Food Protection Trends - July/August 2017 - 275

RESULTS
Statistically valid sample sizes
Three distinct statistically valid sampling plans were
developed with the same confidence levels and reliability.
1. Sampling by ATTRIBUTE:
b. c = 0 (zero tolerance sampling, no non-conforming
core temperatures allowed).
c. Does not assume a normal distribution.
d. Table 3 can be used by operators to determine the
size of the population (n) of pieces in the precooker.
Although Table 3 shows a range of 16 to 8,400 pieces
per precooker, the most common precooker batch
sizes are between 2,000 and 3,000 pieces (5).
e. Requires counting the number of samples.
f. Requires determining the number of samples based on
different population sizes per precooker batch size:
i. For 2400 pieces per precooker, the sample sizes
are listed in Table 4.
ii. For 4800 pieces per precooker, the sample sizes
are listed in Table 5.
iii. For 8400 pieces per precooker, the sample sizes
are listed in Table 6.
iv. For an infinite number of pieces, the sample sizes
are listed in Table 7.
2. Sampling by VARIABLES, using the median
and
range and d2 from the sample data: sample sizes listed
in Table 8.
c. Verify normal or near-normal distribution of the core
temperature data collected.

d. Use sample median
and range/d2 to accept or
reject the lot:
i. Lower limit is y = [ - (3*range/d2)] and must
be at least 60°C to pass. If n = 60 and the lot
passes, we can be 95% confident at least 99% of
the lot is good. If n = 10 and the lot passes, we can
be 95% confident at least 95% of the lot is good.
c. Requires counting, simple addition and division, and
table look-ups.
d. Very easy to use in that it involves only counting and
simple arithmetic.
3. Sampling by VARIABLES using mean (µ) and standard
deviation (σ) estimated from the sample data: sample
sizes listed in Table 9.
d. Verify normal or near-normal distribution of the core
temperature data collected.
e. Uses the sample mean
and sample standard
deviation (s) to accept or reject the lot:
i. Lower limit is y = [ - (3* s)] and must be at
least 60°C to pass. If n = 35 and the lot passes, we
can be 95% confident that at least 99% of the lot is
good. If n = 10 and the lot passes, we can be 95%
confident that at least 95% of the lot is good.
c. Requires multiple calculations and thus is more
difficult to do in real time.
d. Uses the smallest sample size.

TABLE 3. Lot sizes for a precooker load by fish size and precooker size
Precooker lot size - count of individual pieces/fish
Fish per basket and average size
Fish per basket
Size in kg
>>
Racks/
Cooker
1
5
10
15
20
25
30
35

Basket/
Rack
16
16
16
16
16
16
16
16

1

2

3

4

6

8

10

12

15

10 kg

8 kg

6 kg

4.2 kg

3 kg

2.25 kg

1.6 kg

1.2 kg

< 1 kg

16
80
160
240
320
400
480
560

32
160
320
480
640
800
960
1,120

48
240
480
720
960
1,200
1,440
1,680

64
320
640
960
1,280
1,600
1,920
2,240

96
480
960
1,440
1,920
2,400
2,880
3,360

128
640
1,280
1,920
2,560
3,200
3,840
4,480

160
800
1,600
2,400
3,200
4,000
4,800
5,600

192
960
1,920
2,880
3,840
4,800
5,760
6,720

240
1,200
2,400
3,600
4,800
6,000
7,200
8,400

July/August Food Protection Trends

275
Table of Contents for the Digital Edition of Food Protection Trends - July/August 2017
