You are on page 1of 20

Name four non-random sampling method

•J
•R
•C
•Q

Name four random sampling method


•S
•S
•S
•C

1
The tolerance level of risk
• Consumer risk
– The risk to accept defective products; 5% or (α<0.05) is
often used for nonfatal risks.
– It translate to 95% confidence to reject the defective
products.

• Vendor risk
– Has higher tolerance level depending on situations

2
Sample & Population

• Population – Total
quantity from which
sample is obtained sample
• Sample - Portion that
represents the whole
• Sample size – Number
of sampling units in a
sample

3
The closeness of estimation

• Population mean u
– the true value that we will estimate
• Sample mean
– Obtained after analysis of samples
• With sound sampling procedure
– larger sample size will allow a better estimation of u
using
– So let’s sample 100% of the population
4
The optimum sample size

Population mean
100%
95%
Sample mean curve

50%

0%
0% 4% 10% 100%

Sample size (% of population) 5


Sampling for attributes
• You are a manager of plant that produce
tomato juice
• 21 CFR § 156.145 : tomato juice can contain
no more than two “Pieces of peel 3.2
millimeters (0.125 inch) or greater in
length” per 500 mL.
• You are wondering if products in a lot meet
the standard

• There is only one way to find out

6
Decide your sample size

• The complicated way:


− follow ANSI / ASQC Z1.4-2003:
Sampling Procedures and Tables for
Inspection by Attributes

• The easy and safe way:


− follow 21 CFR § 156.3

7
21 CFR § 156.3

Lot size Sample size Acceptance number


<4800 13 2
4801-24000 21 3
24001-48000 29 4
48001-84000 48 6*
84001-140000 84 13

Note: Cans are the containers in the lot. Each contains less than 2.2 lb of juice

* If you have more than 6 defective cans, reject the lot

8
Sampling plan for variables
• You are a manager of plant that produce
grape juice
• Customers are concerned about fructose
in their diet which may cause high blood
pressure and high blood triglyceride
• You like to evaluate fructose content in a
lot of products

• Time for another lab work!

9
Start from a preliminary experiment
• Population mean (the true value) u
• Pilot (preliminary) experiment
– Sample size n ; sample mean

10
Leaps of faith

Now, sample size n can be resolved 11


The final equation

Relative accuracy: set by yourself, at 0.05 or 1%, your sample mean will fall
in u±5%u range

12
Calculation !
• A preliminary experiment for n=10 bottle of juice
– Mean = 6.3 (% w/v) SD = 1.2 CV = 19%

• Desired significance level α=0.05; desired relative


accuracy 0.05%. Calculate sample size
• Find t on t-distribution table

tα, n-1= t0.05, 9 = 1.83

13
Shall We Review ?
• Reliability of analytical data
– Accuracy
– Precision
• Data reporting
– Mean (mean of replicated tests of single sample, sample mean, population mean)
– Standard deviation
– CV
– Standard Error

14
What else ?
• Performance of analytical method
– LOD, LOQ
– Sensitivity
– Specificity

• Type of error
– Random
– Systemic
– Blunder

15
Sampling and sample preparation
• Two categories of sampling based on results
− Attribute
− Variable

• Three basic types of sampling plans


− Single
− Double
− Multiple

16
Statistical consideration in sampling
• Non-probability sampling
− Judgment
− Convenience
− restricted
−Quota
• Probability or random sampling
− Single random
− Systemic
− Stratified
− Cluster
− Composite

17
Decide the sample size
• Sampling for attributes
− ANSI method
− CFR

• Sampling for variables


− Preliminary experiment
− Set desired confidence level and accuracy
− Calculation

18
The ultimate solution

Follow the AOAC procedures for different foods


And the CFR

19
Will continue on Chapter 13

20

You might also like