QA sampling chart

Back | Back to QA page

QA Sampling Chart

print this page
This chart is taken more or less from the Internet Archive Procedures document [here].
Sampling chart below is a simplified version of the ANSI tables found here: ANSIsamplingTables.pdf
For our purposes, the bin can be defined as whatever number of items are on a given cart. If not using carts, a bin can be the total shipment received back from IA.

Items in set (‘bin’)	9-15	16-25	26-50	51-90	91-150	151-280	281-500
number of items to QA	3	5	8	13	20	32	50
# of errors to Fail the set	1	1	1	2	3	4	5
of major errors that are acceptable	0	0	0	1	1	2	3
minor errors for fail	1	1	2	2	3	3	4
minor errors that are acceptable	0	0	1	1	2	2	3

Unfortunately, I have no instructions as to how to select the sample set.

An item fails QA if one error is found. Multiple errors in one item, however, only count as one failed item. We are being generous.
What is an error?
IA "Freeze Errors" include:

• pdf, DjVu, or flip book are missing or corrupt
• truncated files
• missing files (meta.xml, marc.xml etc etc)
• Cropped text
• Blurred page(s)
• Missing page(s), goal is zero pages missing per book. IA will note if pages are missing in the books to be scanned.
• Front cover missing
• Back cover missing
• Text is washed out or overly dark (bad light-comp)
• Evidence of scanner (fingers/shadows/etc) visible on page
• Glass not centered in gutter; text is distorted or cropped

Metadata minor errors (title truncation, volume information incorrect, etc.) are NOT counted as a 'Freeze error' that will fail a book. For now, we are not correcting these errors in IA, but are waiting until the item is ingested into the portal and correcting the metadata there.
Metadata major errors (scan is attached to incorrect MARC record) is a Freeze error.