QFD Math
The QFD model uses several different scores to calculate the relative and absolute weights. These
numbers are used for aiding in decision-making.
One of the key problems inherent in the system used is that the scales are ordinal scales. That is they
are used to create order between the items but do not reflect the distance between any two items.
Ideally interval or ration scales should be used. It is recommended that the Analytic Hierarchy Process
(AHP) be used in place of the traditional 9-3-1 or 1-5 scales.
Traditional QFD Math
A Relative Weighting (CRW) of all the Customer Requirements (Whats) is calculated using the below
formula.
Im por tan ce Rating
RW = ´100
å Im por tan ce Ratings
An Absolute Weight can also be calculated. The formula for the absolute weight is:
Absolute Weight = (Im pro tan ce Rating ) ´ ( Scale - up Factor ) ´ ( Sales Po int)
There is also a Technical Relative Weight (TRW) calculation. This is a more complex and requires that
you first calculate the CRW. Next the Technical Importance Weight (TIW) is calculated. To calculate the
TIW, the Interrelationship Matrix must first be completed. The value or Interrelationship Matrix Weight
(IMW) associated with the how to what relationship is multiplied by the corresponding CRW.
TIW = å (IMW ´ CRW )
The last step is to then calculate the Technical Relative Importance weight. This relative weight is
calculated just like the customer relative weight as shown below.
TIW
TRW = ´100
åTIW
Calculated Weight Issues
In Math, there are four types of number scales: Nominal, Ordinal, Interval and Ratio. There are rules in
math based on the type of scale used as to how this data can be mathematically manipulated.
• Nominal Scales are used to put data into buckets.
• Ordinal Scales are like nominal scales, except they are placed in some kind of order.
Nominal and Ordinal Scales are:
• Not continuous data.
• Have a finite number of places to organize the data.
• The interval or spacing between datapoints is unknown.
Therefore, you can NOT do multiplication or division on these numbers. You may count them,
determine percentage per bucket. These data can be analyzed using nonparametric techniques.
• Interval Scales have a set or known interval between each data point. This scaling is consistent
and continuous. Interval Scales do not have a real zero point and therefore allow for negative
values
• Ratio Scales are the same as interval scales except they have a real zero point and therefore, no
negative values
Interval and Ratio data can be multiplied and divided.
Likert scales and scales in which you use a truncated scale and force a choice are nominal scales.
You cannot average these choices together or analyze them in anything other than nonparametric
statistics. Likert scales do have a set of rules created by Likert himself on how to analyze this data.
Weights in Statistics are often used. The rules about weights include that the sum of a set of weights
should =1. In quality, we often forget these rules and create our own. This can result in skewed and
misleading results.