOVA. Post Hoc Tests After One-Way ANOVA. Bonferroni Post Hoc Test. Tukey’s HSD Test.Two-Way ANOVA. [Notebook with solutions 1][Notebook with solutions 2]
- Seminar 6 [Feb 27]:
- Seminar 7 [March 7]:
- Seminar 8 [March 6]:
- Seminar 9 [March 16]:
- Seminar 10 [March 23]: Consultation for Exam.
- Exam – March 31
Attendance + Quiz grades [here]
Extra Material
1. Notes from the University of Sheffield [site]
2. Mostly harmless Statistics [site]
3. Table with formulas[here]
4. Problems from Module 1 [Download]
5. Problems from Module 2 [Download]

- Monte Carlo Method
- Ansari–Bradley test
- Bootstrap confidence intervals
- Kolmogorov–Smirnov test
- Fisher’s exact test
- Motivation for GLMs, Logistic regression, Poisson regression, Interpretation of coefficients
- Cross-validation, Bootstrap methods, Bias–variance tradeoff, Model selection criteria (AIC, BIC)
- Jarque–Bera test
- verifying normality, Pearson-chi-sqrt. Shapiro–Wilk (best default), Anderson–Darling (strong tail sensitivity), Kolmogorov–Smirnov with Lilliefors correction, QQ-plot.
- EDF-based (empirical distribution function) tests:
Kolmogorov–Smirnov (KS),
Kolmogorov–Smirnov with Lilliefors correction
Anderson–Darling (AD),
Cramér–von Mises (CvM), - Kruskal–Wallis test (“non-parametric ANOVA”). Jonckheere’s trend test. Dunnett’s test. Bartlett’s test
- Breusch–Pagan test (for variance in Linear regression). Cook’s distance