What

is Sample??

To check the quality of entire batch or lot of the products, a small sample is
selected randomly (arbitrarily) for the inspection. Its random selection (unbiased)
& its size are very important parameters for statistical quality control. As the size
of the sample increases the accuracy of the analysis also increases but the cost
also increases.

Acceptance sampling

This was the statistical quality control method developed by Dodge & Roming
during WWII in USA military. Sample of small group of items is chosen randomly
from entire lot & inspected. Acceptance limit (C) of the number of defects is
considered. Number of defects (d) from the sample is calculated. Depending upon
the number of defects obtained & the acceptance limit kept, either entire lot is
accepted or rejected.

Need of sampling?

• 100% inspection (very accurate but costly) involving all the items is not
possible & will consume lot of time & resource
• 0% inspection is very risky, faulty items may be supplied to the customers
• Middle path must be selected where some percentage of items must be
inspected

Features of a good sample?

1. Optimum size (neither very small nor large)

2. Random selection (no biasing, chosen arbitrarily, representative of lot)
 Terminologies
N= batch/ lot size (100 items)
K= Sample size (4 items)
n= number of samples (3)
C=acceptance number of defects (1)
d= number of defective items/sample

if d<C accept entire lot/batch
if d>C reject entire lot/batch

As we see that fate of entire lot/batch depends upon the acceptance or rejection
of small sample. Hence this number should be chosen very carefully. This number
should neither be large nor small but optimum so that neither it compromise
accuracy nor become very costly or time consuming. There are risks involved in
wrong selection of this number. These are risks either for the producer or the
consumer.

Types of the Risks in Acceptance Sampling

1. Type- I error (Producer’s risk, α type risk)….. good lot is rejected

2. Type- II error (Consumer’s risk, β type risk)….. bad lot is accepted

Sampling Plan

Most important decision in Acceptance Sampling is the proper choice of the

appropriate sampling plan. Sampling plan mentions about the number of
sampling stages, lot size, sample size, number of samples, acceptance number
etc. as explained below.

1. Single sample plan

2. Double sample plan
3. Multiple sample plan
4. Sequential sample plan

 1. Single Sample Plan Interpretations
If d ≤ C accept entire lot/batch
Sample size =k If d>C reject entire lot/batch
Acceptance number =C It is very simple type of plan
Number of defects =d Cost involved is lesser
Risk is involved for both producer n
customer

2. Double Sample Plan Interpretations

Two samples are taken for analysis If d1 ≤ C1 accept the lot without going
C1=acceptance number for 1st sample for 2nd sample
(lower limit) If d1>C2 reject entire lot without going
C2=acceptance number for both for 2nd sample
samples combined (upper limit) If C1<d1 ≤ C2 go for 2nd sample
K1=1st sample size Now after considering 2nd sample
K2=2nd sample size If d2 ≤ C2 accept the lot
d1=number of defects in 1st sample If d2>C2 reject the lot
d2=total number of defects in both
samples

Qn In Double Sampling plan, acceptance limits for first stage & for combined
both stages is 3 & 6. Sample sizes are 10 & 15. Comment on acceptance,
rejection, number of samples considered for the following defects obtained in
first stage & in combined both stages:

a) 4 & 7
b) 2 & 5
c) 7 & 8
d) 5 & 6

Ans Following analysis can be done for the four situations given:

C1=3 C2=6 k1=10 k2=15

 d1= 4 d2=7 d1= 2 d2=5

nd
since C1<d1<C2 go for 2 sample since d1<C1 accept the lot
since d2>C2 reject the lot without going for 2nd sample

d1= 7 d2=8 d1= 5 d2=6

since d1>C2 reject entire lot since C1<d1<C2 go for 2nd sample
without going for 2nd sample since d2 ≤ C2 accept the lot

3. Multiple Sample Plan

• More than two samples can be taken
• There is lower & upper limit considered for acceptance & rejection
• If cumulative number of defects in each stage is lesser than
acceptance limit then accept entire lot
• If cumulative number of defects at any stage crosses upper limit of
rejection then reject entire lot
• If cumulative number of defect is between lower & upper acceptance
limit go for next sampling till decision is made
• If sample size is decreased then number of samples considered
increases, it further increases time consumed & cost though accuracy
is increased
• If the product quality is either very good or very bad then sample
gets either accepted or rejected at early sample stage, in such
situation multiple sample plan is useful & cheaper & better than
single sample plan
• Risk involved in multiple sample plan is lesser though it may be costly
4. Sequential Sample Plan
It is extension of multiple sample plan in which one item at a time is
considered from the lot & decision to accept or reject the lot is taken after
inspection of each item. Sequential sample plan consumes maximum time
& resources.

The acceptance and rejection numbers usually correspond to parallel
sloping lines on the chart of (number nonconforming) versus (number
inspected). Inspection could continue indefinitely, but is usually terminated
after a reasonable number of items have been inspected.














Operating Characteristic Curve (O- C curves)

The operating
characteristic (OC) curve
depicts the discriminatory
power of an acceptance
sampling plan. The OC
curve plots the
probabilities of accepting a
lot versus the fraction
defective. When the OC
curve is plotted, the
sampling risks are obvious.
The OC curve must be analyzed before using a sampling plan.
OC curve terminologies

AQL= acceptance quality
level
RQL= rejectable quality
level (LTPD= lot tolerance %
defects)
AOQ= Average outgoing
quantity
AOQL= Average Outgoing
Quality Limit.
α=Producer’s risk
β=Customer’s risk

P=Percentage defects or fraction of defects= total number of defects in a
sample/sample size
AOQ is average outgoing quantity & it represents the average percentage
defects in outgoing products after inspection is done. It includes both the
accepted & rejected lots.
AOQL - Average Outgoing Quality Limit. The AOQL of a sampling plan is
maximum value on the AOQ curve. It is applicable for defective units,
defects per unit, and defects per quantity. The AOQL is the maximum or
worst possible defective or defect rate for the average outgoing quality.

Acceptance quality level (AQL)
AQL is the value of P up to which the customer accepts the lot or batch as
good. Normally it is 5%. It means if the percentage defects in a sample are
upto 5 then the customer treats the batch as good & accepts it.
If P≤ AQL customer will accept the lot as good
If P>AQL No comment (customer indifferent)

Rejectable quality level (RQL) or LTPD (lot tolerance % defects)
RQL is the value of P beyond which the customer rejects the lot or batch as
bad. Normally it is 10%. It means if the percentage defects in a sample are
beyond 10 then the customer treats the batch as bad & certainly rejects it.
If P< RQL No comment (Customer indifferent)
If P ≥ RQL Customer certainly rejects the lot


 Hence if
1. P≤ AQL customer will accept the lot as good (Producer’s risk)
2. AQL<P<RQL Customer is indifferent & takes no decision
3. If P ≥ RQL Customer certainly rejects the lot (Customer’s risk)

Producer’s Risk (α) & Customer’s Risk (β)

• AQL represents the producer’s risk (α) as the lot is good near this limit. But
if P>AQL then customer may not take a decision to accept it even if the lot
was good. It is risky for the producer. It is also called type-I error. It is
around 5%. It represents the probability that AQL gets rejected.
• RQL/LTPD represents the customer’s risk (β) as the lot is not very good
here. But if P< RQL then the customer may not reject the lot though lot was
not good. It becomes the risk of customer. It is also called type-II error. It is
around 10%. It represents the probability that RQL gets accepted.

Benefit of OC Curve
The operating characteristics (OC) curve tells us how good our samples are
and the probability of accepting defects. Steeper the curve, better the
sampling plan. OC curve plotted for different sampling plans are useful to
compare such plans. It shows the effectiveness of a plan in reducing the
risk. It also keeps the high cost of inspection as low. Discriminating power
of OC curve is reflected in its steepness.
As steepness of OC curve
increases
• Its discriminating power for
sampling plans increases
• Risk associated decrease
• Sample size (k) increases
• Steeper curve is with larger
sample size & is more effective

The discriminating power of any sampling plan depends largely on the size of the
sample. Graph shows the OC curves for sample sizes of 100, 200 and 300 with the
acceptance number remaining in the proportion to the sample size. Graph shows
that the OC curve becomes steeper as the sample size goes up.

If the discrimination power of all the three plans shown in the graph is compared,
it is found that these would accept lots of about 0.7% defectives about 80% of the
time (the approximate cross over points of the three curves).

However, if actual quality fall to 3 percent defectives, the plan with n = 100
accepts lots about 25% of the time, n = 200 about 10% of the time, and n = 300
less than 1% of the time. Therefore, it shows that the plans with larger sample
sizes are definitely more effective. Steeper the curve more effective it is.

Steeper the graph

• Acceptance limit (C) decreases
• Steeper curve is with lesser
acceptance number & is more
effective.
• A sampling plan that
discriminated perfectly between
good and bad lots would have an
OC curve that was vertical or
steep

By making sampling plans more discriminating (increasing sample size) or tighter

(decreasing acceptance numbers), we can approach any desired level of outgoing
quality that we please but at increasing inspection cost.

Zones of O.C. Curve: O.C. curve can be divided into following 3 zones
Acceptance Zone:
In this zone all the batches are accepted, therefore, the O.C. curve should be so
selected that its acceptance zone accepts what is considered to be satisfactory
lot.

Rejection Zone:
In this zone, all the batches are rejected. Hence the O.C. curve selected should be
such that it rejects what is considered to be an unsatisfactory lot.

Zone of Indecision:
This is the zone where there is no clarity whether any particular batch or lot will
be accepted or rejected. This problem can be solved either by adopting 100%
inspection or by taking larger sample, but these will increase the inspection costs.
A batch or lot in this zone is worse than acceptable lot, and better than those
what is considered as unacceptable. Thus its quality is border-line, and practically
does not matter much whether a lot is finally accepted or rejected from this zone.

Qn The maximum percentage defects that a consumer may find definitely

acceptable is called (ESE 2017)
a)AOQL
b)LTPD
c)AQL
d)AOQ

Ans c) AQL stands for Acceptance quality limit, it shows percentage defects below
which customer finds the lot as good & definitely accepts the lot till this limit



Qn Operating Characteristic curve is plotted between

a) Percentage defects & total items

b) Percentage defects & acceptance limit
c) Probability of acceptance & percentage defects
d) Probability of acceptance & total items

Ans c) OC curve plots probability of acceptance with percentage of defective

items

Qn Consider the following features related to the samples chosen for
acceptance sampling

1. Larger sample size

2. More number of samples
3. Randomly chosen sample
Which of the above is/are true for a good sample
a) 1 only
b) 1 & 2 only
c) 1,2 &3
d) 3 only

Ans d) sample size should be optimum & must be chosen randomly

Qn Which of the following is correct related to multiple sampling plan used in

statistical quality control

a) It is cheaper than single sampling plan

b) Sample size is larger
c) It provides cheaper analysis if sample is either good or bad
d) It involves more risk than single sampling plan

Ans c) If the product quality is either very good or very bad then sample gets
either accepted or rejected at early sample stage, in such situation multiple
sample plan is useful & cheaper & better than single sample plan

Qn Acceptance & rejection lines are drawn to analyze the acceptance or
rejection of the lot in

a) Single sampling plan

b) Double sampling plan
c) Multiple sampling plan
d) Sequential sampling plan

Ans d) The acceptance and rejection numbers usually correspond to parallel

sloping lines on the chart of (number nonconforming) versus (number inspected).
Inspection could continue indefinitely, but is usually terminated after a
reasonable number of items have been inspected.

Qn Which of the following is NOT correct related to sequential sampling plan

used in statistical quality control

a) It is special type of multiple sampling plan

b) It is more time & resource consuming
c) Acceptance & rejection slope line are drawn
d) Samples are drawn indefinitely till it gets accepted or rejected

Ans d) Inspection could continue indefinitely, but is usually terminated after a

reasonable number of items have been inspected.

Qn AQL acceptance quality limit represents

a) Customer’s risk
b) Producer’s risk
c) Both customer as well as producer’s risk
d) Neither customer as well as producer’s risk

Ans b) AQL represents the producer’s risk (α) as the lot is good near this limit.
But if Percentage defective items >AQL then customer may not take a decision
to accept it even if the lot was good. It is risky for the producer. It is also called
type-I error. It is around 5%. It represents the probability that AQL gets
rejected.

Qn Statement I Single sampling plan is cheaper to use

Statement II Risk involved is for both customer as well as producer
a) Both statements I & II are individually true & statement II is correct
explanation of statement I
b) Both statements I & II are individually true & statement II is NOT the correct
explanation of statement I
c) Statement I is true but statement II is false
d) Statement I is false but statement II is correct

Ans b) Single sampling plans are cheaper but not very accurate & involve the risk
for both customer as well as producers

Qn Statement I Single sampling plans are used for small lot size
Statement II Small sample size can be taken in multiple sampling plan
a) Both statements I & II are individually true & statement II is correct
explanation of statement I
b) Both statements I & II are individually true & statement II is NOT the correct
explanation of statement I
c) Statement I is true but statement II is false
d) Statement I is false but statement II is correct

Ans d) Single sampling plan can be used for any size of the lot or batch

Qn Consider the following statements related to Operating Characteristic curves

drawn in acceptance sampling

1. These are drawn between probability of acceptance & percentage defects

2. Are used to discriminate between sampling plans
3. These must be flat to have better discrimination

Which of the above are correct statements?

a) 1 & 2 only
b) 2 & 3 only
c) 1 & 3 only
d) 1, 2 & 3

Ans a) Steep OC curves have better discrimination power for various sampling
plans