Concept Sorting and the Repertory Grid

Outline

Concept sorting is one of the most common and straightfoward methods of knowledge elicitation. The technique is based on using number of items or objects that occur in the domain of interest and comparing and contrasting these things to elicit more abstract concepts that identify and conceptually organize these items.

We will use The Repertory Grid Technique to demonstrate concept sorting. This method was developed by psychologist George Kelly and is based on his 'Theory of Personal Constructs'. The theory and technique assume some sort of resprentational theory of mind and are psychological in their orientation towards knowledge. We will therefore briefly look at the basic idea behind concept sorting as well as Kelly's theory before considering the technique in more detail.

In the tutorial there will be the opportunity to experiment with this method.

Concept Sorting

Personal Construct Theory

Personal Construct Psychology is a way of looking at the world. The founder, George Kelly, presents a theory that gives an account of how people experience the world and make sense of that experience.

The key message of the Personal Construction Theory is that the world is 'perceived' by a person in terms of whatever 'meaning' that person applies to it and the person has the freedom to choose a different 'meaning' of whatever he or she wants. In other words, the person has the 'freedom to choose' the meaning that one prefers or likes. He called this alternative constructivism: the person is capable of applying alternative constructions (meanings) to any events in the past, present or future.

PCT attempts to explain "Why man does what he does" (Kelly 1955). It is a theory about "how the human process flows, how it strives in new directions as well as in old, and how it may dare for the first time to reach into the depths of newly perceived dimensions."

"The person's processes are psychologically channelized by the ways in which he anticipate events". The subject is the process... the individual is essentially a behaving organism and doesn't need an external 'push' or 'pull' to get into motion. The processes include those of our self-definition and our relationships with others, as well as the tasks at hand (Kelly 1955).

George Kelly, the original proponent of PCT suggested that PCT is based on the model of man-the-scientist (1955). Within this model,

• the individual creates his or her own ways of seeing the world in which he lives; the world does not create them for him;
• (s)he builds constructs and tries them on for size;
• the constructs are sometimes organized into systems, group of constructs which embody subordinate and superordinate relationships;
• the same events can often be viewed in the light of two or more systems, yet the events do not belong to any system; and
• the individual's practical systems have particular foci and limited ranges of convenience.

Kelly called his philosophy constructive alternativism:

"We assume that all of our present interpretations of the universe are subject to revision or replacement...
There are always some alternative constructions available to choose among in dealing with the world."

Thus he asserted that reality does not reveal itself to us directly, but can be construed in a variety of ways.

Constructs are the way in which things ore people are either similar or different. A construct thus simultaneously differentiates and integrates.

To construe is both to abstract from past events, and provide a reference axis for anticipating future events based on that abstraction.

Kelly's notion of a personal scientist assumes that all people actively seek to predict and control events by forming relevant hypotheses, and then testing them against their experience.

Kelly also assumes a contrast between individual reality, social reality and shared reality:

• Individuality: "persons differ from each other in their construction of events."
• Communality: "to the extent one person employs a construction of experience which is similar to that employed by another, his psychological processes are similar to those of the other person."
• Socialty: "to the extent that one person construes the construction processes of another, he may play a role in a social process involving the other person."

Over the last 40 years, the theory has found its home in the areas of artificial intelligence, education, human computer interaction, and human learning, and is gradually gaining ground in the sociological disciplines. Of particular interest is the emphasis of the theory that communality (the social reality) and individuality (the personal reality) need to be considered together for developing an understanding of the psychological processes.

The Repertory Grid Technique

The repertory grid was conceived as a tool for the use with Kelly's personal construct psychology. It has been used by clinicians and educationalists to elicit implicit world-structures which clients do not or cannot verbalize. Knowledge engineers find it useful for the same reasons.

The first step is to identify a small set of elements (objects, entities), the subject is asked to define some constructs (attributes, slots, dimensions), which characterise those elements.

Constructs are usually generated by the triadic method. Three elements are selected and subjects are asked to pick the odd one out and provide an explanation for their choice. This forms the basis of the construct againt which all the elements are ranked.

For example, presented with the task of characterising three different wines, you might choose the elements Cotes du Rhones, Lambrusco and Liebfraumilch, with the constructs sweetness, colour, and effervescence. The left and right poles for each of these constructs are defined as dry/sweet, red/white, and still/sparkling. It is sensible to define scales for these constructs because there are wines with intermediate values for each of them.

The values associated with the constructs for the three wines are:

However, we can generalize these constructs across the whole domain by giving them ordinal values whereby a low number characterizes one dimension, and a high number the other dimension. In this way constructs can be used to describe the whole domain. Eg,

A low ranking is highly associated with the left dimension, high ranking with the right.

The mid point can be either equidistant between the two or irrelevant to the construct.

And we can continue to elicit constructs to discriminate and classify the elements :

(Graphic supplied by Webgrid)

From the grid we can build a dendogram to show the classifications more clearly, and also identify more abstract classes implied by the elicitation. Note the simplicity of the denogram shows how little I know about wine.

Strengths and Weaknesses

Advantages

• Elicitation of a grid makes the subject reflect on a problem, and helps clarify the mind.
• Grids can be analysed to find patterns or associations for further investigation. This is normally at a general or concept-level. Large grids tend to be unmanageable.
• The technique is relatively efficient. This is very important in organizations where people's time is at a premium and it is important for the investigator to be perceived as professional in obtaining relevant information.
• The structure is not imposed as in a questionnaire, but represents the subject's own construction. This gives it theoretical integrity and in practice can reduce resistance to the inquiry.
• The elements can by anything which is identifiable as a domain object, be it verbal token, visual image, sound, smell, event or whatever. It can escape the influence of labels.
• Constructs a common language with an associateve structure.
• The investigator does not need to know the domain. This is important where a vocabulary of esoteric or professional terms are used.

Problems

• There are problems generating a representative sample set of elements. Ie, are the elements selected sufficient to capture the relevant properties of the domain, and do the distinctions that pertain to them apply more widely? Eg, weather and light are outside the range of covenience of the construct tall v. short.
• Sometimes not all the elements in the triad are equaly representative. One may epitomise a distinctive characteristic. When one element is salient, compared to the other two, the relevant distinction may be exemplar/non-exemplar.
• Emphasis on the forcing of distinctions may result in missing important and even definitional characteristics of the elements. For instance, in our example the facts that wine is alcholic and made from grapes will never be revealed using this method.
• The method provides little guidance as to the degree of distinction required. Triads can be 'milked' to produce further distinctions with the proviso that subsequent constructs may be weaker or contrived. On the other hand, the further consideration required may produce stronger distinctions, perhaps more tacit and less verbally accessible.
• Not all constructs are equal and the investigator must attempt to ascertain which are truly relevant.

Tutorial

Rob Stephens