Variability

&
Statistical Quality Control
 WORK AS A PROCESS

S C
U U
P S
P T
L Inputs Process Outputs O
I M
E E
R R
S S

2
 POPULATION AND SAMPLE
• The entire set of items is called the Population.
• The small number of items taken from the population to
make a judgment of the population is called a Sample.
• The numbers of samples taken to make this judgment is
called Sample size.

POPULATION SAMPLE OF
SIZE THREE

3
 REPRESENTATIVE SAMPLES

• For conclusions to be valid, samples must be representative.

– Data should fairly represent the population or process
– No systematic differences should exist between the data you
collect and the data you don’t collect

4
 Sampling Strategy
Where Are You Standing?
Process Data

Population Data

Population Approach
• Make probability statements about the population from the sample
» “I have 95% confidence that the mean of the population is between 1.5 and 2.5 seconds.”
• Use sample size formula

Process Approach
• Assess the stability of the population over time
» Are shifts, trends, or cycles occurring?
» Do I take a special or common cause variation approach to process improvement?
• Use rational subgrouping

5
 Sampling Strategy : Random Sampling
Population Study

Population Sample Description

N n

Each unit (“X”) has

an equal probability
XXX of being selected in
XXXXX X X X
a sample
XXXXX
XXX

6
 Sampling Strategy : Stratified Random
Sampling
Population Study

Population Sample Description

Segments Units Randomly sample

within a stratified
Large LLLLL
L category or group
MMM
Medium MMM
MMM MMMM Sample sizes for
MMM each group are
SS
SSSS generally proportional
Small SSSS to the relative size of
SS the group

7
 Sampling Strategy : Systematic Sampling
Process Study

Process Sample Description

X X X X X X X X X X X – Sample
every nth
one
(e.g., 4th
one)
Must select sampling frequency

8
 Sampling Strategy : Rational Subgrouping
Process Sampling

Process
Hour 1 Hour 2 Hour 3
Sample
Subgroup of
X X X X X X X X X X X X X X X X X X X X X X X X samples

Description
Sample at point “A” in the process every Xth hour
9
 POPULATION, SAMPLE AND DATA

NO ACTION

Measurement /
Random Sampling Observation
POPULATION Sample Data

ACTION

10
Data Collection Plan Features
Data Collection Plan Features,
cont.
SAMPLE SIZE RULES OF THUMB
Statistic or Recommended Minimum
Chart Sample Size (n)

Frequency plot 50
(Histogram)

Pareto chart 50

Scatter plot 24

Control chart 24

14
 6.50 6.55 7.00

6.55 a.m.  5 minutes.

This man wants to reach his work place by 6.55 a.m.. But he
can not do so, exactly at 6.55 a.m. daily. Sometimes he reaches
earlier (but almost never before 6.50 a.m.). Sometimes he
reaches later (but almost never after 7.00 a.m.).
WHY ?
15
 THIS IS BECAUSE....
OF CERTAIN FACTORS WHICH
• Affect the time he takes
• He cannot control
• Vary randomly
e.g. The traffic you encounter under normal course of travel

THE VARIATION THAT OCCURS DUE TO THESE KIND

OF FACTORS IS CALLED INHERENT VARIATION OR
COMMON CAUSE VARIATION OR WHITE NOISE.
e.g.. m/c vibration,tool wear etc.

16
UNDER NORMAL SCHEME OF
OPERATION
Inherent Variability

Target
Value

Lower Higher
Specification Specification

TOTAL PERMISSIBLE VARIABILITY

or
Tolerance

17
TODAY HE IS EARLY ! PROBABLY BECAUSE :
• His watch was running fast.
WHY ? • He got a lift.
• His bus driver took a shortcut.
• He stayed over in the colony.
• He had some important work to
be finished before 7.30.
These causes are characteristic of
6.30 a specific circumstance and do
not occur in the normal scheme
of actions.

Variation due to these types of

reasons is called assignable or
special cause variation or black
noise
18
 GRAPHICAL DISPLAY OF
VARIABILITIES
Inherent Variability

Target
Value

Assignable Assignable
Variability Variability

Lower Higher
Specification Specification

TOTAL VAR I A B I L I T Y 19
 WHAT
WHAT IS
IS THE
THE MEASURE
MEASURE
OF
OF CENTRAL
CENTRAL TENDENCY
TENDENCY OF
OF A
A SET
SET
OF
OF NUMBERS?
NUMBERS?

• There are three ways in which Central

Tendency of Numbers can be measured.
• These are the 3 M’s
MEAN
MEDIAN
MODE
20
 Arithmetic Mean

• The arithmetic mean is the sum of a set of

observations, positive, negative or zero,
divided by the number of observations. If we
have “n” real numbers x1 , x 2 , x 3 , ......., x n ,
their arithmetic mean, denoted byx , can be
expressed as: i n

x
x1  x 2  x3  .............  x n (X  X )
i
n i 1
 Median

• Median is that variable which divides the group of data into two equal
parts ,one part comprising all the values greater and the other ,all the
values less than the median.
• Arrange the data in ascending or descending order .
• If ‘n’ is odd median is the middle value of the observations
• If ‘n’ is even, median is given by the arithmetic mean of the two
middle terms
 Mode

• Mode is the value of a distribution for which the

frequency is maximum. In other words, mode is the
value of a variable, which occurs with the highest
frequency.

• So the mode of the list (1, 2, 2, 3, 3, 3, 4) is 3. The

mode is not necessarily well defined. The list (1, 2, 2,
3, 3, 5) has the two modes 2 and 3.
 WHAT IS VARIATION ?
• No two things in nature are alike.
• This is also true for manufactured products.
• This dissimilarity between two products for the
same characteristic is called variation.
• The variation may be or can be made to be so
small so as to make the product SEEM similar.
• When we say that 2 things are similar we actually
mean that it is not possible to measure the variation
present within the accuracy of the existing
measuring equipment.
• Variation between 2 products are compared for
SIMILAR features or characteristics.

24
HISTOGRAMS: VARIATION at a POINT
OF TIME
DEFINITION A Histogram shows the shape, or
distribution, of the data by displaying
how often different values occur.
EXAMPLE “Time Taken for Approval”

Time Taken for Approval

25
 Measures of Dispersion

 The extent of the spread of the values from the mean

value is called Dispersion.

 The measures of Dispersions are

– Range (R)
– Standard Deviation (s)
– Variance (s2)
– Co-efficient of Variation (CV)

 Standard deviation is the most commonly

used measure of dispersion. 26
 Measures of Dispersion

• Dispersion or spread is the degree of the

scatter or variation of the variables about a
central value.
• The degree to which numerical data tend to
spread about an average value is called the
variation or dispersion of the data.
 Variability
8
7
10
8
• Deviation = distance between 9

observations and the mean (or RAM

average)
Observations Deviations
10 10 - 8.4 = 1.6
9 9 – 8.4 = 0.6
8 8 – 8.4 = -0.4
SHYAM
8 8 – 8.4 = -0.4
7 7 – 8.4 = -1.4
averages 8.4 0.0
 Variability

• Deviation = distance between

observations and the mean (or RAM

average)
Observations Deviations
7
7 7 – 6.6 = 0.4 6
7 7 – 6.6 = 0.4 7
7 7 – 6.6 = 0.4 7
6
6 6 – 6.6 = -0.6 SHYAM
6 6 – 6.6 = -0.6
averages 6.6 0.0
 Variability
8
7
10
8
• Variance = average distance between 9

observations and the mean squared Ram

Observations Deviations Squared Deviations

10 10 - 8.4 = 1.6 2.56
9 9 – 8.4 = 0.6 0.36
8 8 – 8.4 = -0.4 0.16
Shyam
8 8 – 8.4 = -0.4 0.16
7 7 – 8.4 = -1.4 1.96
averages 8.4 0.0 1.0 Variance
 Variability

• Variance = average distance between

observations and the mean squared Ram

Observations Deviations Squared Deviations

7
7 6
7 7
7 7
6 Shyam
6
6
averages
 Variability

• Variance = average distance between

observations and the mean squared Ram

Observations Deviations Squared Deviations

7
7 7 - 6.6 = 0.4 0.16 6
7 7 - 6.6 = 0.4 0.16 7
7 7 - 6.6 = 0.4 0.16 7
6 Shyam
6 6 – 6.6 = -0.6 0.36
6 6 – 6.6 = -0.6 0.36
averages 6.6 0.0 0.24 Variance
 Variability

• Standard deviation = square root of

variance Ram

Variance Standard
Deviation
Ram 1.0 1.0
Shyam 0.24 0.4898979 Shyam

But what good is a standard deviation

 Standard Deviation

The square root of the sample variance is the sample standard

deviation.

  x  x 2

s  s2 
n 1
“sample standard deviation”
This is a measure of standard (typical) deviation from the mean.
 THE NORMAL CURVE

Smooth curve
Center of the bar interconnecting the
center of each bar

 Units of
Measure

35
 “Normal” bell shaped curve

Add up about 30 of most things

and you start to be “normal”

Normal distributions are divide up

into 3 standard deviations on
each side of the mean

Once your that, you

know a lot about
what is going on
 THE NORMAL CURVE

The world tends to

be bell-shaped

Even very rare Fewer Most Fewer Even very rare

outcomes are in the outcomes in the outcomes are
possible “tails” occur in the “tails” possible
(probability > 0) (lower) middle (upper) (probability > 0)
 THE NORMAL CURVE

Here is why: Even outcomes that are equally

likely (like dice), when you add
them up, become bell shaped
 THE NORMAL CURVE
Specific Characteristics

34.13% 34.13%

13.60% 13.60%
2.14% 2.14%

0.13% 0.13%

-3s -2s -1s X +1s +2s +3s

68.26%

95.46%

99.73%

68.26% Fall Within +\- 1 Standard Deviation

95.46% Fall Within +\- 2 Standard Deviation
99.73% Fall Within +\- 3 Standard Deviation
39
 THE NORMAL CURVE
• If the frequency distribution of a set of values is
such that :

– 68.26% of the values line within ±1s from the mean

AND
– 95.46% of the values line within ±2s from the mean
AND
– 99.73% of the values line within ±3s from the mean

Then the distribution is normal.

NORMAL DISTRIBUTION IS
CHARACTERISED BY A BELL SHAPED
CURVE.
40
 WhyWhy
normality is important
is Normality Important?

• Most common statistical terms and analysis tools

are based on a normal data distribution
• Tells us whether the process is under the influence
of common or special causes – a normally
distributed data set is said to be under the influence
of common causes
• If we use these tools with a data set that is not
normal, the accuracy of the tools may be
compromised

41
 Quality Improvement and Statistics

• Quality Improvement
Quality improvement is the reduction of
variability in processes and products.

Alternatively, quality improvement is also

seen as “waste reduction”.
 Types of
Statistical Quality Control

Statistical
Quality Control

Process Acceptanc
Control e Sampling

Variable Attributes
s Charts Charts
Statistical Process Control - Identify and Reduce
Process Variability
Lower Upper
specification specification
limit limit

(a) Acceptance
sampling –[ Some bad
units accepted; the
“lot” is good or bad]
(b) Statistical process
control – [Keep the
process in “control”]

(c) cpk >1 – [Design a

process that is in
control]
 Statistical Quality Control (SPC)
• Measures performance of a process
• Uses mathematics (i.e., statistics)
• Involves collecting, organizing, & interpreting
data
• Objective: provide statistical signal when
assignable causes of variation are present
• Used to
– Control the process as products are produced
– Inspect samples of finished products
Quality Improvement and Statistics

• Statistical process control is a

collection of tools that when used
together can result in process stability
and variance reduction.
 Two Types of Variation

Common Cause Special Cause

• Predictable • Not Predictable
• Routine • Exceptional
• No Assignable Cause • Assignable Cause
• Expected • Unusual
 Stable vs Unstable Processes

A stable (or “in control”) process is one in

which the key process responses and product
properties show no signs of special causes.

An unstable (or “out of control”) process has

both common and special causes present.
Common
Causes

Special Causes

49
Causes of Variation
Partial list of special causes
Partial list of common causes
 Chance & Assignable causes of
Variation
• A process that is operating with only chance
causes of variation present is said to be in
statistical control.
• A process that is operating in the presence of
assignable causes is said to be out of control.
• The eventual goal of SPC is reduction or
elimination of variability in the process by
identification of assignable causes.

SPC uses samples to identify that special causes have occurred

Histograms do
not take into
account changes
over time.

Control charts
can tell us when
a process
changes

54
 Introduction to Control Charts
Basic Principles
 Process Control:
Three Types of Process Outputs
(a) In statistical control and
capable of producing
within control limits. A
process with only natural
Frequency causes of variation and
capable of producing
within the specified control
Lower control limit limits. limit
Upper control
(b) In statistical control, but not
capable of producing within
control limits. A process in control
(only natural causes of variation
are present) but not capable of
producing within the specified
(c) control limits; and
Out of control. A process out
Size of control having assignable
(Weight, length, speed, etc. causes of variation.
)
 Introduction to Control Charts
Basic Principles
A typical control chart has control limits set at values
such that if the process is in control, nearly all points
will lie within the upper control limit (UCL) and the
lower control limit (LCL).
 Introduction to Control Charts
Basic Principles
 Design of a Control Chart

• Control limits can be set at 3 standard

deviations from the mean in both
directions.
• “3-Sigma Control Limits”
UCL = 74 + 3(0.0045) = 74.0135
CL= 74
LCL = 74 - 3(0.0045) = 73.9865
Design of a Control Chart
 Design of a Control Chart
Rational Subgroups
• Subgroups or samples should be selected
so that if assignable causes are present,
the chance for detection of differences
between subgroups will be maximized,
while the chance for detection of
differences due to these assignable causes
within a subgroup will be minimized.
Subgroups Design of a Control Chart
 Design of a Control Chart
Rational Subgroups

Constructing Rational Subgroups

• Select consecutive units of production.
– Provides a “snapshot” of the process.
– Good at detecting process shifts.
• Select a random sample over the entire
sampling interval.
– Good at detecting if a mean has shifted
– out-of-control and then back in-control.
 Important uses of the control chart

1. Most processes do not operate in a state of statistical

control
2. Consequently, the routine and attentive use of control
charts will identify assignable causes. If these causes
can be eliminated from the process, variability will be
reduced and the process will be improved
3. The control chart only detects assignable causes.
Management, operator, and engineering action will be
necessary to eliminate the assignable causes.
 Important uses of the control chart

One goal of using a Control Chart is to achieve and

maintain process stability.

Process stability is defined as a state in which a

process has displayed a certain degree of consistency
in the past and is expected to continue to do so in the
future.

This consistency is characterized by a stream of data

falling within control limits
 Introduction to Control Charts
Basic Principles
Types of the control chart
• Variables Control Charts
– These charts are applied to data that follow a
continuous distribution.
• Attributes Control Charts
– These charts are applied to data that follow a
discrete distribution.
 Types of the control charts

Variables control charts

 Variable data are measured on a continuous
scale. For example: time, weight, distance or
temperature can be measured in fractions or
decimals.
 Applied to data with continuous distribution
 Types of the control charts
Attributes control charts
 Attribute data are counted and cannot have
fractions or decimals. Attribute data arise
when you are determining only the presence
or absence of something: success or failure,
accept or reject, correct or not correct.
 For example, a report can have four errors or
five errors, but it cannot have four and a half
errors.
 Applied to data following discrete
distribution
 Commonly Used Control Charts

• Variables data
– x-bar and R-charts
– x-bar and s-charts
– Charts for individuals (x-charts)
• Attribute data
– For “defectives” (p-chart, np-chart)
– For “defects” (c-chart, u-chart)

70
 Control Chart Selection
Quality Characteristic
variable attribute
defective defect
no
n>1? x and MR
yes constant
yes constant
p or sampling
sample
np unit?
n>=10 no size?
x and R
yes no
no
yes
p-chart with c u
x and s variable sample
size
 Example
An automobile assembly plant
collected cylinder inside diameter
measurements to assess the quality of Dia 600 + 3 mm

the process with a critical dimension

having.

Five cylinders were measured from

each of four shifts daily for five days.

The five samples that comprise each

subgroup were selected within a short
period of time to minimize variation
from one camshaft to the next.
 Control limits
Control limits represent the limits of variation that should
be expected from a process in a state of statistical control.

Specification limits? Control limits

 Specification limits

Limits that are established for the process - they do not

reflect how the process is actually performing.
Specification limits are based on the customer
requirements and detail how one wants the process to
behave.

Upper Specification Limit = ??

Lower Specification Limit = ??
 Control Chart Construction
• Select the process to be charted;
• Determine sampling method and plan;
– How large a sample needs to be selected? Balance the time and cost to
collect a sample with the amount of information you will gather.
– As much as possible, obtain the samples under the same technical
conditions: the same machine, operator, lot, and so on.
– Frequency of sampling will depend on whether you are able to discern
patterns in the data. Consider hourly, daily, shifts, monthly, annually,
lots, and so on. Once the process is “in control”, you might consider
reducing the frequency with which you sample.
– Generally, collect 20-25 groups of samples before calculating the
statistics and control limits.
– Consider using historical data to establish a performance baseline.
 Control Chart Construction
• Initiate data collection:
– Run the process untouched, and gather sampled data.
– Record data on an appropriate Control Chart sheet or
other graph paper. Include any unusual events that
occur.
• Calculate the appropriate statistics and control
limits:
– Use the appropriate formulas.
• Construct the control chart(s) and plot the data.
 X-Bar/ Range Chart
• Shows sample means over time
• Monitors process average
• Example: Weigh samples of coffee & compute means
of samples; Plot
• The X-bar chart monitors the process location over
time, based on the average of a series of
observations, called a subgroup.
• X-bar / Range charts are used when you can
rationally collect measurements in groups
(subgroups) of between two and ten observations.
X Bar Chart

Range Chart
 X-Bar/ Range Chart

• Each subgroup represents a "snapshot" of the

process at a given point in time. The charts' x-axes
are time based, so that the charts show a history of
the process.
• For this reason, data should be time-ordered; that is,
entered in the sequence from which it was
generated.
• If this is not done, trends or shifts in the process may
not be detected, but instead attributed to random
(common cause) variation.
 Control Chart for Samples of 9 Boxes

Variation due
to assignable
causes
17=UCL
Variation due
16=Mean
to natural
15=LCL causes

Variation due
1 2 3 4 5 6 7 8 9 10 11 12 to assignable
Sample Number causes
Out of
control
 R Chart

• Shows sample ranges over time

• Difference between smallest & largest values in
inspection sample
• Monitors variability in process
• Example: Weigh samples of coffee & compute
ranges of samples
 Steps to Follow When Using Control
Charts
1. Collect 20 to 25 samples of n=4 or n=5 from a
stable process and compute the mean.
2. Compute the overall means, set approximate
control limits,and calculate the preliminary
upper and lower control limits.If the process is
not currently stable, use the desired mean
instead of the overall mean to calculate limits.
3. Graph the sample means and ranges on their
respective control charts and determine whether
they fall outside the acceptable limits.
 Steps to Follow When Using Control
Charts - continued
4. Investigate points or patterns that indicate
the process is out of control. Assign causes
for the variations.
5. Collect additional samples and revalidate the
control limits.
 X-Bar and R Control Chart Construction
• Step 1 - Determine the data to be collected. Decide
what questions about the process you plan to answer.
Step 2 - Collect and enter the data by subgroup. Collect
20 to 25 samples of n=4 or n=5 from a stable process
A subgroup is made up of variables data that represent a
characteristic of a product produced by a process. The
sample size relates to how large the subgroups are. Enter
the individual subgroup measurements in time sequence
in the portion of the data collection section of the Control
Chart
X-Bar and R Control Chart Construction
• STEP 3 - Calculate and enter the average for each subgroup. Use the
formula below to calculate the average (mean) for each subgroup
and enter it on the line
• Calculate the grand mean of the subgroup’s average. The grand
mean of the subgroup’s average (X-Bar) becomes the centerline for
the upper plot.
 X-Bar and R Control Chart
Construction
• STEP 4 - Calculate and enter the range for
subgroup.
• Calculate R Bar to get the centre line of R chart
 X-Bar and R Control Chart Construction

• STEP 5 - Calculate the control limits for both X

Bar and R chart
• STEP 6 – Plot on the chart the limits for both X
Bar and R charts along with the centre lines.
• STEP 7 – With respect to each sub group mean
and range, plot the points on the X Bar chart
and R chart.
 X Bar Chart Control Limits

UCL x  x  A  R From
Table 6.1

LCLx  x  A  R Range for

sample i
Mean for
n sample i
 xi n
 Ri
x  i  i 1
R 
n n
# Samples
 R Chart Control Limits

UCL R  D 4 R
From Table S6.1
LCL R  D 3R

n Range for Sample

 Ri i
R  i 1
n # Samples
Factors for Computing Control
Chart Limits
Table 6.1
Sample Mean Upper Lower
Size, n Factor, A2 Range, D4 Range, D3
2 1.880 3.268 0
3 1.023 2.574 0
4 0.729 2.282 0
5 0.577 2.115 0
6 0.483 2.004 0
7 0.419 1.924 0.076
8 0.373 1.864 0.136
9 0.337 1.816 0.184
10 0.308 1.777 0.223
12 0.266 1.716 0.284
0.184
Step 8- Select the scales and plot the control limits,
centerline, and data points, in each plotting area.

The scales must be determined before the data

points and centerline can be plotted. Once the upper
and lower control limits have been computed, the
easiest way to select the scales is to have the current
data take up approximately 60 percent of the vertical
(Y) axis.

The scales for both the upper and lower plotting areas
should allow for future high or low out-of control
data points.
Plot each subgroup average as an individual data point in
the upper plotting area. Plot individual range data points in
the lower plotting area
UCL

CENTRE LINE

LCL
 Interpretation
• Always look at the Range chart first. The
control limits on the X-bar chart are derived
from the average range, so if the Range chart
is out of control, then the control limits on the
X-bar chart are meaningless.
• Look for out of control points. If there are any,
then the special causes must be immediately
eliminated..
R Chart - Spread
 Interpretation
• There should be more than five distinct values
plotted, and no one value should appear more
than 25% of the time.
• If the range values are repeating too often, then
you have inadequate resolution of your
measurement system
• In this case, you'll have to look at how you
measure the variable, and try to measure it more
precisely.
• Once the effect of the out of control points from the
Range chart is removed, look at the X-bar Chart.
Mean and Range Charts
Complement Each Other
X bar Chart - Center

R Chart - Spread
 Control Chart Interpretation:

• A process can be characterized by:

– Examining its behavior during a sufficiently brief interlude of
time
– Examining its behavior across a greater expanse of time.

• Stable process: one which performs with a high degree of

consistency at an essentially constant level for an
extended period of time
– “In-control”
• A process that is not stable is referred to as being in an
out-of-control state
 Typical Out-of-Control Patterns
• Point outside control limits
• Sudden shift in process average
• Cycles
• Trends
• Hugging the center line
• Hugging the control limits
• Instability

100
The three standard deviations are sometimes identified by
zones.
Each zone’s dividing line is exactly one-third the distance
from the centerline to either the upper control limit or the
lower control limit
 Shift in Process Average
Six or more points in either side of the X Bar chart .
This test looks for long series of consecutive points without a change
in direction.

102
Run of two points above 2 Sigma line

103
 Cycles
Fourteen or more points in either side of the X Bar chart following a
cyclical pattern.
.

104
 Trend
Six or more points in either side of the X Bar chart .

105
Patterns to Look for in Control Charts
EXERCISE 1: A team collected the variables
data recorded in the table below.
Use these data to answer the following
questions and plot a Control Chart:
1.What type of Control Chart would you use with these
data?
2. Why?
3. What are the values of X-Bar for each subgroup?
4. What are the values of the ranges for each subgroup?
5. What is the grand mean for the X-Bar data?
6. What is the average of the range values?
7. Compute the values for the upper and lower control
limits for both the upper and lower plotting areas.
8. Plot the Control Chart.
9. Are there any signals of special cause variation? If so,
what rule did you apply to identify the signal?