Introduction

A knowledge of statistics is required in designing studies for research, analysing the data collected in studies, and writing for publication – as well as for the critical assessment of published research work. Statistical techniques are also a key component of evidence-based medicine, used in systematic reviews and meta-analysis. This chapter reviews the application of statistical methods in:

• Describing data

• Collecting data

• Testing and interpreting data

Data description

Types of data

Data are obtained by recording measurements or observations, and can be classified into different types. The data may consist of numbers (numerical data) or group names/labels (categorical data). It is important to choose the appropriate statistical test to suit the type of data.

Numerical data

Obtained from measurements (e.g. height, weight) or counts (e.g. number of operations, number of children). Numerical data can be divided into continuous and discrete data:

• Continuous data – can take any value over the range measured, depending only on the accuracy of the measurement device.

• Discrete data – can only take whole-number or integer values.

Categorical data

Consists of group names or labels. Categorical data can be further divided into unordered and ordered data.

• Unordered – the data comes from groups which are mutually exclusive and not ordered in any way, e.g. blood group (A, B, O), gender (M, F).

• Ordered – the data groups are mutually exclusive and also ordered or ranked, e.g. socioeconomic class, staging of a disease.