# Line graph/chart - straight or smooth curves?

I'm working on dashboard and researching different kind of graphs that could be used.

I couldn't find any information about usability/UX differences between line charts that are using straight (A) or smoothed curves (B).

I believe that straight angles give a better understanding of relatively small ups and downs on the chart but I'd love to be referred to relevant researches or to hear your thoughts.

• It probably depends on the data and / or use case. For instance if you're showing how often an blog article gets shared across the day then it's likely it can be 'looser' than if you're showing stock value figure fluctuations.
– JonW
Commented Dec 22, 2014 at 17:28

I would say in this case usability should take a back seat to accuracy. Graph B implies a polynomial function of sorts that you have sampled at specific locations. If that is not the case, you should always go with A.

For a great comparison of the two, see: http://vizwiz.blogspot.com/2011/12/when-you-use-smoothed-line-chart-your.html

• Yes, my graphs examples are not representing the same data. My gut was saying to go with A but I see that many successful product like Crashlytics are using the B. Commented Dec 22, 2014 at 17:35

It depends on the specific use case. If it's important to display the exact values, then the straight lines do a much better job. But often you'll find that what matters is the trend and the precise measurements are secondary. In that case a smooth line is much clearer. To figure out the trend, your brain basically does the smoothing itself, it's trying to zoom out and see where the line is going. You can save the user from making this cognitive effort by doing it for him/her.

One criticism I've heard in regards to line graphs is that they can potentially suggest a degree of specificity in the data that is not actually there. Your question actually points to this as well. You can't decide whether to use a straight line or a curved line between data points because you have no data to tell you what the shape of the line should be. A straight line suggests that the value between the two measurements increased linearly, while a curved line suggests otherwise.

No matter what you choose, the shape of the line is going to tell a story to the viewer. What story are you trying to tell?

I usually prefer to use a bar chart in these situations to make it clear that the rate of change between data points is not known.