SQLServerCentral Editorial

When to Use Zero

,

I'm not great at building charts and graphs. I can build a basic chart, but I often depend on the tooling I use to size, scale, etc. appropriately for whatever I'm graphing. That, or I just use a basic graph that starts from zero and has some sort of linear scale.

There are plenty of misleading charts, especially used by the media that want to show some particular aspect of data that suits the story they are reporting. Many of these misleading charts often don't start at zero, and they end up scaling in a way that can confused people.

You wouldn't think it was hard to decide whether or not to scale a graph of data from zero or not, but it can be complicated, as this post shows. There is a flowchart to help you decide when to include zero, or even when to have an inset chart that better explains the data. As many of us know, it can be easy to misinterpret data, especially when someone else is deciding what to show.

In this case, the post talks about some examples of what the data doesn't graph well because the scale is too large and the range of data too small. Or when the scale distorts the relative size of two values in a graph. In all of the examples, it makes perfect sense why you do or do not include zero on your axis, but I don't know that I know when I should or shouldn't do this.

Part of the problem for me is that people often glance at visualizations and charts without spending enough time to really study them. That's part of the idea of a visual, in that we can get information quickly from a picture, as opposed to a chart of numbers. However, it is still easy to look at a broken scale or inset chart and not spend the time to comprehend that one value isn't twice another, but that the chart is zoomed in because the scale changed.

To me, a lot of misunderstanding gets cleared up when we discuss the chart and the data in a group. Often one person will realize when others are not reading the scale when drawing their conclusions and remind others that Canada isn't twice as large as Brazil. That works if people speak up and if others listen. That isn't always the case, especially when the boss is making a mistake. The other issue is that many of us might look at a chart by ourselves. and we are unlikely to tell ourselves we are misreading the values.

There isn't a good way to ensure people read a scale and factor that into any decision they make regarding the chart. I do always like to include some data with a visual, that way I can see raw numbers on the same report, or by drilling in. The combination works well for me, but I'm a geeky, numbers person. I like seeing data, which is why I like working with databases.

Rate

5 (1)

Share

Share

Rate

5 (1)