Skip to Main Content

Data Visualization

Email this link:

Data Curation & Management / Emerging Technologies Librarian

Profile Photo
Danielle Kane
Computational Research Librarian Geographic Information Systems (GIS)
Office: Science Library 226
Phone: 949-824-2024

Top Ten Dos and Don'ts for Charts and Graphs


1. Do use the full axis.

Avoid distortion.

For bar charts, the numerical axis (often the y axis) must start at zero.  Our eyes are very sensitive to the area of bars, and we draw inaccurate conclusions when those bars are truncated.  See the difference between the original media chart and an un-truncated chart as generated by FlowingData.

(But for line graphs, it may be okay to truncate the y axis. |

(But for line graphs, it may be okay to truncate the y axis. |

Wide ranges:

If you have one or two very tall bars, you might consider using multiple charts to show both the full scale and a "zoomed in" view - also called a Panel Chart.


Consistent intervals:

Finally, using the full axis also means that you should not skip values when you have numerical data.  See the charts below that have an axis with dates.  The trend is distorted if you do not have even intervals between your dates.  Make sure your spreadsheet has a data point for every date at a consistent interval, even if that data point is zero.  (Original chart on left from Naomi Robbins at Forbes.)


2. Do simplify less important information.

Chart elements like gridlines, axis labels, colors, etc. can all be simplified to highlight what is most important/relevant/interesting.  You may be able to eliminate gridlines (Cole Nussbaumer, gridlines are gratuitous) or reserve colors for isolating individual data series and not for differentiating between all of the series being presented (Gregor Aisch, Doing the Line Charts Right).


3. Do be creative with your legends and labels.



4. Do pass the squint test.

"When you squint at your page, so that you cannot read any of the text, do you still 'get' something about the page?"

(Andrew Abela, The Squint Test: Creating Simplicity of Design and Complexity of Data)                                                               

  • Which elements draw the most attention?  What color pops out?
  • Do the elements balance? Is there a clear organization?
  • Do contrast, grouping, and alignment serve the function of the chart?                        

(The XLCubed Blog, The Dashboard Squint Test)

Related: Projectors often wash out figures. The squint test can simulate this. Try high contrast designs with clear trends. (James Davenport, The Chart that Wasn't There: Avoiding Disappearing Plots in Presentations)


5. Do ask others for opinions.

Even if you don’t run a full usability test for your charts, have a fresh set of eyes look at what you’ve done and give you feedback. You may be surprised by what is confusing – or enlightening! – to others.




1. Don't use 3D or blow apart effects.

Studies show that 3D effects reduce comprehension. Blow apart effects likewise make it hard to compare elements and judge areas.

(Nathan Yau, World Happiness Report makes statisticians unhappy; Naomi Robbins, Trellis Plot Alternative to Three-Dimensional Bar Charts)


2. Don't use more than (about) six colors.

Using color categories that are relatively universal makes it easier to see differences between colors.

The more colors you need (that is, the more categories you try to visualize at once), the harder it is to do this.

(Ware, Colin. Information Visualization: Perception for Design (3rd Edition). Morgan Kaufmann, p. 132.)


But different colors should be used for different categories (e.g., male/female, types of fruit), not different values in a range (e.g., age, temperature).

Borland, D., & Taylor II, R. M. (2007). Rainbow color map (still) considered harmful. IEEE Computer Graphics and Applications, 27(2), 14-17.

So, no rainbows! We often think that the order of colors in our "rainbow" is easy for everyone to understand, but this order is not universal and will make charts and maps harder to read.

If you want color to show a numerical value, use a range that goes from white to a highly saturated color in one of the universal color categories.

Krzywinski, M., Brol, I., Jones, S., & Marra, M. (2012). Getting into visualization of large biological data sets: 20 imperatives of information design. Poster presented at 2nd IEEE Symposium on Biological Data Visualization (BioVis 2012), Seattle, WA.


And remember, some people have color blindness.

Use Vischeck to test your images.

(Naomi Robbins, Choosing Colors for Graphs that are Accessible to Most Viewers)


Also, print out your charts to test what it looks like in gray scale. (For grayscale to work, you need to vary both hue and saturation.)


(Nathan Yau, Incredibly divided nation in a map)



Additional color resources:


3. Don't change (style) boats midstream.

One of the easiest ways to get the most out of charts is to rely on comparison to do the heavy lifting.

Our visual system can detect anomalies in patterns. Try keeping the form of a chart consistent across a series so differences from one chart to another will pop out.

Use the same colors, axes, labels, etc. across multiple charts.

(Jim Vallandigham, Small Multiples with Details on Demand)


4. Don't make users do "visual math." 

If the chart makes it hard to understand an important relationship between variables, do the extra calculation and visualize that as well.

This includes using pie charts with wedges that are too similar to each other, or bubble charts with bubbles that are too similar to each other.  Our visual processing system is not well suited to comparing these types of visual areas. 

We are also not good at holding precise visual imagery in our memory and comparing it to new stimuli; if you are giving a presentation and want the audience to be able to compare two charts, they need to be on the same slide.

(Andy Kriebel, Stacked Area Cart vs. Line Chart - The Great Debate; Emil Johansson, Character Dialog in the Hobbit: An Unexpected Journey measured)

5. Don't overload the chart.

Adding too much information to a single chart eliminates the advantages of processing data visually; we have to read every element one by one!

Try changing chart types, removing or splitting up data points, simplifying colors or positions, etc.

(Kaiser Fung, Ruining the Cake with Too Much Icing; Cole Nussbaumer, Death to Pie Charts)


Many thanks to Angela Zoss at the Duke University Libraries for allowing us to build off of and customize her Introduction to Data Visualization Research Guide.