Sample size for checking that records, known to be in error, are now corrected.

#1
The Problem

I have audited 1000 records and found 100 records to be non-compliant. I have asked for the 100 non-compliant records to be corrected. The individual responsible has responded and said all 100 of the records have now been corrected.

What sample size of records do I need to check from a population that I know was 100% non-compliant to be confident that 95% of the records have indeed now been corrected.

I need to calculate similar sample sizes for differing numbers of non-compliant records.

The Solution

I am not statistically minded but can drive Excel to a reasonable degree or indeed an online calculator of some sort.

Could somebody help with what I am sure is a straigthforward question for the people on this forum. Your challenge is to be able to explain it to a layman!!

Thanks in advance.

Nick
 

terzi

TS Contributor
#2
Hi Nick,

After reading your problem I think you would need to read about acceptance sampling techniques. These methods have been devised to periodically examine products to decide with certain statistical significance whether you should accept every product after examining a few of them. First, you have to define what will be considered acceptance and how you are going to measure it: are you counting defects? Are you just saying yes or no? Although I'm not an expert in Quality Control, here's a link that should be useful to understand these concepts and also some sample size calculators

http://www.sqconline.com

I hope this helps you a bit.