|
|
Experimental design: introduction
Introduction
In this workshop we will examine some of the issues surrounding the planning of
experiments, particularly in computer science. We will do this by examining a
poorly-executed experiment, and attempting to tease out the factors that
contribute to its failure. This will lead to suggestions for improving the
experiment, and development of a framework for experimental methods.
At the end of this workshop you should be able to
- Recognize weaknesses in experimental methodology
- Describe the adverse effects of inadequate sample size, bias, non-representative
sampling, and premature data collection on experimental results
- Plan an experiment in such a way as to minimize these problems
- Recognize situations where the advice of a statistian should be sought
This workshop is not about statistics; it is about understanding the
basic principles upon which statistical methods rest. Students
often believe that the mathematical manipulations of statistics are difficult
and frightening; in fact the real difficulty is planning the experiment in such
a way that statistical techniques will even be appropriate. I have tried also
to minimize the amount of statistical jargon used in the text, on the basis
that the important ideas can be communicated without any.
|
