16 August 2013 | Pierre Jenkins, Head of InstantAtlas Support
InstantAtlas | 5 top tips for mastering legends
Those who have used maps in any meaningful form will appreciate the importance of the legend.
In interactive mapping visualisations such as InstantAtlas (IA) the humble legend gets a makeover. But just as with more traditional static maps, the importance of getting the legend right is crucial. This is well documented in the modern cartography literature; to quote one research paper, Smart Legend - Smart Atlas!. When done right, there is evidence that interactive legends, which combine the visual cues of the traditional legend and the use of interactive controls, perform favourably when evaluated alongside more standard widgets.
In InstantAtlas reports, the legend component allows the user to interpret the symbols and colours used in the map, to toggle map layers and to make on-the-fly changes to the report via the legend options.
Interactive legends give the user great flexibility in their data exploring but they can also lead to confusion if care is not taken. This may arise from the design of the legend appearance or controls, and the inclusion of options that are unsuitable for the data presented and/or the intended audience. Via our Desktop Designer tool there is a large degree of flexibility to configure the legend and below are my top 5 tips.
1 Choose an appropriate default legend type
The IA report legend component offers a choice of five legend types: equal interval, quantile, natural breaks, standard deviation and continuous. The first four group data into classes whereas the continuous type does not use class intervals.
Figure 1: an IA report that is using the continuous legend. The bar chart clearly shows how visual intensity is proportional to data intensity.
Being able to switch between these legend types is very useful for a user that understands the differences and is able to interpret these correctly. However, for other users, the choice may be bewildering and they are more likely to stick with the default type. The choice of the best default legend may be based on many factors, such as the nature of data in the report, the skill level of the end user, the need to be consistent with other content, etc. This discussion could form an entire blog post in itself so here I will reduce to one key point.
As mentioned in previous posts, I generally prefer to default to the continuous legend type. The argument in favour of class intervals is enhancing the map readability but I prefer the starting point to be an un-generalised view, and for a classification to be applied if necessary. Also, my experience is that many users find a representation where the visual intensity is directly proportional to the data intensity more intuitive than a classification algorithm.
You can set the default legend type by editing the Classification Type property in the Designer. For Flash reports this property belongs to the Legend component; for HTML reports it is a general property.
2 Choose an appropriate default palette
Appropriate use of colour has a big impact on the usefulness of a map. It is worth giving some thought to a report’s default palette, as casual users will stick with this despite having the option to change it. In IA we have included sequential, diverging and qualitative palettes for different types of data. The legend will automatically revert to a qualitative palette, termed a colour scheme in our software, for categorical data. In addition to changing the default palette, you can edit the colour ranges for any of the existing palettes or add your own. To this end, Colorbrewer is a well-known and robust tool for helping you choose appropriate colours for your maps and contains all the background information you should need.
You can set the default palette for a report by clicking the Map Palettes icon in the Designer toolbar and changing the Default Palette value. Customising the palettes is covered in section C.4.4 of the IA Desktop User Guide.
Figure 2: setting the default palette for an IA report using the Designer.
3 Choose an appropriate default point symbol size
If your base layer shows point locations rather than areas, the size of the point symbol can have a big impact on the legibility of the map. Sure, the user can change the symbol size using the legend but you can save them some work and set a default size that suits the distribution and density of your points.
Figure 3: an IA report with a point base geography.
For Flash reports you set the default point size by editing the Symbol Size property of the Legend component in the Designer. For HTML reports the change is not made with the Designer; instead you open the map.js file in a text editor and modify the symbolSize value for the base layer.
4 Consider fixing the legend
By default, InstantAtlas dynamically adjusts the legend for the data being displayed in the map at any point in time. As a result, it is not possible to make direct comparisons between different indicators and/or dates based purely on colour.
But you can overcome this. If having a legend that means the same thing across all data is important - if you are using the time animation feature to see change over time for example - you can "fix" the legend so that the same classes will apply across all data in the report. Note that this works best when all of the indicators in the report use the same units; it is unlikely to work as well if there is a mix of units e.g. counts, percentages, rates per 10,000.
For Flash reports you can set the min/max values by editing the Legend Minimum/Maximum properties of the Legend component in the Designer. You should also set the Classification Type property to equalInterval.
For HTML reports you can simply edit the general property called Legend Breaks and enter a semi-colon delimited list of class intervals e.g. 0;25;50;75;100 that will be applied globally.
Figure 4: setting global class breaks in the Designer for a HTML IA report.
5 Consider a custom legend
You can go a step further and specify your own class intervals, colours and labels on a per-theme or per-indicator basis, rather than use those dynamically generated by the report. This gives authors ultimate control over the map and allows them to exactly replicate legends used in other reports or publications if necessary.
This requires addition of appropriate metadata to the data file for the dynamic report – refer to section E.6.2.3 of the IA Desktop User Guide.
Figure 5: metadata in the Excel Data Manager to define a custom legend for Indicator 1. The legend will have three classes with labels of Low, Average/High and colours of red/yellow/green set by the hexadecimal codes.
 R. Sieber, C. Schmid S. Wiesmann. Smart Legend – Smart Atlas! Proceedings 22nd International Conference of the ICA, A Coruña, 2005
 N. H. Riche, B. Lee, and C. Plaisant. Understanding Interactive Legends: a Comparative Evaluation with Standard Widgets. Computer Graphics Forum, 29(3):1193-1202, 2010
 W. Tobler. Choropleth Maps Without Class Intervals? Geographical Analysis, 3, 262-265, 1973
 C. A. Brewer. Color Use Guidelines for Mapping and Visualization. Visualization in Modern Cartography, Chapter 7, pages 123-147, 1994