Can I visualize a bubble with value 0?

Question

Background:

I got a few values on the scale 0..1 and -1..1 that we have been asked to visualize as bubbles. The value 1.0 would be the largest sized bubble and 0.5 would half the size, and so on. Negativity would be represented by a color, say blue for positive, and red for negative.

Now, for "How to visualize a bubble with value 0?" we've been suggested "a small purple bubble". To me, it's counterintuitive -- if the value directly map to the area -- that a bubble of value 0 would be at all visible.

I know these values can also be visualized as bar charts on a y axis spanning both negative and positive. But who doesn't like bubbles, and I would like to explore any examples or standard ways of visualizing small values with area.

Questions:

To make this question less diffuse, here are my actual questions:

• Any way of showing 0 area?
• Would it be acceptable to show 0 area as small bubble, even though a small bubble actually has a non-zero area? Or would that metaphor impose problems somewhere?
• Are there any standard bubble features we could use to make small values more tangible?

Answer 1

Any way of showing 0 area?

Sure. That’s no bubble at all.

Would it be acceptable to show 0 area as small bubble, even though a small bubble actually has a non-zero area?

I suppose if the bubble were small enough to be regarded as a point, then some users might take it as zero. Probably would help if it has no hue (gray or black), rather that making it a mix of negative and positive hues. A mix of hues is what you get if bubbles overlap.

However, such a small dot would be easy to miss and hard to detect the color. Some users may think it’s just slightly negative or slightly positive, which may have significance depending on the domain this is for. I wouldn’t take this route.

Are there any standard bubble features we could use to make small values more tangible?

Not a standard, but give every bubble a cross of the same size. Scale the cross so the ends correspond to a significant threshold value (could be |x| = 0.5).

By doing this:

• A zero value is a cross with no bubble.

• Zero and near-zero values are relatively easy to detect.

• Users can more easily judge how close a data point is to a significant value.

• Users can more easily see where the bubble is centered along x and y coordinates, which itself often has meaning in a bubble graphic.

Who doesn't like bubbles?

Bubbles have the serious drawbacks. It often unclear if the diameter or the area represents the underlying value. It’s very hard for humans to judge relative amounts of area. Mostly they underestimate: When Bubble A is twice the area of Bubble B, human will tend to think it is less than twice the area. Your users will have a hard time interpreting absolute values represented in the bubbles and comparing differences among bubbles. Don’t use bubbles or other area coding unless you have good reason:

• You’re going to plot the bubbles on a map or grid to show other information.

• You’re showing one variable in the width of the bubble and one in the height of variable (the “bubbles” are ellipses or rectangles), which when multiplied have meaning to users (e.g., probability times severity to yield risk).

“Bubbles are cool and trendy” is not a very good reason for most situations.