There are three types of experiment:
- Ones that yield interesting data, no matter the outcome
- Ones that yield interesting data, but only if it turns out one way
- Ones that yields ambiguous, uninterpretable results regardless of the outcome
The next few pages will lead you step-by-step through the important questions and decision points involved in designing an experiment so that your experiments don’t fall into category 3!
My goal is to instil a series of good practices, shaping the way you think about all your future experiments, so that you take the concepts forward into all the rest of your work in whatever discipline.
Whenever you embark on a new series of experiments it is worth taking time to plan things fully before jumping in. Bad design costs time and money and, in the case of animal or human research is ethically unforgivable! Well designed experiments generally require simpler statistical methods to extract useful information. So time spent at the beginning will ultimately be a solid investment in the longer term.
Before we get going, it’s important to stress that you should go into experimental design with a can do attitude. Experimental design if ~50% about your understanding of the biology of the system, ~40% common sense and 10% an appreciation of statistical approaches and, of course, you can get help!
Downloadable experimental design sheet
To help you to identify the correct things to think about for your experiment I have made downloadable word document file that you can fill in as your progress through these pages (or use a general resource in the future). Available here: Experimental Design Checklist
OK, ready to start?
These pages are designed as a step by step guide to teach you what you should be thinking about at each stage of your experimental design. They are based around lectures that I have given, but presented this way so that you can take your time and go at your own pace. I suggest that while you are reading you also think about your own experiments and how the info relates to your studies. That way the messages will be more firmly absorbed.
If you are returning to these pages as reference material, you can jump straight to the part you want via the links below.
- Preliminaries (2 mins)
- Choice of model system (1 min)
- Choosing between Manipulative and Correlative Studies (6 mins)
- Data collection, variables and measurements (6 mins)
- Controls (10 mins)
- Sample selection, randomisation and sample size calculations (10 mins)
- Finishing up (1 min)
Times are indicative from http://niram.org/read/ not sure how accurate they are… would be interested to do an experiment to find out!