Mustafa Akış- The Analytical Hierarchy Process (AHP)

The Analytical Hierarchy Process (AHP)

Dr. Thomas Saaty was improved The Analytical Hierarchy Process (AHP) to address many of the technical and managerial problems frequently associated with decision making through scoring models. An increasingly popular method for effective project selection, the AHP is a four-step process.(Pinto, 2010) 

The First Step: Structuring the Hierarchy of Criteria 

       The first step consists of constructing a hierarchy of criteria and sub-criteria. Let’s assume, for example, that a firm’s IT steering committee has selected three criteria for evaluating project alternatives: (1) Financial benefits, (2) Contribution to strategy, and (3) Contribution to IT infrastructure. The Financial benefits criterion, which focuses on the tangible benefits of the project, is further subdivided into long-term and short-term benefits. Contribution to strategy,

an intangible factor, is subdivided into three sub-criteria: (a) Increasing market share for product X; (b) Retaining existing customers for product Y; and (c) Improving cost management. 

Hierarchy of Selection Criteria Choices
First Level	Second Level
1.Financial Benefits	1A: Short-term 1B: Long-term
2. Contribution to Strategy	2A: Increasing market share for product X 2B: Retaining existing customers for product Y 2C: Improving cost management
3. Contribution to IT Infrastructure

      Table is a representational breakdown of all these criteria. Note that subdividing relevant criteria into a meaningful hierarchy gives managers a rational method for sorting among and ordering  priorities. Higher-order challenges, such as Contribution to strategy, can be broken down into discrete sets of supporting requirements, including market share, customer retention, and cost management, thus building a hierarchy of alternatives that simplifies matters. Because the hierarchy can reflect the structure of organizational strategy and critical success factors, it also provides a way to select and justify projects according to their consistency with business objectives. This illustrates how we can use meaningful strategic issues and critical factors to establish logic for both the types of selection criteria and their relative weighting. Recently, a large U.S. company used the AHP to rank more than a hundred project proposals worth millions of dollars. Because the first step in using the AHP is to establish clear criteria for selection, 10  managers from assorted disciplines, including finance, marketing, management information systems, and operations, spent a full day establishing the hierarchy of criteria. Their challenge was to determine the key success criteria that should be used to guide project selection, particularly as these diverse criteria related to each other (relative weighting). They found that, in addition to clearly defining and developing the criteria for evaluating projects, the process also produced a more coherent and unified vision of organizational strategy.

       The Second Step:  Allocating Weights to Criteria 

      The second step in applying AHP consists of allocating weights to previously developed criteria and, where necessary, splitting overall criterion weight among sub-criteria. Mian and Dai and others have recommended the so-called pairwise comparison approach to weighting, in which every criterion is compared with every other criterion. This procedure, argue the researchers, permits more accurate weighting because it allows managers to focus on a series of relatively simple exchanges— namely, two criteria at a time.

  

The simplified hierarchy in Figure shows the breakdown of criterion weights across the same three major criteria that we used in Table 3.4. As Figure 3.3 shows, Financial benefits received a weighting value of 52%, which was split between Short-term benefits (30%) and Long-term benefits (70%). This configuration means that long-term financial benefits receives an overall weighting of (0.52)×(0.7) = 36.4%. The hierarchical allocation of criteria and splitting of weights resolves the problem of double counting in scoring models. In these models, criteria such as  Service, Quality, and Customer satisfaction may be either separate or overlapping factors,  depending on the objectives of the organization. As a result, too little or too much may be  assigned to a given criterion. With AHP, however, these factors are grouped as sub-criteria and share the weight of a common higher-level criterion. 

     The Third Step: Assigning Numerical Values to Evaluation Dimensions

For our third step, once the hierarchy is established, we can use the pairwise comparison process to assign numerical values to the dimensions of our evaluation scale. Figure is an evaluation scale with five dimensions: Poor, Fair, Good, Very Good, and Excellent.

Figure also shows that for purposes of illustration, we have assigned the values of 0.0, 0.10, 0.30, 0.60, and 1.00, respectively, to these dimensions. Naturally, we can change these values as necessary. For example, if a company wants to indicate a greater discrepancy between Poor and Fair, managers may increase the range between these two dimensions. By adjusting values to suit specific  purposes, managers also avoid the fallacy of assuming that the differences between numbers on a  scale of, say, 1 to 5 are equal—that is, assuming that the difference between 4 and 5 is the same as  the difference between 3 and 4. With the AHP approach, the “best” outcome receives a perfect score of 1.00 and all other values represent some proportion relative to that score.When   necessary, project managers are encouraged to apply different scales for each criterion. Note, for example, that Figure 3.2 used scale points ranging from Poor to Excellent. Suppose, however, that we were interviewing a candidate for our project team and one of the criterion items was  “Education Level.” Clearly, using a scale ranging from Poor to Excellent makes no sense, so we would adjust the scales to make them meaningful; for example, using levels such as “High School,” “Some College,” “College Graduate,” and so forth. Allocating weights across  dimensions gives us a firmer understanding of both our goals and the methods by which we are comparing opportunities to achieve them.

      Final Step: Evaluating  Project Proposal

In our final step, we multiply the numeric evaluation of the project by the weights assigned to the evaluation criteria and then add up the results for all criteria. 

 

Figure shows how five potential projects might be evaluated by means of an AHP program offered by Expert Choice, a maker of decision software. Here’s how to read the key features of the spreadsheet:

 • The second row specifies the value assigned to each of five possible ratings (from  Poor = 1 = .000 to Excellent = 5 = 1.000 ). 

    • The fourth row specifies the five decision criteria and their relative weights                           (Finance/Short-Term = .1560, Strategy/Cost Management = .0816, and so forth). (Note that three criteria have been broken down into six sub-criteria.) 

    • The second column lists the five projects ( Perfect Project, Aligned, etc.).

     • The column labeled “Total” gives a value for each alternative. This number is found by multiplying each evaluation by the appropriate criterion weight and summing the results across all criteria evaluations.

To illustrate how the calculations are derived, let us take the Aligned project as an example. Remember that each rating (excellent, very good, good, etc.) carries with it a numerical score. These scores, when multiplied by the evaluation criteria, yield:

     The Perfect Project, for example, was rated Excellent on all six dimensions and thus received a score of 1.000. Note, too, the evaluations of the Aligned and Not Aligned project choices. Although both projects received an equal number of Excellent and Good rankings, the Aligned project was clearly preferable because it was rated higher on criteria viewed as more important and thus more heavily weighted. Unlike the results of typical scoring models, the AHP scores are significant. The Aligned project, for example, which scored 0.762, is almost three times better than the Mixed project, with its score of 0.284. This feature—the ability to quantify superior project alternatives—allows project managers to use AHP scores as input to other calculations. We might, for example, sort projects by the ratios of AHP scores to total their development costs. Let’s say that based on this ratio, we find that the Not Aligned project is much cheaper to initiate than the Aligned project. This finding may suggest that from a cost/benefit perspective, the Not Aligned project offers a better alternative than the Aligned project. 

     The AHP methodology can also dramatically improve the process of developing project proposals. In firms that have  incorporated AHP analysis, new project proposals must contain, as part of their core information, a sophisticated AHP breakdown listing the proposed project, alternatives, and projected  outcomes. The Analytical Hierarchy Process offers a real advantage over traditional scoring models, primarily because it reduces many of the technical and managerial problems that plague such approaches. 

     The AHP does have some limitations, however. First, current research suggests that the model does not adequately account for “negative utility”; that is, the fact that certain choice options do not contribute positively to the decision goals but actually lead to negative results. For example, suppose that your company identified a strong project option that carried a prohibitively expensive price tag. As a result, selecting this project is really not an option because it would be just too high an investment. However, using the AHP, you would first need to weigh all positive elements, develop your screening score, and then compare this score against negative aspects, such as cost. The result can lead to bias in the project scoring calculations. A second limitation is that the AHP requires that all criteria be fully exposed and accounted for at the beginning of the selection process. Powerful members of the organization with political agendas or pet projects they wish to pursue may resist such an open selection process.

Here there are some videos about AHP: