Statistical Design and Analysis of Experiments

• formulate an adequate experimental design for an experiment in the laboratory or under similar more or less controlled conditions.
• assess advantages and disadvantages of alternative experimental designs, for example with respect to randomization and/or blocking.
• formulate a correct mathematical model for data obtained froma given experimental design.
• test and estimate the effects in models for experiments with both deterministic and random effects.
• do detailed analyses of factorial effects by means of adequate contrasts and/or methods for grouping of mean responses.
• to construct experimental designs for factorial experiments with many factors each at (most often) two levels by means of block confounding and reduction through effect confounding.
• use special methods for statistical analysis of data from block confounded and/or reduced many-factor designs.
• use and carry out the statistical analysis of data from experiments with measurable sources of experimental noise (covariates).
• determine the necessary number of measurements in simple designs.

Main principles in experimental design: Randomisation and blocking. Balanced designs, square-designs, one or more blocking variables. Balanced incomplete block designs. Factorial designs, blocking in factorial designs, fractional factorial designs and confounding. Multiple regression analysis and analysis of covariance.

02402, or a similar course in elementary statistics

Lectures and tutorials.

The course is primarily aimed at students that are going to design experiments and/or analyse experimental data, but the general methods presented are also suited for sharpening the critical sense in experimental work, method development and empirical model building.