[ | Next | Previous | Up ]

Re: Developing double sampling plans

From: Stan Hilliard
Date: 15 Jul 2000
Time: 00:59:27

Comments

Hi Angela, you said:

> We are interested in developing

> double sampling plans. Will your

> software help us in this development?

Yes. Both the attributes program (TP105) and the variables program (TP414) are very flexible for designing double sampling plans. You can also design triple, quadruple, ... from sequential plans.

The method for double sampling plans is this:

Generate a sequential sampling plan having the oc-curve that you want, and then split the sequential decision table (or graph) in half to form two sample-size groups. The groups may be of equal sample size (n1=n2) but they don't have to be.

DOUBLE SAMPLING PLAN -- ATTRIBUTE EXAMPLE: The attribute sequential decision rule at the following address (case sensitive):

www.samplingplans.com/modern3.htm#SEQATT

has the sample size on the horizontal axis. By forming the sample into two groups, you could use this double plan:

n1=31, cumulative n=31, Ac=0, Re=3

n2=22, cumulative n=53, Ac=2, Re=3

The table below the diagram offers a more precise indication of where to separate the groups.

The single sampling plan with matching oc curve is:

n=43, Ac=1

DOUBLE SAMPLING PLAN-- VARIABLES EXAMPLE: The variables sequential decision rule at the following address (case sensitive):

www.samplingplans.com/modern3.htm#SEQTSS

has the sample size on the horizontal axis. By forming the sample into two groups, you could use this double plan Ac & Re are sample averages:

n1=5, cumulative n=5, Ac=2.79, Re=7.09

n2=5, cumulative n=10, Ac=4.94, Re=4.94

The table below the diagram offers a more precise indication of where to separate the groups.

The single sampling plan with matching oc curve is:

n=10, Ac=4.94

BY THE WAY: The two examples above have the same oc curve and thus offer the same protection to the producer and consumer -- even though the sample sizes for the variables double plan are about 1/4 that of the attribute plan.

Forming groups as was done in these examples does not substantially change the oc-curve, and is good sampling practice. Making a sequential into a double causes some increase in the average sample number (ASN) but this is minor.

This web contains much more information about sequential sampling: go to www.samplingplans.com, click "word search" and search for the word "sequential" - without the quotes.)

Sincerely, Stan Hilliard


Last changed: November 20, 2007