Decision making in today's fast-paced and resource constrained business environment requires that priorities be set as early and accurately as possible. This is necessary so that budgets, schedules, and (wo)manpower can be allocated appropriately.
Decisions are often delayed due to incomplete information so a method for dealing with subjective data as well as robust to later-added data is useful. Decisions are often inaccurate due to poor mathematical models that ignore scalar limitations. There are four common numerical scales used in judgment and decision-making: nominal, ordinal, interval, and ratio.
Nominal scales are do not support +, -, X, / mathematical functions. Ordinal scales give order but the interval between the scale levels are indeterminate. Rating scales, such as the commonly used Likert-type (1-5 or 1-10) are also ordinal and are often misused despite these limitations. In customer satisfaction surveys, for example, when respondents score an 8 for "I was greeted with a smile" and a 4 for "I was seated quickly," is the former truly twice as good as the latter?
Additional problems arise when some respondents tend to cluster their scores toward the bottom 1-5 range, the center 4-8 range (can't make up their minds!), or the top 7-10 range. This means that traditional calculations such as mean and standard deviation (these require addition, division, and/or raising to powers) should not be done with ordinal scale numbers.
Interval scales have defined intervals but are "local" to the scale. For example, 10oC is twice as high as 5oC and 10oF is twice as high as 5oF, but 10oC is not twice as high as 5oF. Additionally, we cannot say that 10oC has twice the heat as 5oC. If 10oC had twice the heat as 5oC, what would the ratio be for 10oC to –5oC? For this type of problem, we use an absolute ratio scale.
On absolute ratio scales, all numbers have defined and equal intervals. This scale supports all the +, -, X, / mathematical functions as well as mean, standard deviation, etc. and is the preferred scale for decision making. In subjective decision-making, however, it is very difficult to get people to accurately assign ratio scale numerical evaluations.
The Analytic Hierarchy Process (AHP) was created by Dr. Thomas Saaty in the 1960s to facilitate decision making using natural language inputs and ratio scale numerical outputs. The natural language scales and decision matrix math were proven with numerous examples where the results were known a priori.
AHP has added features of calculating judgment inconsistency (a>b, b>c, but c>a) and can does not require group consensus to a single score. The benefit of this approach are a faster, more natural way to make decisions, accurate numerical priorities, a trap for those trying to push an agenda, and a way to make progress even when there is disagreement.
For quality professionals, this method can be applied where inaccurate ordinal scale scoring is currently used such as in project selection, customer satisfaction surveys, technology concept selection, make/buy decisions, supplier evaluations, FMEA RPN calculation, equipment or software package selection, hiring decisions and student evaluations, and even personal decisions such as which car to buy, which university for our children to attend, which healthcare option is best, etc.
The QFD Green Belt® Course and QFD Black Belt® Course teach how to use AHP to make important decisions with greater accuracy and ease. The courses include AHP templates.
© QFD Institute