Introduction & Aims
About the Program
'Experimental Design' is structured around a menu, from which sections exploring the key issues in experimental design can be accessed. It is designed to be used in a linear fashion but each option can be accessed independently and in any order.
The program also includes a case study and self-assessment section that you can use to test your understanding of the program's content.
The program focuses on:
There are further issues that, although outside the remit of this program, you must keep in mind when you approach a new research project:
This program is aimed at all research scientists using experimental animals, but the principles of experimental design are applicable to most areas of biological and medical research.
Some knowledge of statistics is assumed, but even if you have no knowledge you should find the program useful.
Aims and Objectives
For ethical, economic and scientific reasons it is important that research scientists use experimental animals efficiently. Good experimental design is fundamental to this. This interactive program uses examples relevant to the pharmaceutical industry and the academic scientific community to teach good experimental design. In using this software package we hope you will gain further understanding of how to:
We hope that the 'Experimental Design' program will enable you to:
However, in most cases your use of statistics and therefore the quality of your research will benefit greatly from consulting a professional statistician.
Why do we need to improve experimental design?
Surprisingly, the principles of experimental design are rarely taught to aspiring research scientists. Statistics courses usually focus on methods of analysis once data has been collected. The result is that some experiments give misleading results, and many could be improved so as to give more reliable information.
Two studies of research experiments on animals involving 133 and 33 published academic papers respectively, found some startling results. See if you can guess what was discovered about these particular papers.
How many experiments had deficiencies in design?
What proportion of experiments were around twice as big as need be?
What percentage of experiments used incorrect statistical analysis?
OK, things may not be as bad as they seem. But those third of experiments which were twice as large as they needed to be could represent between 50 and 100% more animals being used than necessary. There certainly seems to be scope for improving the statistical analysis of such experiments.