7 QC Tools

Background

Statistical methods are effective tools for improving the production process and reducing
its defects. However, you must keep in mind that statistical tools are just tools; they
would not work , if used inadequately.

People often try to reduce production defects by tracing directly back to the cause of the
defect. That is a straight forward approach and, at first glance, it seems to be efficient.
But, in most cases, the causes obtained from that approach not true ones.

7 QC Tools provide right approach for any kind of improvement activity or problem
solving. These tools lend objectivity & accuracy to observation and decision making.

The maxims of statistical way of thinking are:

1. Give greater importance to facts than abstract concepts.
2. Do not express facts in terms of senses or ideas. Use figures derived from specific
observational results.
3. Accept regular tendency which appears in a large number of observational results as
reliable information.

Contents

The Tools which we will learn are:

1. Check Sheet
2. Pareto Chart
3. Stratification
4. Cause-and-Effect Diagram
5. Histogram
6. Scatter Diagram
7. Graphs and Control Charts

For each of the above, the followings will be covered.

 Description of the tool

 When to use the tool (application)
 Procedure to apply the tool
 Interpretation tips for the tool
 Examples/Case study of application with specific reference to MSME
 Exercises to support learning
 Check Sheet

Description:
Check sheets are simply an easy to understand form used to collect data by making
tally marks to indicate,“ number of times something occurs". It starts the process of
translating “opinions” into “facts". Check sheet help standardize the data that is collected
and the data collection process.

When to use: When you need to gather data based on sample observations in order to
begin to detect patterns. This is the logical point to start in most problem-solving cycles.

Check Sheet –Preparation Steps

Constructing a check sheet involves the following steps:
1. Agree on the event being observed. Everyone has to be looking for the same thing.
2. Decide on a time period to collect data. This could range from hours to weeks.
3. Design a form that is clear and easy to use, with columns clearly labeled and
adequate space for entering data.
4. Collect the data consistently and honestly. Make sure there is time allowed for this
data-gathering task.

Types of Check Sheets

Types of check Purpose Description

sheet
1. Defective item Data by type of At the end of day/shift, we can
defect immediately calculate the total number
and types of defects
2. Defect location Data on position of This check sheet easily leads to action
defects because the causes of defects can often
be found by examining the place where
defects occur and carefully observing the
process to determine why defects become
concentrated in these places
3. Process Show dispersion Use this check sheet when you want to
distribution know the variation in the dimensions of
any item
4. Defect cause Stratify - by period, In this stratification is also done in terms
(factor) operator, machine of operator, machine, shift etc.
etc.
Check Sheet – Defective Item

Product: Date:
Manufacturing stage: Final insp. Section:
Type of defect: scar, incomplete crack, misshapen Inspector’s name:
Total no. inspected: 1525 Lot No.:
Remarks: All items inspected Order No.:

Type Check Subtotal

Surface scars |||| |||| |||| || 17
Cracks |||| |||| | 11
Incomplete |||| |||| |||| |||| |||| | 26
Misshapen ||| 3
Others |||| 5
Total: 62
Total rejects |||| |||| |||| |||| |||| 42
|||| |||| |||| |||| ||

If it is suspected that there is a shift to shift variation,

is the design of check sheet good??

Check Sheet - Blowhole Location

Product No. and name:_____________________________________________________________________________________________

Material:_______________________________________________________ __________________________________________________
Maker:__________________________________________________________ _________________________________________________
1. Sketch

2. Defect location matrix

Radial 1 2 3 4 5 6 7 10
Circular | | | | | | | |
A | 1
B
C
D
E ||| |||| | 9
F | II 3
G
H
----   
---- 4 2  13
10 ---- 7

Can you think of application of this in Your Organization?

Check sheet – Process Distribution
Range Tally Marks Frequency
169 <183 |||| ||| 7
183 <197 |||| || 8
197 <211 |||| || 7
211 <225 |||| |||| |||| 7
225 <239 |||| |||| || 27
239 <253 |||| |||| || 12
253 <267 |||| | 6
267 <281 || 2
wedge hardness of the item is being marked in the
check sheet which gives idea about the variation

Check Sheet – Defect Cause

Check Sheet of Paint Defect

Defect
Defect Paint Over Dust Less Scratch Freq.
Paint Peel flow Glazy
Booth off
A || || |||| | | 11
B |||| |||| |||| || | | 19
C |||| |||| ||| | 14
Sum 7 22 10 2 3 44
Paint booth B is the cause which gives maximum defects

Construction/interpretation Tips: Check Sheet

• Make sure that observations/samples are as representative as possible
• Make sure the sampling process is efficient so that people have time to do it
• In doing a check sheet, you may not be aware that the population(universe) being
sampled is non-homogeneous (not from the same machine, person etc.). The
population must be homogeneous. If not, it must first be stratified (grouped), with
each grouping sampled individually.

Exercise
In a grinding process, there are 2 workers, each operating two machines. The fraction
defectives of this process has gone up. The workers are asking for a change of
machinery, saying that the machines currently being used are too old. The staff in
charge of the process says that the workers should be more careful because they are
making careless mistakes.

How check sheet can help in this situation. Design an appropriate check sheet for the
abovementioned situation.

Prepare check sheet for the given situation on this page

 Pareto Chart

Description:
• A Pareto chart is a graphical tool that helps you break a big problem down into its
parts and identify which parts are the most important.
• Doing a Pareto chart based upon either check sheets or other forms of data
collection
helps us direct our attention and efforts to the truly important problems.
• We generally gain more by working on the tallest bar than tackling the smallest bars.
• Developed by V. Pareto, an Italian economist and applied by Dr. JM Juran, a leading
quality consultant

When to Use: when you need to display the relative importance of all of the problems or
conditions in order to :choose the starting point for problem solving, monitor success, or
identify the basic cause of problem. Characteristics of Pareto chart are:
• Identifies vital few from trivial many.
• Helps to bring ‘focus’ by prioritization principle.
• Prepared by plotting integrated percentages.

Type
What is it?
• A Pareto Chart is a Histogram
• + a cumulative line

How does it Work?

• Each bar in a pareto represents a different category of problem. The height of each
bar indicates the relative importance of that problem
• Bars are arranged in descending order from left to right
• The bar for the biggest problem is always on the left

What is the basic principle?

• Pareto Charts are used to apply the 80/20 rule of Joseph Juran which states that
80% of the problems are the result of 20% of the problems. A Pareto Chart can be
used to identify that 20% route causes of problem.
Pareto Analysis

J uly 1 – J uly 31

Cumulative percentage of total defects

-- 100
Number of defective items

150 -- -- 90
-- 80
-- 70
100 -- Vital few Trivial many -- 60
-- 50
-- 40
50 -- -- 30
-- 20
-- 10

A B C D E Ohers
A: Dimensional defectives
B: Pinholes
C: Scratches
D: Cracks
E: Distortion

Pareto
J uly 1 – J uly 31
Percentage of total defects
Number of defective items

80 --
70 --
60 --
50 --
40 --
30 -- V: Fitting position
20 -- W: Working speed
X: Components
10 --
Y: Abrasion of jigs & Tools
Z: Shape of parts
Unclear Others
Pareto Before & After
J une 1 – J ul 31
-- 100

150 -- Total Effect

Number of defective items

-- 80

Sept 1 – Oct 31
-- 160
100 -- 100 –- -- 100

-- 80
-- 40
-- 60
50 -- 50 –- Effect
-- 40
-- 20
-- 20

A B C D E Others C B A D E Others

Steps in constructing a Pareto Chart

1. Select the problems that are to be compared and rank-ordered by:
Brainstorming, e.g., “what are our majority quality problems in department A?”
Using Existing Data, e.g., “let’s take a look at Department A’s quality reports over the
last month to find out the major problem areas”
2. Select the standard for comparison unit of measurement. eg. annual cost, frequency
etc
3. Select time period to be studied, e.g.8 hours,5 days,4 weeks
4. Gather necessary data on the site of each category, eg, “Defect A occurred x times
in the last 6 months” or “defect B cost x dollars in the last 6 months”.
5. Compare the frequency or cost of each category relative to all other categories eg,
“Defect A happened 75 times, Defect C happened 107 times; Defect C happened 35
times” or “Defect A cost Rs.75,000 annually; Defect B cost Rs.98,000 annually”
6. List the categories from left to right on the horizontal axis in their order of decreasing
frequency or cost. The categories containing the fewest items can be combined into
an “other” category, which is placed on the extreme right as the last bar.
7. Above each classification or category, draw a rectangle whose height represents the
frequency or cost in that classification.
Additional features of Pareto Chart
1. Often the “raw data” is recorded on the left vertical axis with a percentage scale on
the right vertical axis. Make sure that the two axes are drawn to scale, eg, 100% is
opposite the total frequency or cost;50% is opposite the halfway point in the raw
data.
2. From the top of the tallest bar and moving upward from left to right, a line can be
added that shows the cumulative frequency of the categories. This answers such
questions as, “How much of the total is accounted for by the first 3 categories?”

Data on Rejection in Foundry

Type of Defect Number Code

Rejected
Blow hole 20 A
Porosity 35 B
Misrun 5 C
Cold shut 11 D
Slag inclusion 3 E
Sand inclusion 4 F
Shift 10 G
Dimension deviation 2 H
Mechanical damage 24 I
Poor strength 2 J
Cracks 2 K
Others 2 L

Arranging data in descending order and calculating cumulative contribution

Code Numbers Percentage Cumulative
Rejected Contribution Contribution
B 35 29.3 29.3
I 24 20.0 49.3
A 20 16.6 65.9
D 11 9.1 75.0
G 10 8.3 83.3
C 5 4.1 87.4
F 4 3.3 90.7
E 3 2.5 93.2
H 2 1.7 94.9
J 2 1.7 96.6
K 2 1.7 98.3
L 2 1.7 100.0
120 100
Pareto Diagram

100

80
% CONTRIBUTION

60

40

20

0
B I A D G C F E H J K L
TYPE OF DEFECT

Construction/interpretation Tips: Pareto

• Use common sense-two key customer complaints may deserve more attention than
100 other complaints, depending on who the customer is and what the complaint is
• Mark chart clearly to show standard of measurement (Rs.,%,number)

Exercise
Given below is number of different types of complaints received from customer. Draw a
Pareto diagram to identify the major causes for complaints

Pareto Data
Customer complaints Numbers
ParetoChart
Delayed delivery 60
Wrong material 20 200 100

Missing 10 180 90
Less quantity 80
160 80
Documentation errors 15
Damaged packages 10 140 70

Others 5
120 60
Total 200
Numbers

Cum%

100 50
Pareto Table
Category Type of Complaint Numbers % Cum % 80 40

A 60 30

B
40 20
C
D 20 10

E
0 0
F A B C D E F G
Catagory
G
Total 200
 Stratification
Description
Stratification is a way of dividing a whole group of data into subgroups. It allows you to
see if there are differences in the data from the different subgroups. Simply, stratification
is a technique for finding major sources of variation in a process.

How to Use (Procedure)

Stratification is taking a set of measurements and placing each measurement into one of
two or more categories that may give some insight to sources of variability. So, instead
of a single data set, there are several smaller sets that can be displayed and compared
using any of the statistical tools seen thus far. The categories could be periods of time,
source, location, system, etc.

How to stratify Data

To stratify data, look for factors you think might help you understand the variation in your
data. Divide the data into subgroups based on these factors. Stratification is taking a set
of measurements and placing each measurement into one of two or more categories
that may give some insight to sources of variability. So, instead of a single data set,
there are several smaller sets that can be displayed and compared using any of the
statistical tools seen thus far. The categories could be periods of time, source, location,
system, etc.

When to choose Stratification factors

It is very important to make stratification decisions before you begin collecting data. If
you decide you want to stratify data after it has been collected, you will usually find, you
lack the information you need on the different stratification factors. Then you have to go
back and collect new data.

Deciding how to stratify the data is part of step 1 of the data collection process.

Stratification should be done with respect to

• Man (Operator - A, Operator - B)
• Machine (M/c 1, 2)
• Material (B.Q. Steel, Gun Metal, Satellite)
• Method (Type of tool, jigs, fixtures)
• Time (Shift 1, 2)
Scene I
50 pieces 90 OK
M/C 1 production Total
Inspection
100
50 pieces 10 NG
M/C 2 pieces
production
Scene II
40 OK
50 pieces
M/C 1 Inspection
production 10 NG

50 pieces
M/C 2 Inspection 50 OK
production

What is the difference between Scene I and II ?

Four Perspectives
Time : When does the problem occur
Place : Where does the problem occur
Type : Are there different types of the problem?
Symptom : What are the symptoms of the problem ?

Exercise

Identify the situations from your work area, where you can do Stratification?

Area/ Department/ Function Topic/ situation Ways to Stratify (categories)

 Brainstorming
Description
All of the charting techniques are aids to thinking. They focus the attention of the user
on the truly important dimensions of a problem. It is equally important, however, to
expand your thinking to include all of the dimensions of a problem or solution.
Brainstorming is used to help a group create as many ideas in as short a time as
possible.
– Brainstorming is a useful technique to harness a team’s collective thinking power in
order to generate a list of process activities and potential root causes.
– It is often used to generate a long list of thoughts and ideas in a relatively short
period of time.
– Brainstorming relies on the synergy of the team and builds on what each person has
to say.

Brainstorming - Procedure
Brainstorming can be used in two ways
– Structured –in this method, every person in a group must give an idea as their turn
arises in the rotation or pass until the next round. It often forces even shy people to
participate but can also create a certain amount of pressure to contribute.
– Unstructured – in this method, group members simply give ideas as they come to
mind. It tends to create a more relaxed atmosphere but also risks domination by the
most vocal members
In both the methods, the general “rules of the road” are the same as mentioned on next
page.

Rules:
• No criticism. Don’t evaluate ideas during idea generation.
• No constraints. Take a freewheeling approach. It’s welcome, even preferred—the
wilder the idea, the better. It’s easier to tame ideas than to make them more and
more outrageous.
• Write on a flip chart or blackboard every idea. Having the word visible to everyone at
the same time avoids misunderstandings and reminds others of new ideas.
• Go for quantity. Generate as many ideas as possible. More ideas increase the odds
of good ideas emerging.
• Build on existing ideas. Don’t leave an idea alone because someone else came up
with it. Use one idea to create others. Combine and improve them.

DO
• Understand the exact issue, topic, area before starting Brainstorming
• Give time to individuals to complete their thoughts.
• Build on existing ideas.
• Record on the flip chart in the words of the speaker; do not interpret.
• Be brief when stating an idea.
• Strive for quantity rather than quantity
• Do it quickly; 5-15 minutes works well.

DON’T
• Kill ideas
• Make judgments, verbal or visual, during the brainstorming session
• Dominate the session.
Exercise

Do brainstorming on “How to improve effectiveness of training

programme” and write your ideas on this page
 Cause and Effect Diagram
Description
Tool that facilitates to identify the results (characteristics or effects) and the possible
causes (factors) in a systematic manner.
• JIS definition: ‘A diagram which shows the relation between a quality
characteristic and factors.’
• Developed by Prof. Ishikawa, hence known as Ishikawa diagram
• Also known as Fish-bone diagram due to its shape/ structure.

When to Use: use this tool, when you need to identify and explore and display the
possible causes of a specific problem or condition

How a Cause and Effect diagram helps solve problems

A cause and effect diagram helps organize possible causes of a problem
• It helps you see the problem and its potential causes
• It makes potential relationships among factors visible
• It helps structure ways to think about potential causes
• It provides a way to document causes verified with data

A cause and effect diagram is not really a data tool because the information it displays is
based on opinions. it helps identify theories about possible causes, which must then be
verified with data to confirm that they either are or are not real causes

Cause a Cause b

effect

Cause c Cause d

Cause effect diagram – procedure

1. Generate the causes needed to build a cause & effect diagram in one of the
two ways
-- structured brainstorming about possible causes without previous
preparation
-- ask members of the team to spend time between meetings using simple
check sheets to track possible causes and to examine the production
process steps closely
2. Construct the actual cause & effect diagram by:
-- placing problem statement in box on the right
-- drawing the traditional major cause category steps in the production
process, or any causes that are helpful in organizing the most
important factors
-- placing the brainstormed ideas in the appropriate major categories
-- for each cause ask, “why does it happen?” and list responses as
branches off the major causes.
3. Interpretation
In order to find the most basic causes of the problem:
-- look for causes that appear repeatedly
-- reach a team consensus
-- gather data to determine the relative frequencies of the different causes

Cause & Effect Diagram-example

Operators Machine
Fatigue illness Stability
Health Item Operation
Concentration
Imbalance
Spirit Inspection
Education Training Deformation
Skill Attentiveness Method J igs & Tools
Abrasion
Experience
Dimensional
variation
Shape
Position
Form Degree of Fitting
tightening material
Component Dimension Angle
Setting
Quality of Diameter
material Order Procedure
Storage Working
Speed
Parts & materials Operational method

PROCED UR E P EOP LE
A n yb o d y c an sen d
I n v en to ry rev i ew M a c h in e d is p o s a l
S to r e/ e m p lo y e e s k e e p o rd e r to su p p l ie rs
S ys te m n o t ava i la b l e c h e c k lis t n o t
E x tr a in v e n to r y
a v a ila b le
I n v en to ry s ta tu s P o o r c o m m u n i c ati o n b etw e en
n o t vi si b l e C o m p eten cy o f U ser /s to re / p u rch ase
S A P s u p p o rt em p l o ye es
N o t av ai l ab le M u l ti co d i fi cati o n
H ig h o b s o le te /n o n- m o v in g
in v e n to ry
M ac h i n e w i se s p a re p a rts
L i s t n o t av ai l ab le P o o r sp are s
av ai l ab i li ty A va il a b i l i ty / n o n - a vai l ab i l ity
M a ch in e re p l ac em en t O f fu n d s
to o freq u e n t
L ess b re akd o w n s
D es i g n c h a n g e i n
E x i sti n g m ac h i n es
M a c h in e Money
Construction/interpretation Tips:C&E Diagram

• On Preparation
– List all factors with involvement of all concerned.
– Use as few words as possible
– If ideas are slow in coming, use the major causes categories as catalysts,eg,
“what in material is causing…..?”
– State the characteristic (effect) very clearly.
– Both characteristics and factors should be measurable
– Select factors where action is possible.
– Look to cure the causes and not the symptoms of the problem

• On usage
– Assign importance of each factor based on data.
– Try to improve the CE diagram continually while using it.

Exercise
Draw Cause and Effect Diagram for “why high number of Accidents on Highways”

Effect
Cause
 Histogram

Description: USL
As we have already seen with the Pareto chart, it is very LSL
helpful to display in bar graph from the frequency with which
certain events occur (frequency distribution). The Pareto
chart, however, only deals with characteristics of a product
or service, e.g, type of defect, problem, safety hazards
(attribute data).

A Histogram takes measurement data, e.g, Temperature, dimensions, and displays its
distribution. A Histogram reveals the amount of variation that any process has within it.

When to use: when you need to discover and display the distribution of data by bar
graphing the number of units in each category

What is it?
• A Histogram is a bar graph usually used to present frequency data

How does it Work?

• Define Categories for Data
• Collect Data, sort them into the categories
• Count the Data for each category
• Draw the Diagram. each category finds its place on the x-Axis.
• The bars will be as high as the value for the category

a. General type b. Comb type c. Positive skew type d. left-hand precipe

e. Plateau type f. Twin-peak type g. Isolated-peak type

Study the above histograms and discuss

what can lead to this type of distribution shape ??
Types of Histogram-Shapes

General Type (symmetrical or bell shaped)

The mean value of the histogram is in the middle of the range of data. The frequency is
the highest in the middle and becomes gradually lower towards the ends. The shape is
symmetrical. This is the shape which occurs most often.

Comb Type (multi modal type)

Every other class has a lower frequency. This shape occurs when the number of units of
data included in the class varies from class to class or when there is a particular
tendency in the way data is rounded off.

Positively skew type (negative skew type)

The mean value of the Histogram is located to the left (right) of the centre of range. The
frequency decreases somewhat abruptly towards the left (right), but gently towards the
right (left) asymmetrical. This shape occurs when the lower (upper) limit is controlled
either theoretically or by a specification value or when values lower (upper) than a
certain value do not occur.

Left – hand precipice type (Right –hand precipice type)

The mean value of the Histogram is located far to the left (right) of the centre of range.
The frequency decreases abruptly on the left (right), and gently towards the right (left).
Asymmetrical. This is a shape which frequently occurs when a 100% screening has
been done because of low process capability, and also when positive (negative)
skewness becomes even more extreme.

Plateau Type
The frequency in each class forms a plateau because the classes have more or less the
same frequency except for those at the ends. This shape occurs with a mixture of
several disruptions having different mean values.

Twin-peak Type (bimodal type)

The frequency is low near the middle of the range of data, and there is a peak on either
side. This shape occurs when two distributions with widely different mean values are
mixed.

Isolated-peak type
There is a small isolated peak in addition to a general-type Histogram. This is a shape
which appears when there is a small inclusion of data from a different distribution, such
as in the case of process abnormality, measurement error, or inclusion of data from a
different process.
Histogram with Specifications
If there is a specification, draw lines of the specification limits on the histogram to
compare the distribution with the specification. Then see if the histogram is located well
within the limits.

Five special cases appears and each of them is described below:

Cases in which histogram satisfies the specification.

Good process Maintenance of the present state is all that is

needed, since the histogram amply satisfies
the specification

SL a) SU

The specification is satisfied, but there is no

Good process extra margin.Therefore,it is better to reduce
with high the variation by a small degree.
variation

SL SU
b)

Cases in which histogram does not satisfies the specification.

It is necessary to take measures to bring Process skewed

the mean closer to the middle of the towards lower
specification. side

SL c) SU

Process with
This requires action to reduce the high variation
variation.

SL d) SU
Process skewed
towards higher
Both the measures described above are side
required in this situation.

SL e) SU

Histogram
1. To visually understand the general tendency of data distribution.
2. To know the approximate data mean.
3. To know the outline of data distribution,
4. To compare data distribution with the specification value or standard value.
5. To compare difference between lots, and,
6. To show data distribution to those who have little experience with statistics. On the
contrary, a histogram also has a disadvantage that the time change of data and
individual data aspects are difficult to represent.
Bore Diameter of Bushes

10.77 10.76 10.73 10.75 10.78

10.76 10.79 10.75 10.75 10.76
10.77 10.78 10.76 10.78 10.75
10.80 10.77 10.74 10.79 10.74
10.72 10.75 10.82 10.76 10.73
10.76 10.77 10.79 10.76 10.77
10.75 10.76 10.77 10.78 10.75
10.81 10.74 10.81 10.74 10.73
10.75 10.78 10.76 10.80 10.77
10.74 10.79 10.78 10.77 10.80

Specs :10.75+0.05

Histogram
LSL USL
20
18 17

16 14
14
12
FREQUENCY

10 8
7
8
6
3
4
1
2
.72

.74

.76

.78

.80
10.70

10.82

BORE DIA

Procedure for Making Histogram

1. Obtain a set of 50 to 100(N) observations.
2. Obtain the maximum value and minimum value.
3. Determine the number of classes `K'
K = √N
4. Class width c = Max - Min, rounded off to a convenient figure.
K
5. Calculate the class boundaries as, [min - 1/2 (Least-Count)] to be the starting point.
6. Make a frequency table, i.e., number of data falling in each class.
7. Histogram : Erect rectangles over the class interval having area proportion
to the frequencies.
8. Frequency polygon: Starting from zero
Join the midpoints corresponding the top of each rectangle by straight lines.
Frequency Table

Class Class Boundaries Mid value Frequency Tally Frequency

No.
1 3.275 – 3.325 3.3 /// 3
2 3.325 – 3.375 3.35 /// 3
3 3.375 – 3.425 3.4 //// //// 9
4 3.425 – 3.475 3.45 //// //// //// //// //// //// // 32
5 3.475 – 3.525 3.5 //// //// //// //// //// //// //// /// 38
6 3.525 – 3.575 3.53 //// //// 10
7 3.575 – 3.625 3.6 /// 3
8 3.625 – 3.675 3.65 / 1
9 3.675 – 3.725 3.7 / 1

Histogram

40

35

30

25

20

15

10

0
3.3 3.35 3.4 3.45 3.5 3.55 3.6 3.65 3.7
Normal distribution

- For variable measurement

- Bell-shaped curve
- Uniform about centre
- Uni-peak curve
-
+3

+2

+

         

Recollect what you know about Normal distribution and list its characteristics.

Construction/Interpretation Tips: Histogram

1. The number of classes (bars in the graph) determines how much of a pattern will be
visible.
2. Some processes are naturally skewed; don't expect every distribution to follow a bell-
shaped curve.
3. Get suspicious of the accuracy of the data if the classes suddenly stop at one point
(such as a specification limit) without some previous decline in number.
4. Always look for twin peaks indicating that the data is coming from two or more
different sources, e.g., shifts, machines etc
Histogram – exercise

There was complaint from customers that a particular property of the product is not
consistent.

A team was constituted to study the problem. The team brainstormed for causes and
came up with a strong probable cause. They felt the process is being tuned in every
shift for aligning this property to meet customer requirement based on lab analysis.
There seems to inspector error which is leading to this. The new set of inspectors who
joined recently does not know the method well and probably causing the error.

They randomly selected 50 readings from the lab log book and plotted a histogram to
validate this theory. Data collected is given below. Draw a histogram and interpret it.

497 504 504 486 503 488 517 480 498 523
493 508 503 499 508 507 498 505 493 486
503 490 509 507 501 501 498 493 506 487
495 503 496 515 504 529 492 501 502 506
501 510 502 499 500 501 501 512 516 502

Construct Histogram on this page for the situation given on previous page
 Scatter Diagram

Description:
A Scatter Diagram is a graph that helps you visualize the relationship between two
variables. It can be used to check whether one variable is related to another variable
and is an effective way to communicate the relationship.

A Scatter diagram shows one variable along the vertical axis and another variable along
the horizontal axis.
• Each data point represents a pair of measurements (such as age and height of
school children)
• The resulting pattern shows how the two measurements are related

When to use:
• To check out possible causes of a problem
• To see if there is a relationship between two variables
• When you can get pairs of data
• When both variables are measured on scales that have a numerical order
– Measurements like time,count,and size all have numerical order
– Categories like color or kinds of errors do not have numerical value

Scatter diagrams – characteristics

1. The plot uses two axes, one horizontal and one vertical
2. The plot is square; both axes are about the same length
3. Two variables are shown, a potential cause variable and an effect variable
4. Each plotted point represents a pair of measurements
5. The resulting pattern of points shows how the two variables are related

Scatter diagrams – procedure

1. Collect 50 to 100 paired samples of data that you think may be related and construct
a data sheet.
2. Draw the horizontal and vertical axes of the diagram. The values should get higher
as you move up and to the right on each axis. The variable that’s being investigated
as the possible “cause” is usually on the horizontal and the “effect” variable is usually
on the vertical.
3. Plot the data on the diagram. if you find the values being repeated, circle that point
as many times as appropriate.
4. The resulting diagram may look like as shown on next page.
Scatter diagram
Scatter Diagrams (Oct. 1 – Nov. 9)
(%)
n = 30
0.93 --
0.92 --
0.91 --
0.90 --
0.89 --
0.88 --
0.87 --
0.86 --
0.85 --
| | | |
8.0 8.5 9.0 9.5 (kgf/cm2)
Blowing air-pressure
What do you conclude from above scatter plot??

Y --
n = 30
--

6 --

--

4 --

--
Outliers
2 --

--

0-- | | | | | | | |
0 2 4 6 x
Blowing air-pressure

Construction/Interpretation Tips: Scatter

1. A negative relationship (as y increases, x decreases) is as important as a positive
relationship (as x increases, y increases)
2. You can only say that x and y are related and not that one causes the other.
3. There are statistical tests available to test the exact degree of correlation but are
beyond the scope of QC Tools.

Correlation and Causation: Background Information

When we create scatter plots, we are usually looking to see if changes in one variable
cause changes in another variable. This is called Causation. Unfortunately, scatter
plots alone do not prove causation. They only show that there is some relationship
between the two variables. This is called Correlation.

When two variables are correlated, it means that there is some predictability or pattern
in the relationship between them.
Correlation does not mean Causation
When two variables show a relationship on a scatter plot, they are said to be correlated.
But this does not always mean they have a cause-effect relationship.
• Correlation means two things vary together.
• Causation means changes in one variable cause changes in the other variable.

Scatter Diagram- Various Possibilities

Y -- n = 30 r = 0.9
6 -- An increase in y may depend on an increase in
-- x.If x is Controlled, we might have a good
4 --
--
chance of controlling y.Example,training vs
2 -- performance
--
0 | | | | | | |
0 2 4 6 x
Positive correlation

Y -- n = 30 r = 0.6
6 --
--
If x is increased,y may increase somewhat, but
4 -- y seems to have causes other than x
--
2 --
--
0 | | | | | | |
0 2 4 6 x
Positive correlation, May be Present
Y -- n = 30 r = 0.0
6 -- There may be no correlation. y may be
-- dependent on another variable.
4 --
--
2 --
--
0 | | | | | | |
0 2 4 6 x
No correlation

Y -- n = 30 r = - 0.9
An increase in x may cause a decrease in 6 --
y.therefore,as with situation 1 above,x may --
4 --
be controlled instead of y --
2 --
--
0 | | | | | | |
0 2 4 6 x
Negative correlation

Y -- n = 30 r = - 0.6
6 --
An increase in x may cause a tendency for a --
decrease in y. Example: quality vs. customer 4 --
--
complaints, training vs. rejection 2 --
--
0 | | | | | | |
0 2 4 6 x
Negative correlation, May be Present
Y -- n = 30
There may not be a straight relation 6 --
--
between x and y.upto a certain extent,y 4 --
increase with increase in x and then y start --
decreasing.example:vehicle speed and fuel 2 --
--
efficiency 0 | | | | | | |
0 2 4 6 x
Scatter Diagram - Exercise
The impurity level was suspected to increase on a day from start of the shift to end of
the shift.

A team was constituted to study the problem. The team collected data on the amount of
impurity in the line every half an hour on 5 different days and drew a scatter diagram to
establish the cause and effect relationship. The data is given be below. Construct the
scatter diagram and help the team to arrive at a conclusion.

Operator/No of Hours of work Day1 Day2 Day3 Day4 Day5

0.5 20 40 25 30 42
1.0 40 32 21 35 21
1.5 24 31 35 25 40
2.0 25 22 30 25 20
2.5 42 35 25 45 35
3.0 25 45 45 50 35
3.5 45 22 35 34 42
4.0 19 18 17 22 21
4.5 35 45 50 42 25
5.0 44 33 22 51 35
5.5 44 33 38 22 28
6.0 25 22 24 19 20
6.5 44 52 31 35 47
7.0 42 34 45 34 22
7.5 22 45 48 17 38
8.0 23 45 38 23 45

Draw a Scatter Diagram here for the situation given on the previous page
 Graphs
Description:
• A graph is a picture of a set of numerical data that shows pattern and relationships in
the data. Graphs are best used to get an overall picture of what’s going on. The
exact value of any given dot on a plot is less important than the pattern formed by all
the dots.
• Graphs helps in understanding of the trends etc. so that gap/problem is easily
defined.
• Use data in the form of charts and graphs.
• See relationships pictorially, with sketches and boxes.
• Define the problem clearly
• Demonstrate the importance of the problem
• Show the background of the problem.

When to use Graphs:

When lot of data is available and information need to be presented pictorially for quick
and easy understanding.

Other names for a graph are Chart, Plot & Diagram.

Fundamental graphs

(A) BAR GRAPH

(B) PLOTTED LINE GRAPH
(C) CIRCLE GRAPH
(D) STRIP GRAPH

There are other type of graphs are also available, but these are the most prominently
used fundamental graphs.

A Bar graph is a graph that uses bars of different heights to show how frequently
different data value occur.

Identify categories on
F one axis and quantity on
E other axis and the
D prepare Bars of
appropriate heights
C
depending upon the
B quantity for the
A particular category,
A B C D E F
Bar graphs are used for comparing quantity.
 PLOTTED LINE GRAPH Identify categories on
(1,000 t) one axis and quantity on
other axis and put dots
7 Production at appropriate heights
6 depending upon the
quantity for the
5
particular category, and
then join the dots to plot
4
line graph
3 This type of graph is
particularly useful for
2 tracking changes over a
period of time.
1

(Month) 3 4 5 6 7 8 9 10 11 12 13

Used for observing the quantity changes

CIRCLE GRAPH

%contribution of
each element is
calculated and the
Miscell Miscell
- - circle graph is made.
aneous aneous Complete circle
E A E A shows 100%.

D D
C B C B

Used for showing the breakdown of percentages.

This graph is also called Pie Chart

STRIP GRAPH

A B C D E MISC

% contribution of each element is calculated and the strip graph is made. Complete strip
shows 100%.

Used for showing the breakdown of percentages.

 (104t) BAR GRAPH + BAR GRAPH Shipment

5 Stock

March April May June

Shipment and stock

This type of graph is made for comparison of 2 factors of interest.

BAR GRAPH + STRIP GRAPH Injuries caused

by carelessness
Injuries caused
by poor facilities
and tools
160
140
120
100
80
60
40
20

1975 1976 1977 1978 1979 1980

Injuries in a Plant
This graph is also called Stacked Bar
 STRIP GRAPH + STRIP GRAPH

1985 30% 25% 22% 12% 11%

D
H OL ES E R
TV RADIO
U SE ANC TAP RDE
I
HO PPL CO OTHERS
A RE

1990 27% 16% 15% 15% 27%

SALES DURING PAST FIVE YEARS

Exercise
The number of complaints received by a company are as given below:

Month Number of complaints Month Number of complaints

January 2007 18 January 2008 12
February 2007 22 February 2008 12
March 2007 09 March 2008 15
April 2007 14 April 2008 17
May 2007 12 May 2008 19
June 2007 16 June 2008 13

Prepare appropriate graph for the abovementioned data on next page

Prepare appropriate graph for the given data on this page

 Control Charts

Run Charts
Measurement

Time

What is it?
• Run Charts are representing change in measurement over a sequence or time

How does it Work?

• Gather Data
• Organize Data
– Measurements (y) must be confronted with time or sequence of the events.
• Chart Data
• Interpreting Data

What is its use?

• Determining Cyclic Events and there average character

Example
• Oil consumption of a specific machine over a period of time.
Measurement

Time
 Control charts
Description:
• Statistical tool, showing whether a process is in control or not

How does it Work?

• Define Upper limit, lower limit and medium value
• Draw Chart.
• Gather values and draw them into chart

When to use: when you need to discover how much variability in a process is due to
random variation and how much is due to unique events/individual actions in order to
determine whether a process is in statistical control.
• A control chart is simply a run chart with statistically determined upper control limit
(UCL) and lower control limit (LCL) lines drawn on either side of the process
average.
• Mathematical and graphical method to illustrate performance of a machine or
process or activity over time.
• It has three lines parallel to x-axis.
• One line is in the centre and is called average line. It is firm line.
• Other two lines are at an equidistant from centre line. These are called upper and
lower control limits. These are broken lines.

UCL
Characteristics
Quality

CL
LCL

Time

Variability
• Is the cause of all quality problems.
• Is a natural phenomenon
- in nature
- in industry.
• Its detection depends upon
- Precision of measurement
- Skill of personnel.
• It can be reduced but not eliminated.
• Lower variability means higher possibility of conformance.
Chance vs. Assignable causes

Chance (Common) Causes Assignable (Special) Causes

• Consists of many individual causes. • Consists of just one or few individual
• Any one cause results in only a minute causes.
amount of variability. • Any cause can result in large variability.
• Example : • Example :
- Slight variation in raw material. - Batch of defective material.
- Lack of human perfection in reading - Untrained operator.
instruments and setting controls. - Faulty set up.
• Cannot be economically eliminated. • Easy to detect and generally economical to
eliminate.
• Process follows a predictable (statistical) • No specific pattern.
pattern.

Total Variability

Controlled Variability Uncontrolled Variability

Because of chance causes Because of Assignable causes
Or
Inherent causes
Or
Random causes

• Live with these causes • Use statistical tools to

till up-gradation of work identify, eliminate, or
standard / design / reduce these causes.
machine.

Process control

What it means ?
• Bringing process under the influence of chance causes alone; by identifying and
eliminating ASSIGNABLE CAUSES.

Advantages
• Process operates at its best when under chance causes alone.
• Exhibits minimum variability.
• Predictable process.
• Lower level of Non- Conformance.
• Sampling inspection possible.
Variable Control Charts
Generally applied where the process and their out puts can be characterized by variable
measurements

Outcome Example Control Chart Example

• Shaft O. D. (inches) X for the Average of the Measurement
• Hole distance from reference surface (mm)
• Circuit resistance (Ohms) R Chart for the Ranges of the
• Railcar transit time (hours) Measurement
• Engineering change processing time (hours)

Variable control charts are useful for several reasons

 A quantitative value (e.g. “the diameter is 16.45 mm”) contains more information than
a simple yes-no statement (e.g. “the diameter is within specification”);
 Although collecting variables data is usually more costly than collecting attributes
data (e.g. go/no-go), a decision can be reached more quickly with a smaller sample
size. This can lead to lower total measurement costs due to increased efficiency;
 Because fewer parts need to be checked before making reliable decisions, the time
delay between an “out-of-control” signal and corrective action is usually shorter;
 With variables data, performance of a process can be analyzed, and improvement
can be quantified, even if all individual values are within the specification limits. This
is important in seeking continual improvement.

Attribute Control Charts

Outcome Example Control Chart Example
Vehicle does not leak p Chart for Proportion of
Lamp lights does not light Units Nonconforming
Hole diameter undersized or oversized
(evaluated using a go/no-go gage) np Chart of Number of
Shipment to dealer correct or incorrect Units Nonconforming
Bubbles in a windshield c Chart for number of
Paint imperfection on door Nonconformance per Unit
Errors on an invoice u Chart for Number of
Nonconformities per Unit

Attribute control charts are particularly useful for several reasons:

 Attributes data situations exist in any technical or administrative process, so
attributes analysis techniques are useful in many applications. The most significant
difficulty is to develop precise operational definitions of what is conforming.
 Attributes data are already available in many situations – wherever there are existing
inspections, repair logs, sorts of rejected material, etc. In these cases, no additional
effort is required for data collection. The only expense involved is for the effort of
converting the data to control chart form.
 Where new data must be collected, attributes information is generally quick and
inexpensive to obtain. With simple gauging (e.g., a go/no-go gage or visual
standards), specialized measurement skills are often are not required. There are
many occasions where specialized measurement skills are required especially when
the part measured falls in the “gray” area.
 Much data gathered for management summary reporting are often in attributes form
and can benefit from control chart analysis. Examples include scrap rates, quality
audits and material rejections. Because of the ability to distinguish between special
and common cause variation, control chart analysis can be valuable in interpreting
these management reports

Control Charts

• Variable
- Run-chart
- X- MR chart (Individual observation – Moving range chart)
- X - R chart (Average – Range chart)
- X - R chart (Median – Range chart)
- X - s chart (Average – Standard deviation chart)
- Sloping control chart
- Group control chart
- Pre-control chart

Attribute
- p chart (Proportion defective chart).
- np chart (Number defective chart).
- c chart (Number defect chart).
- u chart (Number defect per unit chart).
X-Bar – R Chart

Formulae for Control Limits

Rules for Interpreting Special Causes of Variation - 1
A lack of control is indicated whenever a single point falls outside the control limits.

Unexpected points indicate a

change in the process

Rules for Interpreting Special Causes of Variation - 2

A lack of control is indicated whenever at least two out of three successive values fall on
the same side of, and more than two sigma units away from the center line

UCL
Zone A
Zone B
Zone C
Zone C
Zone B
Zone A
LCL

Rules for Interpreting Special Causes of Variation - 3

A lack of control is indicated whenever four out of five successive values fall on the
same side of, and more than one sigma unit away from, the center line.

UCL
ZoneA
ZoneB

ZoneC
ZoneC
ZoneB

LCL ZoneA
Rules for Interpreting Special Causes of Variation - 4
A lack of control is indicated whenever at least eight successive values fall on the same
side of the center line
UCL

LCL
Rules for Interpreting Special Causes of Variation - 5
Run is the state in which points occur continually on one side of the central line and the
number of points is called the length of run. Seven –point length of run is considered as
abnormal.

Even if the length of run is under 6 the following cases are considered to be
abnormal.
-- at least 10 out of 11 consecutive points occur on one side of the central line
-- at least 12 out of 14 consecutive points occur on one side of the central line
-- at least 16 out of 20 consecutive points occur on one side of the central line

Rules for Interpreting Special Causes of Variation - 6

Trend when the points form a continuous upward or downward curve ,this is said to
have a trend
Rules for Interpreting Special Causes of Variation - 7
Approach to the control limits considering points which approach the 3-sigma control
limits, if 2 out of 3 points occur outside of the 2-sigma lines, this case is considered to be
abnormal.

Rules for Interpreting Special Causes of Variation - 8

Approach to the central line when most of the points are arranged within the central 1.5
sigma lines(the bisectors of the central line and each of the control limits), this is due to
an inappropriate way of subgrouping. Approach to the central line does not mean a
controlled state, but it means mixing of data with a different population in subgroups,
which makes the width of control limits too wide. At this situation it is necessary to
change the way of sub grouping.

Construction/Interpretation Tips: Control Charts

1. Generally, collect 20-25 groups of samples before calculating the control limits.
2. Upper & lower control limits MUST be statistically calculated. Do not confuse them
with specification limits, which are based on product requirements
3. Management, through support and action, has the opportunity to control/reduce the
natural variation between the control limits.
4. Make sure to select the right type of control chart for the right type of data
5. Data must be kept in the exact sequence as it was gathered, otherwise it is
meaningless.
6. Do not “tweak” the process beyond standard procedures while you are gathering the
data; the data must reflect how it runs naturally.