CHAPTER FOUR: DECISION MODELS
4.5. Decision Models
4.5.4. Personal Construct Theory (PCT)
The basic idea of Kelly’s (1955) Personal Construct Theory (PCT) is that “a person's processes are psychologically channelized by the ways in which he anticipates events" (p. 46). Walker (1996) refers to the above postulate as the act of construing, which he defines as the act of “meaning making, appraising, applying a theoretical framework, ways you are likely to make sense of things, or adopting new perspective” (p. 8). This implies that a person is constantly striving to develop his view towards himself and his environment as he learns new things. This indicates the main assumption of PCT which views the decision maker as a scientist (Salmon, 1980; Walker, 1996). Thus, PCT views the decision maker not as an object (Walker, 1996), and recognizes the autonomy of the decision maker to control events based on her/his own experience. For an individual, the result of considering a problem is the formation of a ‘construct’ that guides actions. Thus, a construct is akin to a heuristic, or rule of thumb.
There are 11 corollaries within PCT, and they indicate a hierarchical relationship of how a person evaluates events, develops and tests some hypotheses (constructs) and, finally, shapes her/his own values and perception of the events. The summary of PCT’s corollaries is as follows (adapted from Jangu, 1997, pp. 47-58):
a) construction: a person identifies the pattern or order of events and construes accordingly;
b) individuality: each person construes and discriminates events based on his
unique ways of construing and ordering;
c) organization: people develop different ways of discriminating different
constructs;
d) dichotomy: a person compares constructs based on similarities and differences;
e) choice: a person chooses an alternative that enables her/him to deal with future complexities and to anticipate future events;
f) range: a person selects an alternative based on a certain range, whether it is related to relative ease, extent of ease and circumstances;
g) experience: a person construes, learns new things and re-plans his construct in an evolving way;
h) modulation: a person may easily or resistantly change his construct;
i) fragmentation: a person may consecutively construe events in different and
incompatible ways;
j) commonality: people may construe events and take actions in similar ways; and
k) sociality: people construe one another as a way to build relationships, understanding and communication.
Measurement of the component of a PCT is focussed on two aspects. These are: a) elements, which are people, objects, events and activities (Stewart & Stewart,
1981, cited in Jangu, 1997) and used as the focus of analyzing constructs; and b) constructs, which are bipolar statements, or discriminations (Fransella, Bell, &
Bannister, 2004), used for examining similarities or differences among
elements. Besides bipolar in nature, Kelly (1955) also asserts that the constructs should be (i) permeable (can be applied to an individual or a population); (ii)
pre-existing (permanent); (iii) communicable (descriptive); (iv) understanding of other people (appreciative); and (v) non-dissociation. Types of constructs may include (i) pre-emptive constructs; (ii) constellatory constructs; (iii)
proportional constructs (“as if” type); and (iv) corresponding, consensus, conflicting and contrasting constructs.
The procedures usually start with the selection of elements. The criteria for selecting elements are that they must be (i) unambiguous, (ii) uniform, (iii) independent, (iv) representative and (v) detailed (Kelly, 1955). The selection processes include (i) providing a list of considered elements and the relevant clues on the context to the respondents, (ii) describing the scope and category, and (iii) determining elements from discussions with respondents (reviewed from various sources by Jangu, 1997, pp. 63-69). For example, in the case of agricultural technology adoption, farmers are asked to classify some types of fertilizer (the elements), according to their intention or adoption behaviour. The farmers are also asked to explain the reasons associated with their preferences. This is then followed by the elicitation of constructs.
According to Fransella et al. (2004), the 10 ways to elicit personal constructs are3: a) eliciting from triads of elements;
b) eliciting with dyads of elements; c) eliciting with single elements;
d) eliciting using the Full Context Form (a direct comparison of a full list of elements);
e) eliciting from self-characterizations;
f) eliciting from other written material (e.g. an essay about different elements chosen by the subject);
g) eliciting through an interview; h) eliciting using non-verbal materials; i) eliciting using computer; and
j) eliciting using constructs (e.g. laddering, pyramids, ABC model).
The triad procedure is more common. The procedure starts with asking the subject to discriminate between three elements presented to her/him using bipolar statements (constructs). After the constructs are elicited, one element will be replaced by another element, which is randomly selected. The subject is asked again to compare the new set of three elements. The process is repeated until all constructs are elicited. Fransella et al. (2004) refer to this process as the Sequential Form of the triad procedure. This process can be time consuming, as there are many possible triads when many elements are used.
The dyad and interview procedures are commonly used in the case of people who have limited cognitive ability (children or disabled persons) (Barton, Walton, & Rowe, 1976, cited in Fransella et al., 2004). Non-verbal materials may also be used for the same reason; however, the subject often finds it difficult to relate the picture or drawing with a matched verbal expression (Baillie-Grohman, 1975, cited in Fransella et al., 2004).
To elicit constructs in more detail, Epting, Probert, and Pittman (1993, cited in
Fransella et al., 2004) used an essay format which permits the subjects to fully express her/his idea about an event. In a more informal way, this procedure may be akin to Ross’s (1996) qualitative conversational approaches which are based on participant observation such as through “listening to and joining in conversations… and observing everyday incidents and behaviour” (p. 185). This method is believed to enable the elicitation of constructs in groups of people with different cultural backgrounds (Ross, 1996).
A more detailed personal construct system can also be defined using a self-
characterization technique (Epting et al., 1993, cited in Fransella et al., 2004). In farm management, Salmon (1980) discovered that this technique can provide insights for understanding farmers’ cognitive systems. The self-characterization procedure can be followed by hierarchically classifying the constructs, e.g. through laddering or
Compared to other methods, the computer-based construct elicitation is more
commonly used (Dunnett, 1988, cited in Jangu, 1997), and a software package called RepGrid is already available. The reason is that the RepGrid software allows a time- saving procedure. This software also includes the triad procedures, as well as
laddering or pyramiding procedures (i.e. using the FOCUS function). In addition, the RepGrid software can reveal the pattern of constructs, using its PrinCom function, and shared constructs, using the Socio sub-program. Nevertheless, this method is quite challenging for some respondents. For example, most respondents in a study
conducted by Sjah (1998) found it difficult to discriminate between elements during a computer-based interview. He construed the reason may relate to the “marginal differences” between elements that were randomly presented in the triad form. However, the problem in this study could stem from the selection of elements that may not be independent, or clear enough for the respondents.
All the above techniques basically follow the same procedures. All of them involve two-way interactions with the subject. Some interview techniques and factor analyses can replace the functions in the RepGrid software. To describe how this works, a triad procedure using the previous example of fertilizer adoption is presented. In an interview, the farmers are asked to determine the similarities between two out of three types of fertilizers (elements) that are presented. The responses may reveal the constructs, for example profitability, price, etc., which represent the farmers’ decision motives. Then, the farmers are asked to explain the relationships among the
constructs (e.g. profitability and prices) and among the elements (different types of fertilizer). Next, a hierarchical cluster analysis is performed to test the relationships. The farmers are also asked to decide on the effect of using the fertilizers on their farming practices. The responses are analyzed using a principal component analysis, or a factor analysis, to elicit the correlation between constructs and elements, as well as their variances. The results provide the pattern of constructs used by each farmer and may suggest shared constructs among farmers or, in other words, the social constructs.
PCT is commonly applied within the areas of clinical psychology, psychotherapy, marketing, education, tourism and agriculture (Salmon, 1980; Walker, 1996). Salmon (1980), Murray-Prior (1994, cited in McGregor et al., 2001), and Jangu (1997) are among the analysts who have successfully applied PCT using a farm case study. The advantages of PCT mainly relate to (i) its applicability in various behavioural and motivational analyses, and (ii) its bottom-up nature enabling a more accurate description of an individual’s learning process and reducing outsiders’ influence during the eliciting process (Jangu, 1997).
Combining PCT with other methods is also plausible. For example, Jangu (1997) and Murray-Prior’s study (1994, cited in McGregor et al., 2001) combined PCT with the Ethnographic Decision Tree Model. Jangu, in particular, applied PCT in the case of technology adoption and analyzed different constructs among different groups of farmers (adopters and several groups of non-adopters). The results showed that the combination was able to simplify an individual’s decision making structure, reveal the decision rationales, as well as supply information for prediction. The combined procedures were also perceived to be able to overcome some procedural and analytical limitations when applying only one of the methods.