The 'New' Tools
The following table displays the list of the so-called 'new' Japanese tools. These tools have
been in widespread use in Japan since the mid-1970s. It is some measure of America's
devotion to quality that these tools have only become of interest to American business in the
1980s and 1990s. The tools described are only a few among those that can be potentially
developed. In fact, it is likely that many more tools are currently in use in Japan but not
known in America! The term 'new' tools was used in Japan as the tools were first employed.
More recently (Brassard, 1989) these tools have been renamed as the Management and
Planning (MP) tools. The designation 'new' is retained in this document since it is more
widely used than the 'MP' term.
Tool Name Utilization
Affinity Diagram Used to Organize abstract thinking about a problem.
Relations Diagram Used for determining causalities among parts of a problem.
Systematic Diagram Planning tool.
Matrix Diagram (many types)
Used to organize knowledge in a matrix format; sometimes
includes intercell relationships.
Matrix Data Analysis Method Principal components technique is performed on matrix data.
Process Decision Program
Determining which processes to use by evaluating events and
Arrow Diagram Used to do 'what-iffing' on flow of process.
The tools listed in table above provide a relatively complete capability of analyzing and
understanding a problem. While not essential, the tools can be used in the order shown from
top to bottom of the table to move from more abstract analysis (affinity) to the explicit detail
provided by the arrow diagram. These tools can be usefully combined with the 7 standard
tools (see discussion at end of chapter).
Table of Contents
• Affinity Diagrams
• Relations Diagrams
• Systematic Diagrams
• Matrix-Related Tools
• Process Decision Program Chart
• Arrow Diagrams
• Other Japanese-Origin Tools
Affinity Diagrams (KJ Method)
The affinity diagram tool is also known as the KJ method from the name of the inventor,
Kawakita Jiro (note: Japanese reverse their names, usually writing the surname first). An
affinity diagram is useful for brainstorming about various possibilities about how to solve a
difficult problem or when a team cannot clearly decide what to do next. Figure 2.11 shows an
example of the structure of an affinity diagram. The name of the problem considered is
displayed at the top of the diagram and subproblems are grouped internal to the main
problem. The basic idea is for the team to suggest, at random, areas of concern, then arrange
these concerns in groupings and name the groupings. One way of manually conducting an
affinity diagram session is to write areas of concern on post-it notes, then place the notes on a
large board and have team members discuss and arrange the notes into groupings. The
benefit from the technique, like all the tools, is the enhanced visualization of the problem that
the use of the tool provides. In the example given in the figure below, there are shown four
grouped problem areas that contribute to the main problem. Also shown are three
subproblems within the problem designated in the problem area at the upper right of the
Using the Affinity Diagram as a Team
It is relatively simple to try out using affinity diagrams yourself using the post-it note method
mentioned above. Use the steps shown below.
1. Gather a team; make sure the right people are on the team - that is that the team has
common goals and interests.
2. Discuss and select a specific problem area. For example, lack of productivity.
3. Have each member jot down as many contributing factors to the selected problem as
possible on a post-it note.
4. Post the note on a board and begin as a team to logically group the notes.
5. Name the logically selected groups.
6. Have a period of quiet reflection and permitting any team member to move any note
to any other desired group.
7. Discuss the changes. Iterate these last two steps until a steady state is reached.
8. Next, analyze the different groupings and decide which things to focus on in the
This another example used by a shipping company.
The affinity diagram can be improved by computer automation. Techniques for
computerization are discussed in a subsequent chapter.
Relations diagrams specify the relationships among things. More specifically, these
diagrams are used to map and analyze problems where causes of the problem have
complex interrelationships. In contrast to the Ishikawa diagram in which all causes of
a problem are assumed to be hierarchically decomposable, the relations diagram
promotes discovery of relationship among causes. In plain terms, this means that a
single thing might influence two or more other things, a situation that cannot be easily
shown in a fishbone diagram. Consider the figure below in which multiple
relationships among causes of a specific problem are shown. When teams work in a
group to create relation diagrams, the 'card' or 'post-it note' technique can be used in
which causes are specified by team members and arranged on a table or board and
arrows drawn between cards or notes as needed to show the relationships.
The figure below shows how relations diagrams and Ishikawa diagrams are related.
Ishikawa diagrams are arranged hierarchically, that is, there is a main node, subnodes,
and subnodes of the subnodes, and so forth. The arrangement in which no branches
are permitted to crossover is called a strict hierarchy; those cases in which crossover
is permitted is termed a tangled hierarchy. The relations diagram is most like a
tangled hierarchy. Crossover is permitted, in fact, encouraged. The relations diagram
and the Ishikawa diagram are identical if the structure of either graph does not permit
crossovers between the hierarchically organized structures. In Figure 2.12, the top
diagram shows a simple Ishikawa diagram with 6 causes shown (Cn). Shown below
the Ishikawa diagram is the identical relations diagram with one additional link,
shown to the left. Note that the primary difference between the two diagrams is
simply the arrangement of the drawing - and the ability to have additional
interconnectivity between causes. In truth, there is no reason that such
interconnectivity cannot be shown on the fishbone.
Systematic diagrams help teams or individuals think systematically about how to
achieve a goal. Affinity diagrams assist in identifying a problem, relations diagrams
help to figure out what is related to a problem, and systematic diagrams organize the
aspects of the solution of the problem. Figure 2.12 shows a general form of a
systematic diagram, in this case a structure for assisting in breaking a goal down into
subgoals. Subgoals in this case can become sets of methods and plans which assist in
achieving the subsequent goal. Another way to use the systematic diagram is to
decompose components or physical processes into their subparts. This type of
hierarchy is sometimes known as a whole-part hierarchy, in reference to breaking the
whole into it's constituent parts. The term systematic diagram is also referred to, by
some, as the tree diagram due to the way that the diagram appears.
Once the systematic diagram is organized and completed, the tip nodes (those to the
right in the figure) represent the specific methods and actions that are to be taken. In
this configuration, one can use a matrix diagram to prioritize and organize the work to
be done. When a matrix diagram is used in this fashion it is sometimes termed a
Matrix diagrams permit organization of knowledge so that relationships between
factors, causes, objectives, (or any thing that one wants to show) can be shown.
Matrices, of course, provide rows and columns with intersecting cells that can be
filled with information that describes the relation between the items located in the
rows and columns. Several basic forms of matrix- related tools have been developed
including, an L-type matrix, a T-type matrix, Y-type matrices, and X-type matrices.
The L-type matrix is just a two dimensional table that places contrasting elements in
the rows and columns of the matrix. The L term is used to describe the upside-down
shape of the labels of the row and columns. The T-type matrix forms the labels in the
form of a T adding an additional matrix above the top of the matrix. X and Y-type
matrices are simply composed of multiple L and T matrices. As shown in the diagram
below, the L is discernable in the column and row labeled Cn and Rn; for the T the
additional set of Cns gives the appearance of a T.
R1 R2 R3 R4
L1 1 2 3 4
L2 1 2 3 4
L3 1 2 3 4
L4 1 2 3 4
L4 1 2 3 4
L3 1 2 3 4
L2 1 2 3 4
L1 1 2 3 4
R1 R2 R3 R4
L1 1 2 3 4
L2 1 2 3 4
L3 1 2 3 4
L4 1 2 3 4
The next figure shows another matrix style display technique with a 'rooftop' on the
top of the matrix. This addition makes the diagram into what is sometimes called the
"House of Quality." The additional diagram provides the capability of making
additional relationships between factors that are listed in the columns.
The typical means for showing relationship in a matrix diagram is to place a symbol
in a cell in the matrix, for example:
- strong relationship
- possible relationship
Naturally, it is plausible to use any type of symbol that has meaning to the user.
Different types of symbols are used by different authors and software manufacturers
of matrix toolsets.
One popular methodology based on matrix-style representations is known as Quality
Function Deployment (QFD). QFD designates the columns as hows and the rows as
whats, that is, how to accomplish something and what to accomplish. The whats are
customer requirements or objectives; the hows are the different things that can be
done to achieve the objectives. The roof of the hows show any interrelationship
among the hows. In one software package by Qualisoft (1991), the whats are
segmented into different types of requirements and the hows into different categories
of things to be accomplished. To use this system, the user places in cells various
symbols indicating the relationships between the whats and hows. Weights indicating
the relative importance of the whats permit assessment of which of the columns
(hows) provide the most benefit.
Matrices and Principal Components Analysis.
One of the seven 'new' tools described by Mizuno (1979) is Matrix Data Analysis, a
technique which creates correlation matrices for preferences of individuals for a
product. Brassard (1989) drops this tool from the list of new tools, indicating that the
tool is too complex for inclusion in this set of tools. The technique employs an
analysis of the correlation matrix that is analyzed by principal components methods.
In principal components, information is extracted from the matrix that shows how
much each feature contributes to the principal factors which account for the variance
in the matrix. Vectors of composite preferences then give a perspective on the
characteristics for particular items in the correlation matrix, which can be used for
decision making about which features to include or not include in a product. This
technique is quite similar to the analytical hierarchy process (described below) for
general decision making. The fundamental methodology is to convert the matrix of
information about interrelationships between variables into a simpler form that
creates a set of eigenvectors which are made up from a sum of fractions of the
contributing features. By scanning which features are included, an assessment can be
made about the relative importance of each feature to the overall preference matrix.
Process Decision Program Chart (PDPC)
Process decision program charts are charts that help the user select the best processes
to be used to accomplish a desired task. While the method has many variations, the
most simple explanation of the method is that one shows the possible tasks in a
process and the task sequences of all the relevant alternatives. For example, this
figure shows starting at some beginning point and proceeding toward a goal. Each
circle (node) is a separate task. In this figure, four different pathways are possible
from start to goal. The fundamental idea is that the goal can be reached in various
ways; however, only one of the sequence of tasks will be optimal. The PDPC assists
in visualizing the alternatives when planning a sequence of tasks in a process. Two
major alternatives for analyzing PDPCs are used - forward planning and backward
planning. The first type is illustrated in the figure to the right and the latter in figure
below. The method in forward planning is to begin at the start node and work toward
the goal node, attempting to select the best path as one moves forward. In contrast,
backward planning starts at the end node and moves toward the start.
Arrow diagrams display information about the operation of a process by using arrows
and nodes. Arrow diagrams can be thought of as a way to understand the operation of
a process using time, cost, or other metrics. As a simple example, consider the figure
below which displays a simplified process of creating a new product. In this very
simple diagram, two paths are shown, the top path for creating the hardware
component of the product and the bottom path, the software component. Each link
(i.e., arrow) has a numerical label on it which designates the time required to proceed
from one node to the next. In this example, the values used for the arrows are in
weeks. Hence, the time required for completing the initial software for this product is
much longer than the time required for the hardware. The arrow diagram, as in all the
other tools, simply permits an explicit visualization of the bottleneck problem that
will occur in this process!
Gantt charts have been used for many years to permit visualization and scheduling of
parallel activities. The figure below shows a typical Gantt chart. Tasks are listed on
the right side of the diagram. The times during which the tasks are scheduled are
shown by the extent of the arrow in the diagram. The abscissa in this example might
be weeks or months. In contrast, the arrow diagram attaches the name of each task to
a node and shows the times between the tasks. Parallelism is easily seen and,
moreover, disconnects in time are easily observed so that delays can be identified and
plans can be altered to achieve the minimum time from start to completion of the
process. Note that the same type of analysis could be accomplished using cost or any
other metric that is appropriate to the process.
Other Japanese-Origin Tools
There are undoubtedly many other Japanese-origin and related tools than the ones
discussed above. Two more that have gained prominence are the poke-yoke
methodologies and the Taguchi methods. These two tools are briefly discussed next.
Poke-Yoke. The term poke-yoke is a hybrid word created by Japanese manufacturing
engineer Shigeo Shingo. The word comes from the words yokeru - (to avoid) and
poka (inadvertent errors). Hence, the combination word means avoiding inadvertent
errors. The term can be further anglicized as mistake-proofing, that is, making it
impossible to do a task incorrectly. The text (NKS/factory magazine) referenced at
the end of this chapter contains many examples of mistake-proofing. It is a very, very
simple concept, yet not widely used. For example, if one part fits into a hole and it
must fit in only one orientation, then fool-proofing the assembly of the parts requires
that the part fit in the hole in only the correct orientation. This assembly constraint
can be implemented, for example, by placing a small extrusion on the side of the
inserted part that matches with a key on the part in which the first part is inserted.
There are many examples of this type which can be found in processing, assembly,
measurement, and other tasks. Human workers will make mistakes if it is possible.
Poke-yoke simply removes the possibility.
Taguchi. Dr. Genichi Taguchi, a statistician, has become well-known for his method
for assessing quality. Taguchi defines quality in terms of a loss function which
assesses the loss to society for not having a high quality product. Hence, the higher
the quality of the product, the lower the loss. The basic idea is that variables which
influence a product can be varied to determine the performance of the product in
various situations. In what might be thought of as a poor man's statistical design, the
Taguchi method assesses variation in a matrix formulation which varies each
significant variable around an operating point, keeping track of the products output
for each variation captured in the matrix. A homely example will help explain.
Consider a simple electronic circuit made up of several resistors and transistors,
perhaps an amplifier. Each one of the components in the circuit when varied slightly
can effect the gain of an amplifier. Taguchi's method essentially systematically varies
the values of the circuit components, permitting the experimenter to see what effect
variations will have on the performance of the electronic circuit. In short, the
technique is an empirical means of determining exactly how a product will work and
what variation it is likely to present.