We have talked before about how the absence of evidence is not evidence of absence and how to plan sample sizes. This short video by Matt Asher, from the University of Toronto, takes an amusing look at power and sample size. As you can tell, it is written by a frustrated statistician!
Core Samples by Adam Winsor, Attribution-NonCommercial License
One of the biggest influences on the conclusions you can draw from your research, or anybody else’s, is the source and size of the sample used. The sample refers to the source of the data. It might be the animals you collected data from, participants in surveys or interviews, or even documents you analysed (for example, previous literature).
We will talk about the source of your sample another time, but for now lets just think about the size of the sample. When you are planning your research you will need to calculate the sample size you need in order to determine if your plans are feasible and will give you useful data. In this previous post I have given you information about how to do the calculations, but before you can do them you need to collect up some information. In this post, Karen Grace-Martin from the Analysis Factor shares a list of the information you need.
http://www.theanalysisfactor.com/5-steps-for-calculating-sample-size/
Red Dot by Sarah Laval Attribution-NonCommercial License
If you are planning a piece of experimental work, one of the first tasks is to figure out if the project is feasible. To do this you need an idea of the numbers of samples you need for the study. The cost of research is often proportional to the numbers you need to study, and the time it takes to accomplish the research can also depend on the numbers involved, especially if you have to wait for cases.
How do you work this out? You can get an idea from your reading of similar types of research, but in the end you need to work it out exactly as this is how you will justify you project to your supervisor, to funding bodies and to an ethics committee. What you need to show is how many samples will be needed to have a reasonable expectation that the study will give a useful result. Too few samples and the whole project is a waste of time, money and may cause harm for no benefit. Too many samples is not usually the problem, but is also equally costly in terms of money, time and ethics.
How many samples you need depends on how big a difference you are looking for and how much natural variation you can expect in the populations you are studying. This article by Martin Bland explains different methods of establishing the sample sizes for studies.
Bland M (2009) The tyranny of power: is there a better way to calculate sample size? British Medical Journal 339:b3985 http://www.bmj.com/content/339/bmj.b3985
MVM learning 