SAMPLING INSPECTION TABLES 17 



the process average value thus determined is well within the range of process 

 average values listed in the selected sampling table then sampling can ad- 

 vantageously be introduced. If it is beyond this range, it would be quite 

 satisfactory from a protection standpoint to use the last process average 

 column of the selected table but the samj)ling plan itself would force rejec- 

 tion or a screening inspection of such a large proportion of the lots that 

 the introduction of sampling probably would not pay. If the process aver- 

 age value is but poorly estimated, the amount of inspection will be some- 

 what larger than need be but the specified degree of protection will still be 

 realized. Where there is uncertainty it is better to overestimate than to 

 underestimate the process average value since, for a given magnitude of 

 error, a lesser amount of excess inspection will thereby be incurred. 



It should be especially noted that the tables may be safely applied whether 

 quality is well controlled or not. If, for example, the usual level of per cent 

 defective is well within the range of process average values listed in the 

 selected table but individual lots are frequently well outside this range, 

 the sampling plan will usually permit acceptance by sampling while quality 

 is good but force 100% inspection when it is bad. 



Experience with the tables indicates that where the procedures are used 

 by a manufacturer within his own organization or by a consumer who 

 rejects lots that are not accepted by sample, the general plan forces correc- 

 tive action whenever quality becomes poorer than normally expected. 

 The attendant increase in overall inspection costs provides a compelling 

 argument, in a language well understood by all, for determining the cause 

 of trouble in the manufacturing process and for instituting measures for 

 eliminating it as speedily as possible. Thus, while the inspection pro- 

 cedures have as their immediate purpose the provision of a curative tech- 

 nique whereby product already made is cleared of abnormal proportions of 

 defects, they are found by experience to enforce the adoption of a preven- 

 tive technique one that exerts economic pressure to track down and 

 remove causes of abnormal quality variations, thus enforcing control of 

 quality in the process and assuring better health in the product of tomor- 

 row. Because of these factors the long term average outgoing per cent 

 defective may rarely be expected to exceed half the AOQL value associated 

 with the inspection plan in use. 



Quality control is achieved most efficiently, of course, not by the inspec- 

 tion operation itself but by getting at causes.^ It may be expedited by 

 carrying out regular statistical control analyses of the cumulative results 

 of sampling inspection — preparing quality control charts ' for "per cent 

 defective" with subgrouping of results on a lot-by-lot, a day-by-day, or a 

 week-by-week basis — and making the findings available to those directly 

 responsible for manufacturing processes. 



