Interpretive ranking process (IRP)

IRP is a novel ranking method that combines and uses the strength of both the logic choice process with the intuitive process of decision-making. It builds on the strength of a pair-wise comparison approach which minimises the reasoning overload. It also relies on an interpretative matrix as a basic tool and paired comparison of interpretation in the matrix to generate the ranking model. Sushil (2009) suggested that IRP is a powerful method when compared to the existing logic methods such as the Analytic Hierarchy Process (AHP). The AHP depends on an expert judgment about the importance of one element over another one in a pair-wise comparison, along with its intensity. However, the interpretation of these is left in an implicit manner with the expert and thereby the interpretive logic of a decision remains opaque to the implementer. On the other hand, the IRP method presents clearly the interpretive logic of the decision as the expert is

86

supposed to spell out the interpretive logic for dominance of one element over the other for each pair-wise comparison. This logic is usually documented on the knowledge base for future use by decision makers. In addition, IRP does not require the extent degree of the dominance which is difficult to be interpreted and its validity is questionable. Instead, it makes an internal validity check via the vector logic of the dominance relationships by developing a dominance system graph.

Sushil (2009) stated that the interpretive approach to decision-making has been used by different authors who use different constructs such as organisational culture, mental models, sense making, managerial frames, critical thinking and argument mapping. He also presented the steps of the basic IRP process which are illustrated in Figure 4-4 and explained next in more detail.

Identifying ranking variables (X) and reference variables (Y)

Establishing the contextual relationship between ranking and reference variables

Developing cross-interaction matrix of ranking and reference variables

Interpreting interactions (Cross interactions – Interpretive matrix)

Pair-wise comparison to identify the dominating interactions

Developing the dominance matrix

Displaying ranking in a diagram exhibiting all dominance relationships and interpretation

(Interpretive ranking model)

Knowedge management for

further use Recommendations for

87

Figure 4-4 Steps of IRP

Step 1: Variables identification: the first step in this process is to identify two sets of variables. One set will comprise the alternatives to be ranked, and the other set will comprise the criteria that are to be used for ranking the alternatives. In this research, the ranking set consists of five ‘GBM elements’ and the reference set consists of three ‘benefit areas’. The five GBM elements are: GVP, TG, KA, KR, and FL, while the three benefit areas are: B1, B2, and B3. This will be explained in detail in Chapter 6. The decision problem is designed to rank GBM elements with reference to their dominance and influence on various benefit areas.

Step 2: Contextual relationship: this step can be considered the backbone for IRP and needs great attention. In this study, the contextual relationship is the 'influence of GBM elements in different benefit areas'. The elements having more influence will be ranked higher. These relationships have been extracted from the interview analysis. Although there were no direct questions about these relationships, the researcher was able to document them during the analysis process.

Step 3: Cross-interaction matrix: this matrix will be dependent on the previous step, and the relationships developed can be presented as a binary matrix. In the binary matrix, ‘1’ denotes relationship between the two variables while ‘0’ denotes no relationship. In other words, the binary matrix questions the existence of a relationship between each GBM element and the benefit areas. In some cases all the pairs of interactions might exist, thus making the cross-interaction matrix a 'unit matrix'. Step 4: Interpretation of interaction: the binary matrix can be converted into a cross- interaction -interpretive matrix by interpreting the interactions with entry ‘1’. That means all the possible interactions between the pair(s) of variables are to be interpreted in terms of the contextual relationship. The interpretive matrix becomes the essential data for comparison for the purpose of ranking of the variables. In the case of GBM elements and benefits, the interpretation was based on interviewee responses.

Step 5: Pair-wise comparison: The interpretive matrix is used as a basis to pair compares the ranking variables with reference to the reference variable(s) one by one. It is worth noting that the GBM elements (ranking variables) are not directly compared, rather their interaction with reference to. the benefit (reference variables) is compared.

88

All the dominating interactions will be summarised in the dominating interaction matrix, as explained in the next step.

Step 6: Dominance matrix: The numbers of dominating interactions will be summarised in the form of a dominance matrix, which gives the number of cases in which one ranking variable dominates or is being dominated by another ranking variable. The concept of a dominance matrix is taken from the fuzzy set techniques. The sum of rows gives the total number of cases in which the individual ranking variable(s) dominates all other ranking variables. The sum of a column indicates the total number of cases in which a particular ranking variable is being dominated by all other ranking variables. The difference of the number dominating in a column and the corresponding number being dominated in a row gives the net dominance for a ranking variable. The positive net dominance will mean that the variable under consideration has more numbers dominating than being dominated, while the net negative dominance will suggest that the concerned variable is being dominated in a higher number of cases than dominating other variables. The variable having net positive dominance in the maximum number of cases is ranked 1, followed by lower numbers of dominance relationships. The variables with more negative net dominance will be ranked lower, as these are being dominated more by other variables.

Step 7: Interpretive ranking model: The ranks obtained will be diagrammatically represented in the form of an 'Interpretive Ranking Model’. This model displays the final ranks of the ranking variables.

The IRP application is presented and explained further in Chapter 6, with an illustration of ranking GBM elements with reference to various benefit areas.