Whether you are undertaking qualitative or quantitative research as part of your PhD, and whether your study is on communities, people, rats or test tubes, you must justify the size of the sample you intend to use. Epidemiologists love breaking things into two: for example, dead or alive, smoker or non-smoker. So following on from this tradition, studies can also be broadly broken down into two types, descriptive and analytic.

With descriptive studies, we are simply trying to describe something about a group of people. Examples of descriptive studies include: Surveys, Case series, and Qualitative research. Of the descriptive study designs, it is only surveys that need concern us with respect to sample size. Sample sizes for case series or qualitative research are usually based on the need to obtain saturation, that is further interviews would provide no additional new information.

Analytic studies are different in that they usually involve hypothesis testing. Examples of analytic studies include: Randomised Controlled Trials (RCT), cohort studies, case-control studies and cross-sectional studies. Sometimes there is a bit of overlap. For example, you might want to compare groups as part of a survey. However, the primary objective is still descriptive, and the analytic part a secondary objective, or hypothesis generating exercise. Similarly, when reporting the results of an RCT, you still need to describe the subjects in each study arm.

Topic Objectives

On completion of this topic students should be able to:

1. Appreciate the different requirements for sample sizes for descriptive and analytic studies
2. For descriptive studies, be able to calculate sample sizes based on desired accuracy
3. For analytic studies, be able to calculates sample sizes based on power
4. Have an understanding of available software and websites for sample size calculations, and when to see further help