Card Sorting – Introduction
Note that if you want someone to conduct IA or card-sort-based research and design, visit our Information Architecture page. Otherwise, read below about obtaining and using our tools yourself.
On this page:
Card sorting is a powerful technique for assessing how people group related concepts together. It can be much more effective than usability testing in determining how effective web site navigation is, but it also has applications in many other aspects of design and research.
However, card sorting has its drawbacks - many of which we have addressed in the approach and tools described here:
While preparation and data capture can be simplified by using screen-based sorting tools, our approach allows a wide range of approaches and analyses. The items being sorted can be:
while the methods of data collection are also much wider:
Our processing software, SynCaps, provides highly effective visual analyses and exports spreadsheet files to allow bespoke investigations using off-the-shelf spreadsheet packages such as Microsoft Excel.
Many practitioners will argue that face-to-face paper/card-based sorting will give the best results. In fact, that is why we originally developed this approach, since it allows paper, card or adhesive labels to be prepared using only Microsoft Word and a printer (in addition to the material to be printed).
There are some real benefits to a paper-based approach: Participants are less intimidated by paper and pens, plus they can provide much more in the way of qualitative data by making notes on the cards, changing item names, adding new items or groups and so on.
Here are the steps for CAPS (Computer-Aided Paper Sorting):
To prepare the cards, you download our free Word template(s), supply the item names and other details and then laser-print the results. You then end up with four cards per sheet that look something like this:
Alternatively, you can print onto self-adhesive labels and apply these to index cards, Post-It Notes, photographs or physical objects. An adhesive label looks like this:
We supply perforated card for this purpose in the UK and EU, but there are other companies who provide similar products (try searching the web for "perforated card stock"). Notice that the card templates include a 'quality of fit' measure. This allows participants to say how well an item fits within a group and significantly improves the quality of results.
The labels for which we provide free templates are standard Avery sizes, available in most parts of the world.
Our analysis software, SynCaps comes in two versions. V1 is the original, low-cost edition while V2 provides more sophisticated visual analyses and features for manipulating sorted data, such as merging items or groups. Both provide a traditional "dendrogram" - a tree-like graph that shows the relative strengths of relationships between items (based on how frequently they occurred together in groups. A simple dendrogram is shown below:
(The user-selectable background colour simply allows alternate groups to be more readily identified.) Also, both versions of SynCaps now support Unicode so alphabets other than Latin can be used such as Greek, or in the example here, Cyrillic:
Note that because the dendrogram is based on a cluster analysis of how often pairs of items appeared in the same groups, the group names are not available (in an open sort, there might have been dozens of different group names used by participants). However, this information is shown in an Items x Groups chart in SynCaps V2:
In this chart, the outer part of each cell shows how much agreement there was between participants while the inner part shows agreement with a reference or 'expert' data set. So, in this case, the 'expert' put Groundhog Day into Fantasy/Science Fiction while the participants primarily thought it should be in Humour. This can be seen interactively by clicking on the appropriate cells:
SynCaps V2 also provides a 'Pairs Map' that provides an effective visual summary of the relationships between items:
Again, the central core of each cell shows the 'expert' dataset. This can be any reference sort that would provide a useful contrast. In navigation sorts, the current organization of a web site or intranet is often used as the 'expert' set. This provides the basis of a set of measurements we call 'navigational alignment'.