What it is
A scatter plot places each observation at the point given by two numeric values: one along the x axis, one along the y axis. Patterns in the cloud of dots reveal relationships — a rising band means the measures increase together, a falling one means one rises as the other falls, and no pattern means no relationship.
It also shows what a summary number hides: clusters, outliers and curves.
✓ Use it when
- You have two numeric columns and want to know whether they are related.
- You want to spot outliers or groups.
- You have many observations — scatter plots handle hundreds of points.
✗ Avoid it when
- One of the measures is time and you care about the sequence — use a line chart.
- You have only a few points with names — a labelled bar chart may communicate better.
Tips for a better chart
- Put the measure you think of as the cause on the x axis.
- Don’t force axes to start at zero; let them fit the data.
- Use translucent dots when points overlap, so dense areas look darker.
- Remember that correlation is not causation.
The maths
The strength of a straight-line relationship is measured by Pearson’s correlation coefficient, r = Σ(xi − x̄)(yi − ȳ) / √[Σ(xi − x̄)² · Σ(yi − ȳ)²], which runs from −1 (perfect falling line) through 0 to +1 (perfect rising line).
Anscombe’s quartet shows why the plot matters: four datasets with the same means, variances, correlation and regression line look completely different when drawn (Anscombe 1973).
A little history
In 1833 the astronomer John Herschel plotted the positions of double stars against time to estimate their orbits, an early scatterplot with a curve fitted by hand; Friendly and Denis trace the form’s origins to it (Friendly & Denis 2005).
Francis Galton made the scatter plot the tool for studying relationships when he plotted parents’ and children’s heights and discovered regression towards the mean (Galton 1886). Karl Pearson’s correlation coefficient followed within a decade.
Questions
Can I name the dots?
Yes. If your table has a column of names, choose it under “Point labels”.
Can I show more than one series?
Yes, up to three, each in its own colour. Beyond that the colours become hard to tell apart.
References
- Friendly, M. & Denis, D. (2005). The early origins and development of the scatterplot. Journal of the History of the Behavioral Sciences 41(2), 103–130.
- Galton, F. (1886). Regression towards mediocrity in hereditary stature. Journal of the Anthropological Institute of Great Britain and Ireland 15, 246–263.
- Anscombe, F. J. (1973). Graphs in statistical analysis. The American Statistician 27(1), 17–21.