# Why Line Chart Baselines Can Start at Non-Zero

There is a recurring argument that line chart baselines must start at zero, because anything else would be misleading, dishonest, and an insult to all that is good in the world. The critique is misguided.

Line chart baselines do not have to start at zero.

Bar charts must start at zero because of geometry. The lengths of the bars directly represent values, so a bar that is twice the length of another means a value is two times greater than another. On the other hand, line charts use slope and position to make comparisons, so you don’t break the chart by truncating the axis above zero.

For example, the simple line chart below shows the percentage of shots that were three-pointers in the NBA, during the 2010-11 to 2021-22 seasons. Use the slider to shift the baseline and change the steepness of the trend line.

The rate of three-pointers increased from 22% to 40% during the time span, which you see by reading the tick marks on the y-axis and looking at how the line fits in that context.

Decrease the baseline towards zero, and the change looks less dramatic. Increase the baseline beyond 22% and the data is off the chart. The stretch of the trend line changes, but the geometry still works.

Some say it’s misleading to only show the range of the data and that you should show the range of all possible values. In this case, that would mean a y-axis that shows 0% to 100%. Try the adjustment below.

Shifting to 100% flattens the line more — even though the three-pointer rate nearly doubled. Does that mean the absolute difference, 18 percentage points, between 2010 and 2021 is insignificant? The answer is based on the context of the data. If you’ve watched basketball over the past decade, you know the increase in three-pointers is very real, sometimes to the detriment of watchability.

But if you never watch basketball and know nothing about the sport, then the data doesn’t mean anything. You must rely on the chart (and the person who made the chart) to communicate the weight and the context of the change.

The weight in a line chart — the significance of some unit of change, large or small — is reflected in the aspect ratio of the chart. Usually when people criticize y-axis limits on a line chart, they’re actually referring to the aspect ratio. A tiny change can look huge if you stretch out the chart enough. Try it below.

People also like to criticize the range on the x-axis. All the charts so far show 2010 to 2021, but you can go further back in time. The NBA’s first season was in 1947.

The haters’ claim is that you have to provide data until the beginning of time for full context. This is partially true. Maybe you’re talking about historical records, in which case it makes sense to compare now to all of history. But if you’re talking about a year-over-year change or comparing against a useful marker in time, then you don’t need to include everything. Sometimes you just want to know what’s been going on the past few years.

Well-meaning critiques treat visualization as a purely technical task. Implementation is of course part of the work. There is code, and the code creates definitions and sets of rules.

However, there are many ways to show even the simplest of datasets, so to avoid the infinite abyss of choices, we ask questions and use charts to answer. The questions we ask and what we want to show define the axis limits, the range of data to apply, the aspect ratios, and design decisions to highlight the interesting parts. The following charts show how questions change the choices.

When reading or using visualization, not just line charts, think about the questions and answers. Think about the data and how it dictates the type of chart to use. Sometimes that includes a zero-baseline. Sometimes not.

The slider-heavy format was inspired by Bartosz Ciechanowski articles. I made the charts with D3.js. Find out more about the rise of the three-pointer and death of the mid-range. Bar chart baselines still must start at zero.

Become a member. Support an independent site. Make great charts.

My work on FlowingData is supported by paid memberships. Since 2007, I have been analyzing data and making charts to help people understand and appreciate data in their work and everyday lives. I hope to keep it going for many more years.

If you liked this (or want to learn how to make similar data things), please consider supporting this small corner of the internet with a membership. You get unlimited access to visualization courses, tutorials, and extra resources. Thanks. — Nathan