# USL calculation Problem

Six Sigma – iSixSigma Forums Old Forums General USL calculation Problem

Viewing 8 posts - 1 through 8 (of 8 total)
• Author
Posts
• #48701

Aguilar
Member

Thanks a lot to Pete and stan for your opinion in this case.
I need a help with this please let me know.
In the next table your can see that customer in the department B in a range 16 – 30 consider their waiting time good.  This is what they want.  This means that I can consider less that 31 min my USL ?
I will explain what a professor told me about the calculation of the media.  I am not sure about this because I did it in a different way.  The professor got two X.. .
X = (46 (0 ) + 47 (16) +  11(31) + 1 (61)) / 105 ( for the LCL)
the other was
X = (46(15) + 47(30) + 11 ( 60) + 1(90)) / 105 ( for the UCL)
Now he told me that I need to calculate the standard deviation of the distribution of the patients.   I calculate it ( thanks to Pete for your comments about this ) and X = 105/4 = 26.25 and standard deviation = 20.56.  The one million question is how can I calculate the USL.

Time

A

B

C

D

0-15

69

46

65

76

16-30

36

47

44

44

31-60

13

11

11

6

61-90

3

1

0

0

91-120

2

0

0

0

121-150

3

0

0

0Victor

0
#164991

Cordell Fife
Participant

Hi Victor
I think decission on the spec limits of a process is a matter of design and what the critical limits for the customer are. So you did well to ask customers about their tolerable limits. If you want to derive spec limits from those data, I think you should choose 15min. (Because less is always better in this case, as mentioned in previous replies) When you select the 30th min as USL, you will have 50% unhappy customers in departmen B.
I am really confused what you are trying to do with the sigmas. I think sigma has nothing to do with spec limits. But if you want to see the capability or performance of your process, then you need to use sigma and already defined spec limits. But do not forget your data are durations which are not normally distributed.
Good luck
Kim

0
#165007

Ward
Participant

I agree going to the customer is what you would like to do, but asking them is not always effective. Customers really don’t have a good sense of time. They may “say” 15 minutes is tolerable, but they just pulled that number out of a hat, so to speak. A better way might be to question them after service has been provided. You will have to capture their wait time, and compare that to the level of satisfaction for the wait time. You should obviously see greater satisfaction for the the wait time when it is short and the decline when longer, and you can then determine what an appropriate wait time might be.

0
#165011

Dr. Scott
Participant

Victor,
First, in your case the LSL should probably always be zero. In healthcare, patients want the service ASAP, which is zero. In other industries, such as work I have done with restaurants, the LSL isnt always zero. That is, at “higher end” restaurants, the customer often appreciates a wait time to gather with friends and have a drink at the bar. At a fast food restaurant people want the order filled ASAP.
Second, another observation I have is that you seem to have four groups of customers (A, B, C, and D). And each group has more progressively higher standards. I would suggest setting a different USL for each group, in the same manner that a chemical company, or restaurant, or a tire store does. If you can identify the segment (group) you are servicing, then you know what their USL is. Then you must improve the processes offered to satisfy each.
Third, look at Petes advice on how to truly identify what the USL is for each of the four groups. His idea is a good one, one which is preferable over merely asking the customer before the experience.
And finally, if given only the information you present in your table, I would very subjectively set the USL at 30 minutes to hopefully satisfy up to 90% of the patients.
Hope this is of some value. As far as your professor goes, s/he apparently has their own way (formula) for determining the USL. All I can say for certain is that the LSL is zero. And don’t overlook what Pete said.
Regards,
Dr. Scott

0
#165039

Chris Seider
Participant

I would suggest having only a one sided spec for the process as the USL.

0
#165070

Dr. Scott
Participant

C Seider,
That is what I meant by saying the LSL should be zero (a hard or fixed limit) and setting the USL.
Regards,
Dr. Scott

0
#165073

Ward
Participant

If you are using Minitab, remember to set “0” as a boundary, so it does not calculate the likelihood of performing less than “0”.

0
#165078

Dr. Scott
Participant

Pete,
You are absolutely correct, and I should have mentioned that early.
Thanks,
Dr. Scott

0
Viewing 8 posts - 1 through 8 (of 8 total)

The forum ‘General’ is closed to new topics and replies.