All decisions are made after drawing a sample from a population. The process of sampling involves risk: risk of deciding that sample is different from population when in fact it is not and risk of concluding that sample is representing population when in fact it is not.
1. 1
Naresh Chawla
Visiting Professor –TQM, Punjab Engineering College (Deemed to be University)
Principal Trainer, TQM & Operational Excellence
“Sampling”
Basic Concepts of Sampling
2. Sampling
2Naresh Chawla, Principal Trainer, TQM & Operational Excellence
▪ Sampling
o A process of choosing a
representative portion
from a population
Representative means it is
covering all the
characteristics of the
population
▪ Objective: To draw conclusion/inference for the
population based on the sample analysis
3. Need for Sampling
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 3
3. 100% testing may not be possible or
sometimes even not practical
2. Faster results
1. Limited resources
4. Application
▪ Social sciences: for
various surveys
▪ Commerce & Trade :
assessing mean quality
of bulk material
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 4
▪ Manufacturing: acceptance of incoming
material /BoIs, intermediates, finished
products/goods, packaging material
7. ▪ Mistakes made during data acquisition
Systematic error (or bias)
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 7
▪ Selection bias
▪ non response error
(information bias)
8. ▪ Errors caused by chance
in selecting a random
sample
o Finding a difference (in
sample and population)
that doesn't exist really
o Not finding a difference
that actually exists
between sample and
population
let us understand this by a
grid on next slide
Random Sampling Errors
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 8
9. Random Sampling Errors
In reality
Decision
based on
sample
Sample belongs
to population
Sample doesn’t
belong to
population
Sample
belongs to
population
Sample
doesn’t
belong to
population
Type II Error
Type I Error
9Naresh Chawla, Principal Trainer, TQM & Operational Excellence
10. ▪ Type I Error
o Also known as a error
o General level of acceptance is 5% (or 0.05)
▪ Type II Error
o Also known as b error
o (1-b) is also known as power of the plan
(ability to find a difference when actually it
is there)
Random Sampling Errors
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 10
11. Minimizing Sampling Errors
▪ Sample should be representative i.e. it
should have all the characteristics of the
population
▪ Sample size should be adequately large.
▪ Sample size depends upon the inherent
variation in the population and the
accuracy required
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 11
12. How much to sample?
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 12
1.96 = Za/2 for 95 confidence level
s = Standard Deviation
E = Margin of Error
13. Early uses of Sampling
▪ Social Sciences
o Market Research
o Economics
o Exit/opinion polls
o Employee survey
o Customer survey
▪ Agriculture
o Estimating produce
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 13
14. Application in Manufacturing
You just received a shipment of 2000 units
from a new supplier
How will you decide?
Is the shipment good enough?
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 14
16. Arms & Ammunitions
▪ Acceptance Sampling was first introduced to the
Ordnance Department of the U.S. Army during
World War II,
▪ It was impossible to qualify every bomb to
determine if it would work in the field
▪ Harold F. Dodge and Harry G. Romig of Bell
Research lab developed the standards.
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 16
17. Automotive Industry
IATF 16949: 2016
9.1.1.1 Monitoring & Measurement of
Manufacturing Processes
o Organization shall perform process studies to verify
process capability
o If it is not possible to demonstrate process
compliance through process capability use alternate
methods such as batch confirming to specifications
o Sampling plan, acceptance criteria and records of
actual measurement values and/or test results for
variables may be part of customer’s part approval
process
17Naresh Chawla, Principal Trainer, TQM & Operational Excellence
18. Acceptance Sampling
▪ Process of evaluating a portion
(sample) of the product from a lot
(population) for the purpose of
accepting or rejecting the entire lot.
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 18
19. ▪ Plans for attributes were first
developed for the US Army
during World War II.
▪ MIL-STD-105D was first
issued by the U.S.
government in 1963.
Various Standards for Acceptance
Sampling
Naresh Chawla, Principal Trainer, TQM & Operational Excellence
▪ MIL-STD-105D was revised to MIL-STD-105E in
1989. Its use has been discontinued since 1995
▪ MIT-STD-105D forms the basis of ANSI/ ASQC Z
1.4 -1981:Sampling Procedure and Tables for
Inspection by Attributes. Version 2008 is the
current version of this standard
19
20. ▪ MIL-STD-105D also forms the basis of ISO
2859 (issued by ISO)
▪ Standards for variable inspections were also
introduced in the form of ANSI/ASQ/Z1.9
and ISO 3951-1:2013 is complementary to
ISO 2859-1:2008
▪ In India IS 2500 series is used for
acceptance sampling
Various Standards for Acceptance
Sampling
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 20
21. ▪ IS 2500 is in three parts:
o IS: 2500 (Part-1) : 2000 –for inspection
by attributes based on AQL for Lot by Lot
Inspection (aligned fully with 3rd revised
edition of ISO 2859 (Part-1) : 1999)
o IS: 2500 (Part-2): 1965 – For Inspection
by variables (based on MIL-STD-414)
o IS: 2500 (Part-3) : 1995 –Attribute
Sampling Plans based on LQ for isolated lot
inspection (protecting consumer risk)
Various Standards for Acceptance
Sampling
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 21
22. ▪ ISO 2859-1 or ANSI/ASQ Z1.4 for
series of lots; ISO 2859-2 or
ANSI/ASQ-Q3 for isolated lots)
▪ Material to be sampled:
1. Unitary material (finished product,
leaflets, test tubes, vials, stoppers,
medical devices etc
2. Particulate material (Powders of
API, granulate, blend etc
3. Bulky-continuous material (water,
liquid solution, viscous solutions,
gel, cream, compressed gas etc.
FDA Guidelines for Pharma
companies
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 22
23. ▪ In medical device manufacturing the key
point is to have the plan accept on zero
defectives, C = 0 plans by Nicholas L.
Squeglia
▪ This point is not FDA but legalese. It is
based on past lawsuits
Medical devices
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 23
24. Bulk Sampling
▪ Bulk materials are essentially continuous
and do not consist of populations of
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 24
o Discrete
o Constant
o Identifiable
o unique units or
o terms
that may be drawn
into the sample
25. Bulk Sampling
▪ Type A Bulk material
o a pile, a truck, a
railroad car, or a
conveyer belt
▪ Type B Bulk material
o Bags of fertilizers
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 25
Iron ore and coal belong to
Type A material
26. Objective of Bulk Sampling
▪ Characterization of the material in place as to amount,
content or value (as in a natural deposit)
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 26
1. For grading and need for
further processing
2. Determination of mean value
for purposes of taxation or
payment
3. Determination of properties
that must be known so that
the end use will be
appropriate
4. Control during processing
▪ Acceptance on a lot-to-lot basis
Parameters UOM Values
Moisture %ar 10.0
Volatile
matter
%ar
%daf
23.5
30.3
Fixed Carbon %ar 54.1
Ash
%ar
%dry
12.4
13.8
Net CV KJ/Kg 25050
Sulphur %ar 0.48
Chlorine %ar 0.01
27. Foods, food production & food safety
▪ FV-Q Visible defects in fruits/vegetables - ISO 2859-1:1999: Sampling
procedures for inspection by attributes
▪ MI-Q Fat contents in milk - ISO 3951-1:2013: Sampling procedures for
inspection by variables
▪ FV-P Pesticides Residues in fruits/vegetables - CAC/GL 40-1993
C O D E X A L I M E N T A R I U S international food standards,
Naresh Chawla, Principal Trainer, TQM & Operational Excellence 27
Fruits/
Vegetabl
es
(FV)
Fats/
Oil
(FO)
Fish/
Fishery
product
(F)
Milk/Milk
product
(MI)
Meat/
Meat
product
(M)
Natural
Mineral
Waters
(MW)
Cereals
(C)
Qualitative/Quantitative
Characteristics / Sensory
inspection (Q)
FV-Q FO-Q F-Q MI-Q M-Q MW-Q C-Q
Food Hygiene (FH) FV-FH nr* F-FH MI-FH M-FH MW-FH nr*
Pesticide residues (P) FV-P FO-P nr* MI-P M-P nr* C-P
Contaminations (C) FV-C1/2 FO-C F-C MI-C M-C MW-C C-C
Residues of veterinary drugs
(R)
nr* FO-R F-R MI-R M-R nr* nr*
* nr – Not Required