Types of Data and Their Presentation

Types of Data and Their Presentation

1. Why the data type comes first

• The first step in any analysis — before any calculation, summary, or plot — is to decide the type of each variable, because the data type dictates which summary statistic is legitimate and which display is appropriate; choosing the wrong display for the data type is a primary error in biomedical reporting (Swinscow & Campbell 10e Ch.1, p.1).
• A variable is any characteristic that varies from one member of a population/sample to another (e.g. age, sex, blood concentration, diagnostic category); the recorded value of a variable for an individual is an observation (Swinscow & Campbell 10e Ch.1, p.1).
• The data type determines three downstream decisions at once: (i) the measure of central tendency (mean vs median vs mode), (ii) the measure of variation (SD vs interquartile range vs range), and (iii) the tabular/graphical method (bar diagram vs histogram vs pie chart vs scatter) (Swinscow & Campbell 10e Ch.1, pp.1–2).
• Postgraduate framing: the data type is a property of the measurement scale, not of the phenomenon — e.g. blood pressure measured in mmHg is continuous, but the same construct dichotomised at a threshold ("hypertensive yes/no") becomes binary; reducing a continuous variable to categories discards information and is justified only when clinically meaningful (Swinscow & Campbell 10e Ch.1, p.2; Ch.2, p.19 on transformation/categorisation).