Hello,
This relates to Apparels industry. We had an inspector appointed by the customer to do a Pre Shipment Inspection.
They used ANSI/ASQ Z1.4-2008 sampling level II for what they call 'visual' inspection which was essentially attributes inspection. Major 2.5 and Minor 4.0. Based on the lot size of 2000 pcs they inspected 125 pcs as the random sample size. So far so good.
They then used special inspection level S4 for measurements of the garments and selected 32 garments with Accept as 2 and reject 3.
I was not convinced that this was right as would have thought that to determine conformance, they should have used variable sampling plan Z1.9 based on MIL-STD-414 and should do the standard deviation calculation to determine percent conformance/ non-conformance.
They chose 7 measure points. 1. Chest width 2. Body length 3, Shoulder width 4. Bottom opening (sweep) 5. Hood height 6. Sleeve length and 7. cuff opening.
They classified a garment as defective if any one of the measurement was out of tolerance on any one garment. For e.g. body length is specified to be 27 inches. The process tolerance given is +/- ½ inch. Thus if any one garment measured + or - ¾ inch it was counted as one defective piece. They ended up counting 22 out of 32 in this way and failed it.
The overall report was PASS for Visual FAIL for Measurement.
My view was that their measurement inspection was not correct. If they were counting out of tolerance points then statistically they are asking a score of 2 out of 32x7 = 224 points and not 2/32. If they have to convert their variable into attribute then the garment should count as a defect if it falls under 'Major' classification, and a length of +¼ inch does not make a garment fall in that definition and should not be counted. Hood is cinched for functionality and its measure of +½ inch or ¼ inch over tolerance is not 'Major'
Below is a part of the report that shows the sizes, tolerances and what they measure. The yellow highlights are out of tolerance and is considered major defect. This is a pullover, not a dress shirt.
In the table below they inspected 3 pcs size XS, 4 pcs size S, 4 pcs Size M, 4 pcs size L, 4 pcs size XL and 4 pcs size XXL. Total 23 garments and their count was major = 20. I Have not included the table for the rest of 9 pcs but it looks similar.
So my questions are:
1. Is using Z1.4 attributes sampling valid for variable inspection as above?
2. If answer to Q1 is Yes, then Is their classification of 'Major' count correct?
This relates to Apparels industry. We had an inspector appointed by the customer to do a Pre Shipment Inspection.
They used ANSI/ASQ Z1.4-2008 sampling level II for what they call 'visual' inspection which was essentially attributes inspection. Major 2.5 and Minor 4.0. Based on the lot size of 2000 pcs they inspected 125 pcs as the random sample size. So far so good.
They then used special inspection level S4 for measurements of the garments and selected 32 garments with Accept as 2 and reject 3.
I was not convinced that this was right as would have thought that to determine conformance, they should have used variable sampling plan Z1.9 based on MIL-STD-414 and should do the standard deviation calculation to determine percent conformance/ non-conformance.
They chose 7 measure points. 1. Chest width 2. Body length 3, Shoulder width 4. Bottom opening (sweep) 5. Hood height 6. Sleeve length and 7. cuff opening.
They classified a garment as defective if any one of the measurement was out of tolerance on any one garment. For e.g. body length is specified to be 27 inches. The process tolerance given is +/- ½ inch. Thus if any one garment measured + or - ¾ inch it was counted as one defective piece. They ended up counting 22 out of 32 in this way and failed it.
The overall report was PASS for Visual FAIL for Measurement.
My view was that their measurement inspection was not correct. If they were counting out of tolerance points then statistically they are asking a score of 2 out of 32x7 = 224 points and not 2/32. If they have to convert their variable into attribute then the garment should count as a defect if it falls under 'Major' classification, and a length of +¼ inch does not make a garment fall in that definition and should not be counted. Hood is cinched for functionality and its measure of +½ inch or ¼ inch over tolerance is not 'Major'
Below is a part of the report that shows the sizes, tolerances and what they measure. The yellow highlights are out of tolerance and is considered major defect. This is a pullover, not a dress shirt.
In the table below they inspected 3 pcs size XS, 4 pcs size S, 4 pcs Size M, 4 pcs size L, 4 pcs size XL and 4 pcs size XXL. Total 23 garments and their count was major = 20. I Have not included the table for the rest of 9 pcs but it looks similar.
So my questions are:
1. Is using Z1.4 attributes sampling valid for variable inspection as above?
2. If answer to Q1 is Yes, then Is their classification of 'Major' count correct?
