3. Bivariate data

So far this week you have looked at ways to present multiple datasets about different populations. Recall from Week 2 that a population need not be a population of people; it is the word used to describe all members in a dataset. Investigating bivariate data, also known as paired data, is where we study two properties about the same population member, for instance, a town’s monthly rainfall and monthly sunshine over a given time period.

The ‘bi’ part of the word bivariate refers to two things, just as a bicycle as two wheels. The ‘variate’ part refers to a numerical quantity or variable. So bivariate means ‘involving or depending on two variables’ (OED, 2025) and the earliest use of it is from the 1920s.

In statistics, researchers are often trying to identify if there is a link between two sets of data or between two properties of the same object. However, even if a link is found, we cannot say if this link is causal i.e. that one variable is solely dependent on the other. How often do you see news reports claiming, ‘if you do A, B will occur’?! These headlines are often based on research looking at the relationship between different information in the form of bivariate data. What the researchers have usually found is there is some kind of relationship between A and B; this relationship is known as correlation, and you will look at this in more detail next. The key point to remember is that correlation is not causation.