Rejection rate statistical measurement

S

siouxan

Here is my question, what make sense in measuring total QA for statistical measurement purposes:

For example, if a certain part (SMD) has 5000 pcs order, the production sends 1000 pcs with an outgoing record (sheet) every day to tape & reel packing room. QA operators in T&R do a visual inspection every day for a total of 5 times Monday through Friday (5 X 1000=5000 pcs, 5 record sheets, 5 lots).

Which formula / method is correct?
(1) If one lot (1000 pcs on certain day) is rejected, then the reject rate is 1/1=100% for all 5000 pieces?
(2) Or If one lot (1000 pcs on a certain day) is rejected, the reject rate is 1/5=20%, not 100%?
 

Bev D

Heretical Statistician
Leader
Super Moderator
It depends on how you subgroup the data. if you are 'plotting' or tracking each day AND you are considering each delivery event of 1000 pcs as a "lot" AND you are only inspecting one 1000pc lot each day, then the failure rate is either 100% or 0%. This situation - as you have described it - doesn't really constitute a "rational subgroup"* It might be better to plot and track by week. OR even by defect rate instead of by lot defective rate: record, track and plot the proportion of pcs that are found to be non conforming in the lot. The denominator would be the sample size adn it woudl be best if the inspectors always sampled the same amount even if they are rejecting the lot on the first nonconforming piece...in this case you could use a c chart or u chart to track the failure rate.


*rational subgroup is a statistical term not a comment on anyone's mental state. :)
 
S

sixsigmais

You need to define the reject is through lot basis or time basis.

You shall listed down the lot and the accumualte lot.
Day,total lot, lot accept, lot reject, reject rate, accumulate reject rate
Then everything is clear.

Eg. the day 1 lot reject, day2~5 lot all accept
Day total lot lot accept lot reject reject rate accumulate reject rate
1 1 0 1 100% 100%
2 1 1 0 0% 50%
3 1 1 0 0% 33%
4 1 1 0 0% 25%
5 1 1 0 0% 20%
 
Top Bottom