M3S10 and M4S10

DESIGN OF EXPERIMENTS AND SURVEYS

OUTLINE OF THE COURSE

Introduction

What is an experiment? What is a survey?

EXPERIMENTAL DESIGN

Requirements of a Good Design

Aims. Systematic Effects. Randomisation. Precision. Clinical Trials.

Block Designs

Balanced Incomplete Block Designs.

Latin Squares

Properties of Latin Squares. Graeco-Latin Squares. Mutually Orthogonal Latin Squares. Youden Rectangles.

Factorial Experiments

Complete Factorial Experiments. Experiments with all factors at 2 levels. Confounding. Partial Confounding. Fractional Replication. Aliasing. Pseudo-Factors. Construction of Factorial Experiments from Mutually Orthogonal Latin Squares. Taguchi Methods.

Optimal Designs

Design Measures. D and G optimality. The General Equivalence Theorem. Construction of Optimal Designs.

• Booklist

SAMPLE SURVEYS

Introduction to Finite Population Sampling

Random Samples. Stratified Random Sampling. Cluster and Multi-Stage Sampling

Design of Surveys and Estimation

Estimation of Totals and Proportions. Ratio and Regression Estimators. Choice of Sampling Method. Sample Sizes.

• Booklist

---

• Exercises

• M4S10 Enhanced Coursework

• Army Experiment
• Map Reading
• Map Reading Solution

• Main Effects and Interactions in the 2x2x2 Experiment
• Optimal Designs for Polynomial Regression in One Variable

• Old Exam Papers

---

Lynda White
Room 657
Email: l.white@imperial.ac.uk