From: Stan Hilliard
Date: 23 Sep 1999
Time: 01:10:15
Greetings Dan,
I think TP105 is the choice, but to be sure, consider the pros & cons of TP105 versus Audit Sample Planner (ASP).
TP105 will allow you to have a pre-established decision rule based on a probability statement. It can be either a fixed-n or a sequential decision rule.
ASP doesn't develop decision rules. One thing that ASP has that TP105 doesn't have is the ability to print a data sheet containing random record numbers to use in selecting the sample. It also will write the record numbers to a file in *.CSV format.
The confidence statement is slightly different than you state.
> "Based on the confirmed accuracy of x-number of > records I am 95% confident that > all records are accurate."
It should be:
"Based on the confirmed accuracy of x-number of records I am 95% confident that at least X% of the records are accurate."
The X% could be stated as the actual number by multiplying it by the number of records in the database.
You calculate X% with the RQL that you use to develop the plan using TP105:
X%=100*(1-RQL)
Your sampling plan design job will be to strike a satisfactory balance between the conflicting needs of high X% (low RQL fraction defective) and low sample size.
Sincerely, Stan Hilliard