Colors can be effective tool to help readers quickly discern the key insights in charts. Often, however, we simply use default colors or our company color scheme when building charts and don’t take the time to understand the impact of our selection. Here are a few simple rules you many want to consider when building your next chart:
- Don’t use colors when they are not necessary. Colors should only be used if they add information to the chart and not just for visual appeal. In the examples below, you’ll notice that colors in 1. add no information over 2. and could simply confuse the reader.
- Test your colors on your presentation medium. Colors can look different on different screens, projectors and printers. The safest method is to test that your colors work before presenting.
- Consider your audience when using colors. A surprising 7-10% of males suffer from Red-Green color blindness (the rate 0.5-1% in women). If an important member in your audience has some type of color blindness consider adding redundant labels to ensure they can effectively read the chart. Charts 1., 2., 3. highlight three ways that Aploris lets you label colors on a chart.
The science of choosing colors is quite detailed with a variety of studies. We are planning to tackle this issue in upcoming sessions.
Aploris charts default to the color scheme set in PowerPoint. However, colors can easily be edited and Aploris can also be set to use customized schemes.
The primary strength of the pie chart is the obvious message of the “part-to-whole” relationship in visually appealing way that can be appreciated by all (see Choosing the right chart type). Bar charts with stacking and/or proper labeling can also achieve this message but in a slightly less intuitive manner. Beyond this small advantage, however, it’s often argued that pie charts are the least effective form of graphs (e.g. on Wikipedia or in Edward Tufte’s “The Visual Display of Quantitative Information”).
The size of elements in pie chart can be fairly easily estimated when the element begins at 0, 90, 180, or 270 degrees. If, however, it starts at a different angle the size of the element becomes much more difficult to estimate. In the first example below, the blue element is easily approximated at 25%. In the second example, when the chart has been rotated, the size of the blue element is more difficult to estimate.
The eye can quickly compare differences location and line length (as used in bar charts) but has difficulty discerning differences in angles and areas (as used in pie charts). In the first three examples below, you can quickly arrange the elements from largest to smallest. In the third example, however, it takes a little longer to ascertain that that D is larger than B.
Furthermore, it’s easier to extend bar charts into Mekkos where you can quickly compare the width of each bar. To extend the pie chart you have to create multiple pie charts where the total area or diameter is used to compare data sets; a comparison that is not always easy to make. In the first example below, you can quickly discern that the bar W is about 5 times wider that than bar Q. The second example, however, it is much more difficult to determine that pie W is 5 times the area of pie Q.
Pie charts can be used to effectively display very simple data in a visually appealing style. However, if the data has any complexity at all, forgo the stylistic points and opt for the more effective bar graph.
This posts concludes the previous post on choosing a chart type and describes the categories relationship/correlation and frequency/ranking.
A relationship between two dimensions is often displayed with a scatter chart. For instance sales growth of products can be plotted in relation to their profitability. A bubble charts adds a third dimension to the visual data. So for the sales numbers the size of each bubble could represent last year’s sales volume.
In some scenarios a correlation between two dimensions may be expected that would be visible in the scatter or bubble chart. Imagine that management realigned sales incentives to focus on the most profitable products. Hopefully, this results in a positive correlation between sales growth and product profitability. For this application some charting tools allow you to automatically insert a trend line for subsets or all data points.
Scatter chart sample: Relationship GDP/Olympic medal count
Another chart type that may be useful to relate two dimensions is a bar-mekko. Effectively, this is a column chart with variable column widths. Or a Marimekko without a 100% axis to put it another way. This could be another way display the sales data with the column width depicting revenue and its height profitability. For a limited of categories, say products in this case, this type may be easier to read than a scatter chart. At the same time it describes a composition, e.g. how the company sales are split between products.
Frequency or ranking
This category is for you if data needs to be presented for multiple items but it is not desirable or possible to combine the numbers to a composite value. A typical application would be a histogram or a distribution curve if displaying many data points. Opposed to a composition chart, median or quartile information can easily be highlighted. This again shows that it’s not primarily about the data you want to show but the point that you want to make.
If columns are ordered by their frequency values the result will be a ranking chart. Values will be displayed from highest to lowest or vice versa. To visually display which are the “top” values it may be a good idea to rotate the chart to a bar charts with the best items standing literally at the top. Of course, frequency can be substituted by any other value describing the quality of an item.
Work plans and timelines
For building out timelines, building schedules, or understanding dependencies in a workplan, the Gantt chart offers a valuable visualization. This article provides a guide to Gantt charts.
Charting tools typical offer a number of different types of charts. You may see bar or columns charts, pie charts and Marimekkos. So which chart type would I use to display certain data?
As usual the answer is: it depends. In this case on the message that you would like the slide to tell. Imagine having sales data at hand. You might want to display how the sales revenue developed over time and is projected to grow. It may also be meaningful to analyze the distribution between multiple regions or division. Or you might want to know how the sales volume of multiple products relates to their profitability.
The outcome of your analyses usually falls in a certain category. For a certain category some chart types may be better suited than others. In this blog post and the next we are going to describe the categories comparison, composition, relationship/correlation and frequency/ranking.
Comparison, often a development over time
To see how numbers develop over time typically column and line charts are most powerful. You may add visual elements to highlight or summarize overall growth rates. Tabular data can be used to indicate growth rates for multiple lines or segments in stacked columns.
Stacked bar chart sample: US food consumption
For a comparison of multiple items along several criteria spider charts or line charts may be helpful. Use a rotated line chart with lines running top down to visualize a tabular comparison of multiple items.
Line chart template comparing items on multiple criteria
Composition, sometimes comparing between instances
Composite numbers may tell you how sales are distributed between several divisions. The simplest way to display a 100% composition is to use a pie chart. A single 100% stacked column carries the same information and provides a natural way to add the total number that equals 100%. Showing multiple columns adds a second dimension that allows to compare compositions over time or between several entities. If many periods are displayed it may make sense to use an area chart instead of many stacked columns.
A special chart explaining a composite value is a waterfall chart. Unlike the aforementioned chart types it is capable of showing positive as well as negative values incurring into a composite number. A waterfall chart is especially powerful to display how an input value (e.g. last year’s sales) transforms into a new value (current sales plan) taking into account multiple influencing values (delisted products, new products, additional sales channel).
For breaking down a number into two dimensions a Marimekko can be a good choice. Think about sales figures split across geographies and by product line within each geography.