This chapter will provide background to enable the reader to understand basic statistics and be able then to follow more complex statistical ideas. Although statistics is more than the mere analysis of data, it is a subject largely about data, so this will be discussed first. Data can be categorical or numerical, and in these two classifications there are various different types of data. This is the allocation of the individual to one of two categories. Often these relate to the presence or absence of some attribute. These data also have many other names such as binary, dichotomous and attribute data. Examples of such categorisations for patients include: ◆ Male/Female ◆ Smoker/Non-smoker ◆ Anaesthetist/Surgeon ◆ Married/Single. Each of these can be only be one or the other – they could be coded ‘1’ or ‘0’ to be binary (or on, off). For example male = 0, female = 1, or vice versa. Many classifications require more than two categories, such as: blood group, type of doctor, country of birth. Also the two categories, such as described previously, might be expanded into several categories. For example the married/single could be expanded to: married/single/divorced/separated/ widowed. This sort of data is called nominal data where there are several categories, but with no logical order. When there is a natural order (such as in seniority), the data are then called ordinal data. For example, anaesthetists could be divided into: ‘Foundation year 1’, ‘Foundation year 2’, ‘speciality doctor’, consultants’, ‘senior consultants’ and ‘clinical directors’. Ordinal data can be reduced to two categories, with possibly a considerable loss of information (e.g. ‘senior doctors’, ‘junior doctors’). Discrete numerical data are where the observation takes exact numerical values. Counts or events are discrete values. For example: number of children, number of ectopic beats in a time period and so on. Continuous (or analogue) data are usually obtained by some form of measurement. Examples are body temperature, blood pressure, height and weight. These values have an infinite number of possibilities, depending on the measurement interval, and variation. Although there are infinite possibilities, measurement systems usually round the continuous data up, or down, to discrete values. Blood pressure is often rounded up to the nearest 5 mmHg, for example.
Keywords: Welch test, arithmetic mean, attribute data, averages, binary data, categorical data, centiles, confidence intervals, contingency tables, correlation, data variability, diagnostic tests, dichotomous data, discrete data, frequency polygons, histograms, linear regression, null hypothesis, numerical data, ordinal data, outliers, probability
Oxford Scholarship Online requires a subscription or purchase to access the full text of books within the service. Public users can however freely search the site and view the abstracts and keywords for each book and chapter.
If you think you should have access to this title, please contact your librarian.