看看MSA手册第99页的注46:
The total number of “ranges” generated ought to be > 15 for a minimal level of confidence in the results. Although the form was designed with a maximum of 10 parts, this approach is not limited by that number. As with any statistical technique, the larger the sample size, the less sampling variation and less resultant risk will be present.
“为了获得结果中最低的置信水平,总的极差数量应该大于15。尽管表格被设计为最多容纳10个零件,但这种方法并不会因此限制零件数量。如同任何统计技术,较大的样本容量,会降低抽样变差和结果的风险”
所以还是尽量取样本数量超过15。
最新回复
parrot_cqj (2008-7-29 14:22:12)
20个零件,3个人,重复3次。
georgecho (2008-7-29 20:57:41)
The total number of “ranges” generated ought to be > 15 for a minimal level of confidence in the results. Although the form was designed with a maximum of 10 parts, this approach is not limited by that number. As with any statistical technique, the larger the sample size, the less sampling variation and less resultant risk will be present.
“为了获得结果中最低的置信水平,总的极差数量应该大于15。尽管表格被设计为最多容纳10个零件,但这种方法并不会因此限制零件数量。如同任何统计技术,较大的样本容量,会降低抽样变差和结果的风险”
所以还是尽量取样本数量超过15。
wsbg (2008-7-30 11:28:44)
那么,能否计算出他的置信水平?
另外,对于计数性的MSA,其样本量又当如何?请指教!
logingx0304 (2008-7-31 16:23:07)