Let’s just look at some examples of using planar and retinal variables. I’ll be adding more as we go along.

Planar variables are x and y, yes, but you can make use of the y axis without there being two planar variables. A bar graph is measured on the y axis, but that’s only because it has size/length.

First, a simple example of a planar variable without any retinal variables. Each mark shows up on the plane in accordance with the year it’s associated with.

Boring, right? Let’s add another variable: the number of ducks in each of those years.

We could use another planar variable - y - but instead we’re going to represent the number of ducks with a retinal variable. First we’ll use brightness/value (…don’t harp about that nomenclature just yet). More ducks = darker green.

We could also use size in the form of area, which generally means circles. More ducks = larger circle.

We could have also used size/length instead of size/area, which would use lines (…thick lines, which we tend to call bars). More ducks = longer lines/bars.

Not that these are not on 2 planes. The number of planes refer to how the marks are positioned - all of our bar chart marks are positioned against the bottom of the x-axis, and it’s only their length that changes according to the number of ducks, not their position.

We can add more information to these by upping the number of variables to three - adding color/hue to the bars seems like a good choice.

If we went back to size/area instead of size/length, we could do the exact same thing while taking up much less space.

The only problem being that according to Cleveland area is much less accurate than area - you need to decide what’s important to you!

If we want to look at two planes, we need to position our marks according to both the x and y axes. If you have a flat x axis, this usually looks like a scatterplot.

You can then drop give the marks retinal variables to show more information. Some are easily to notice than others!

For example, shape…

…versus color…

Along with color, we could also bring back size to describe the number of ponds visited by each president to court the duck vote. In this one I used size/repetition instead of size/area because I wanted to mix things up a bit.

We could also try it with brightness/value. I think size/repetition looks a bit better than brightness/value in this case, but with many more data points I bet size/reptition might look too cluttered.

Backing up a bit, you can sometimes use fewer retinal variables to get across more information. For example, the graphic below is just two planar variables, but the opacity can show you the density of different sections (if you had a ton of dots this would look a lot more interesting).