A simple scatter plot and a triple scatter plot

To the left, a regular scatter plot (Cartesian plot) and to the right a triple scatter plot. Reproduced from Cleveland, W.S. & Mcgill, R. 1984. Graphical Perception - Theory, Experimentation, and Application to the Development of Graphical Methods. Journal of the American Statistical Association, 79, (387) 531-554.

The scatter plots show the relationship between two random variables X and Y. The simple or regular scatter plot is shown on the left, and a variation of the scatter plot known as the triple scatter plot is shown on the right. Some overlapping dots or symbols may not be visible on scatter plots e.g. with solid-filled dots or symbols on the left. The scatter plot on the right tackles this issue using empty circles to make overlapping symbols more visible. Circle is a good choice since overlapping circles form shapes that are not circles, and therefore can be easily distinguished (see Cleveland 1984).A simple scatter plot cannot show the frequency of occurrences for data points that have the same values, as they would appear as one dot. The triple scatter plot has been proposed to circumvent this issue by scaling the size of the dots or symbols according to the frequency, for example. However, the triple scatter plot requires users to judge areas, and thus shares some features seen with an area graph. A triple scatter plot could also introduce cognitive burden to the users having to estimate the centre of the circles, and therefore compromising on accuracy. This weakness may be generally true for simple scatter plot with very large or irregular shaped symbols.