This document describes the practice and application of the card sorts elicitation technique. Card sorting is a very powerful tool, used to identify unknown or concealed perceptions held by groups of people.
A person is presented with a diverse set of cards with pictures on them. Actual objects may also be used, as well as smells, taste or touch. For example, a set of a dozen or so DVD movies; a set of celphones; a group of photos of vacation destinations, photos of a group of automobiles or a set of drinking vessels, as detailed below.
.
The person is asked to sort the items into a ranked order, or into a set of stacks, that is meaningful to him. The type of sorting chosen is meaningful; the way the items are sorted is also significant. The correlation of selections across several people adds further depth to the interpretation of the card sort.
The subject may be asked to perform several such exercises in a series, with details of every choice carefully recorded. This process may be done in person, or through software, and is often rated as highly enjoyable by the people who do it.
Results are analyzed and statistically cross-correlated, to reveal the underlying thinking, assumptions and prejudices of a group of people and to draw subtle connections between them. Quite often, card sorts reveals startling, unexpected facts about how people perceive and act; for example, it may be discovered that the perceived value of a celphone among teenagers is influenced more by the size of the central navigation button than by style, features or brand.
.
Pictures of automobiles might be spontaneously sorted according to their size, their color, their price, or the perceived modern nature of the design. A group of insurance executives sorted a set of common everyday objects according to the perceived risk of insuring them; a group of school children might sort the very same objects entirely differently, for example whether they would like to play with them.
Above is an example, showing pictures of different types of drinking vessels.
The sort chosen by the user here is the materials the vessels are made from – glass versus pottery.
The cards are numbered, and the details of each sort are recorded. The name of each group, and the numbers of the cards in each group, are captured. The overall criterion for the grouping (e.g. “material” or “number of doors”) is also recorded. For instance, the criteria of cost, size and fragility could all use the group names of “low/medium/high”.
Here’s what the recording looks like for the card sort pictured:
| Participant number: 12
Date: 12-12-2010 Topic: drinking vessels Pack: “drinking vessels” pack . Sort 1: what material they are made from Glass: 1, 2, 3, Pottery: 4, 5, 6, 7, 8 |
Next, the person sorts the cards again, using a different set of spontaneous criteria. For example, one subject chose to sort the same drinking vessels in terms of “what I might drink out of them,” shown here:
The recording sheet for this card sort would now look like this:
| Participant number: 12
Date: 12/12/2010 Topic: drinking vessels Pack: “drinking vessels” pack. . Sort 1: what material they are made from Glass: 1, 2, 3, Pottery: 4, 5, 6, 7, 8 . Sort 2: what I might drink out of them Wine only: 3, 4 Wine or beer: 1, 2, 5 Beer only: 6, 7, 8 |
The subject may be asked to keep running new card sorts this until he runs out of ideas. People generally enjoy this process, and look forward to coming up with new and creative sort criteria. And they often reveal things about themselves that would never be drawn out by traditional surveys or interviews.
Analysis
There are many different ways of analysing card sorts results. The most common are:
- Qualitative analysis of the criteria and group names.
- Which criteria do the respondents use?
- Which group names do the respondents use?
- How many of the criteria and group names are new and unexpected?
- How many of the criteria and/or group names involve subjective versus objective features?
- How many of the criteria and/or group names involve intrinsic versus extrinsic features?
- Quantitative analysis of criteria and groupings
- How many criteria do respondents use – few, or many?
- How many groups do respondents use within each criterion – few, or many?
.
Statistical analysis of co-occurrence
How often is each card grouped together with each of the other cards?
How strongly does each criterion for grouping correlate with each of the other criteria for grouping?
Card sorts many be combined with:
Laddering downwards, to better understand subjective terms that the respondents use;
or laddering upwards, to better understand which features respondents prefer, and why
Video Demonstration
A very simple card sort of commercial media, using a collection of music CD’s, DVD’s and books to show how a card sort is created. Narrated by Dr. Gordon Rugg, Chief Scientist at Prevocative at a professor of knowledge modelling at Keele University, Staffordshire England. (1:23)
Applications:
- Which attributes are important to respondents: useful for marketing campaigns
- Which attributes are often ranked positively, or negatively
- Which attributes the experts are spotting which the non-experts don’t spot: useful for risk assessment
- How often a particular card is sorted next to others, regardless of the actual criteria in different sorts; useful for identifying competition, and for identifying product differentiating factors, competitive advantage
