Prioritisation Methods
Introduction
Method prioritization “is a process whereby an individual or group places a number of items in rank order based on their perceived or measured importance or significance” (Brown, C. 2007). Prioritisation allows for an organisation to identify the issues on which to focus its limited resources. This paper will explore two of these methods, and provide a critical analysis where appropriate.
Evaluation
Paper 1 is an essay entitled “The Analytic Process and its Foundation” (Forman, E. H. 2002). AHP was developed in the late 60s by Thomas Saaty, and involves structuring a hierarchy by defining goals, criteria and alternatives. Once a hierarchy has been established, priorities are identified by performing pair wise comparisons against two elements, all the way through the hierarchal chain. Next, the user calculates weighted criteria scores that combine all of the ranking data. And finally, in using these calculated scores, a final figure can be obtained for each alternative.
AHP is widely used whenever any complex situation requires measurement, structuring or synthesis, and is based upon “the welldefined mathematical structure of consistent matrices and their associated righteigenvector's ability to generate true or approximate weights” (Merkin. 1979). AHP is based around three simple axioms. The first axiom, referred to as the reciprocal axiom, states that if element A is five times larger than element B, then element B is one fifth as large as element A. This is a straight forward paired comparison of A and B, with the parent element C representing the calculated outcome.
The second axiom, referred to as the homogeneity axiom, states that any two elements being compared should not have a large difference on the order of magnitude. The basis behind this theory is to increase accuracy whilst decreasing any chance of inconsistencies. If element A has a magnitude value of 1, and element B has a magnitude value of 9, the range of the compared elements represents a value of 8. This large value, when incorporated within complex matrixes, increases the likelihood of human error.
The final axiom, referred to as the rank reversal axiom, states that preference for alternatives is almost always dependent on higher level elements, and the importance of the objectives might be dependent on lower level elements. This can be explained better if we provide elements A and B with a weight and speed property. If elements A and B both have the same weights, but element A is much faster than element B, then speed would be a more important property than weight. However, if elements A and B had the same speed, but differed in weight, then weight would be seen as being more important than speed.
Together, these three axioms make up the foundation of what is a fairly widely used method for prioritization. The question one has to ask oneself however, is how a method that relies largely on human subjects performing complex mathematics can be considered a reliable process in obtaining prioritization? If we look at our first axiom, the mathematic function deriving from this theory is Pc(Eb,Ea) = 1/Pc(Ea,Eb). Although simple for the mathematicians amongst us, this is a function which has to be performed by general stakeholders of the system. Furthermore, when combined with a function matrix of size n*n, the scale of error inflates as the number of requirements increase. It is important to understand that I am not passing judgement on the accuracy of the mathematics, but on the way the mathematics needs to be performed by general stakeholders.
Through the exploration of Forman’s paper, it had been identified that AHP is rarely used in isolation, but alongside other methodologies (Forman, E. H. 2002). This then ask a further question as to why AHP is unable to satisfy completely the goals set within an organisation. Surely one methodology which is able to provide full comprehension of a given situation is better, and more reliable, than a mixture of many. A single method will certainly reduce any likelihood of communication being lost in translation.
In order for AHP to become more sustainable, a happy medium needs to be incorporated. There is a need for AHP to become more self sufficient, without having to work alongside different methods. I feel that there is also a strong prerequisite for AHP to be designed as a program, so that machines can carry out the mathematics for us. This is further elaborated within a case study carried out by the Software Engineering Institute (2006), whereby they state the need to “develop a robust tool that provides both analysis and documentation support”. Until then, it is my personal opinion that AHP should only be used in the simplest of circumstances, and should not be relied upon for multi object prioritization.
Paper 2 is an essay entitled “The Art of Requirements Triage” (Davis, A. 2003). Requirements Triage is the process of determining which requirements a product should satisfy given the time and resources available for its development. For many organisations, the marketplace has become an extremely competitive place where only the strongest survive. In order to make their product more economical, businesses are forced to add additional features and compress delivery schedules. This generally leads to a complete mismatch of requirements and resources, with an end result of a product that fails to meet the customer’s needs.
RT works in a similar way to a triage in the medical profession, whereby in the aftermath of a disaster, three groups of requirements are identified; those beyond saving, those who don’t need saving, and those who do need saving. The resulting analogous groups of requirements are those that the product; absolutely has to satisfy, need not satisfy, should satisfy, but for which there might not be sufficient resources to accomplish.
In order to ascertain where each requirement should be allocated, three activities are carried out. The first is to establish relative priorities for each requirement, the second is to estimate the resources needed in order to carry out each requirement, and the last is to select a subset of requirements that optimizes the probability of the product's success relative to the resource constraint. With these tasks carried out, the user is left with three groups of requirements, ordered by relative importance.
Although RT may seem like a simple, yet effective means of prioritising requirements, it does have some downfalls. Firstly, there has not been much written up about this method within Davis’s paper, which makes it difficult to truly understand the proper techniques of establishing priorities. Secondly, very few companies actually implement this method (Davis, A. 2003), which makes you think as to the true stature it holds amongst other methodologies. Lastly, any mistake within the prioritization of not just RT, but every prioritisation method, would trigger a major loss of revenue. As RT does not employ any complex mathematics to gain a weight of a requirement, it leaves you wondering whether judgement alone can compensate for a rational decision in requirement grouping. Although this may seem like a contradiction to what I critiqued within the first paper, I feel that mathematics is important for accurate results, but as to who performs the mathematics needs to be reconsidered.
However, with many obstacles to overcome, it is my opinion that there is potential for RT to make a stand as a major prioritization method. With the right knowledge and team behind them, organisations can use RT with a systematic approach. And once the method has been implemented within a business, it can be easily adjusted and redesigned to accommodate any new requirements which may arise.
Conclusion
This paper has looked at two different prioritization methods. Although one is purely mathematically based, and the other based around human judgement, they both depend on the perfect actions of humans to produce a viable solution. It is easy to see why these methods are incorporated within organisations, but at the same time it is hard to ignore the fact that they have the potential to cause a dramatic economic fluctuation. The difficulty lies in creating a method which has complete accuracy whilst at the same time producing perfect consistency.
References
Brown, C. 2007. http://www.cdc.gov/o...oritization.pdf
Davis, A., 2003. The Art of Requirements Triage. IEEE Computer, pp.4249.
Forman, E. H. 2002. The Analytic Hierarchy Process and its foundation. School of Business and Public Management, George Washington University. Washington, DC.
Merkin, B. G. 1979. Group Choice, John Wiley & Sons, NY.
Software Engineering Institute, 2006. https://buildsecurit...ts/534BSI.html
Introduction
Method prioritization “is a process whereby an individual or group places a number of items in rank order based on their perceived or measured importance or significance” (Brown, C. 2007). Prioritisation allows for an organisation to identify the issues on which to focus its limited resources. This paper will explore two of these methods, and provide a critical analysis where appropriate.
Evaluation
Paper 1 is an essay entitled “The Analytic Process and its Foundation” (Forman, E. H. 2002). AHP was developed in the late 60s by Thomas Saaty, and involves structuring a hierarchy by defining goals, criteria and alternatives. Once a hierarchy has been established, priorities are identified by performing pair wise comparisons against two elements, all the way through the hierarchal chain. Next, the user calculates weighted criteria scores that combine all of the ranking data. And finally, in using these calculated scores, a final figure can be obtained for each alternative.
AHP is widely used whenever any complex situation requires measurement, structuring or synthesis, and is based upon “the welldefined mathematical structure of consistent matrices and their associated righteigenvector's ability to generate true or approximate weights” (Merkin. 1979). AHP is based around three simple axioms. The first axiom, referred to as the reciprocal axiom, states that if element A is five times larger than element B, then element B is one fifth as large as element A. This is a straight forward paired comparison of A and B, with the parent element C representing the calculated outcome.
The second axiom, referred to as the homogeneity axiom, states that any two elements being compared should not have a large difference on the order of magnitude. The basis behind this theory is to increase accuracy whilst decreasing any chance of inconsistencies. If element A has a magnitude value of 1, and element B has a magnitude value of 9, the range of the compared elements represents a value of 8. This large value, when incorporated within complex matrixes, increases the likelihood of human error.
The final axiom, referred to as the rank reversal axiom, states that preference for alternatives is almost always dependent on higher level elements, and the importance of the objectives might be dependent on lower level elements. This can be explained better if we provide elements A and B with a weight and speed property. If elements A and B both have the same weights, but element A is much faster than element B, then speed would be a more important property than weight. However, if elements A and B had the same speed, but differed in weight, then weight would be seen as being more important than speed.
Together, these three axioms make up the foundation of what is a fairly widely used method for prioritization. The question one has to ask oneself however, is how a method that relies largely on human subjects performing complex mathematics can be considered a reliable process in obtaining prioritization? If we look at our first axiom, the mathematic function deriving from this theory is Pc(Eb,Ea) = 1/Pc(Ea,Eb). Although simple for the mathematicians amongst us, this is a function which has to be performed by general stakeholders of the system. Furthermore, when combined with a function matrix of size n*n, the scale of error inflates as the number of requirements increase. It is important to understand that I am not passing judgement on the accuracy of the mathematics, but on the way the mathematics needs to be performed by general stakeholders.
Through the exploration of Forman’s paper, it had been identified that AHP is rarely used in isolation, but alongside other methodologies (Forman, E. H. 2002). This then ask a further question as to why AHP is unable to satisfy completely the goals set within an organisation. Surely one methodology which is able to provide full comprehension of a given situation is better, and more reliable, than a mixture of many. A single method will certainly reduce any likelihood of communication being lost in translation.
In order for AHP to become more sustainable, a happy medium needs to be incorporated. There is a need for AHP to become more self sufficient, without having to work alongside different methods. I feel that there is also a strong prerequisite for AHP to be designed as a program, so that machines can carry out the mathematics for us. This is further elaborated within a case study carried out by the Software Engineering Institute (2006), whereby they state the need to “develop a robust tool that provides both analysis and documentation support”. Until then, it is my personal opinion that AHP should only be used in the simplest of circumstances, and should not be relied upon for multi object prioritization.
Paper 2 is an essay entitled “The Art of Requirements Triage” (Davis, A. 2003). Requirements Triage is the process of determining which requirements a product should satisfy given the time and resources available for its development. For many organisations, the marketplace has become an extremely competitive place where only the strongest survive. In order to make their product more economical, businesses are forced to add additional features and compress delivery schedules. This generally leads to a complete mismatch of requirements and resources, with an end result of a product that fails to meet the customer’s needs.
RT works in a similar way to a triage in the medical profession, whereby in the aftermath of a disaster, three groups of requirements are identified; those beyond saving, those who don’t need saving, and those who do need saving. The resulting analogous groups of requirements are those that the product; absolutely has to satisfy, need not satisfy, should satisfy, but for which there might not be sufficient resources to accomplish.
In order to ascertain where each requirement should be allocated, three activities are carried out. The first is to establish relative priorities for each requirement, the second is to estimate the resources needed in order to carry out each requirement, and the last is to select a subset of requirements that optimizes the probability of the product's success relative to the resource constraint. With these tasks carried out, the user is left with three groups of requirements, ordered by relative importance.
Although RT may seem like a simple, yet effective means of prioritising requirements, it does have some downfalls. Firstly, there has not been much written up about this method within Davis’s paper, which makes it difficult to truly understand the proper techniques of establishing priorities. Secondly, very few companies actually implement this method (Davis, A. 2003), which makes you think as to the true stature it holds amongst other methodologies. Lastly, any mistake within the prioritization of not just RT, but every prioritisation method, would trigger a major loss of revenue. As RT does not employ any complex mathematics to gain a weight of a requirement, it leaves you wondering whether judgement alone can compensate for a rational decision in requirement grouping. Although this may seem like a contradiction to what I critiqued within the first paper, I feel that mathematics is important for accurate results, but as to who performs the mathematics needs to be reconsidered.
However, with many obstacles to overcome, it is my opinion that there is potential for RT to make a stand as a major prioritization method. With the right knowledge and team behind them, organisations can use RT with a systematic approach. And once the method has been implemented within a business, it can be easily adjusted and redesigned to accommodate any new requirements which may arise.
Conclusion
This paper has looked at two different prioritization methods. Although one is purely mathematically based, and the other based around human judgement, they both depend on the perfect actions of humans to produce a viable solution. It is easy to see why these methods are incorporated within organisations, but at the same time it is hard to ignore the fact that they have the potential to cause a dramatic economic fluctuation. The difficulty lies in creating a method which has complete accuracy whilst at the same time producing perfect consistency.
References
Brown, C. 2007. http://www.cdc.gov/o...oritization.pdf
Davis, A., 2003. The Art of Requirements Triage. IEEE Computer, pp.4249.
Forman, E. H. 2002. The Analytic Hierarchy Process and its foundation. School of Business and Public Management, George Washington University. Washington, DC.
Merkin, B. G. 1979. Group Choice, John Wiley & Sons, NY.
Software Engineering Institute, 2006. https://buildsecurit...ts/534BSI.html
0 Comments On This Entry
Trackbacks for this entry [ Trackback URL ]
Tags
My Blog Links
Recent Entries

Prioritisation Methods
on Feb 03 2012 06:21 PM
Search My Blog
1 user(s) viewing
1 Guests
0 member(s)
0 anonymous member(s)
0 member(s)
0 anonymous member(s)