Disadvantages of DPMO into sigma level
Six Sigma – iSixSigma › Forums › Old Forums › General › Disadvantages of DPMO into sigma level
 This topic has 9 replies, 8 voices, and was last updated 13 years, 8 months ago by J. Knauf.

AuthorPosts

March 27, 2009 at 3:17 am #52115
Hi all,
Understanding process in terms of Sigma level by converting them from DPMO is not always right. I wanted understand What are the disadvantages in converting DPMO into Sigma level. Will that result in a wrong intrepretetion of a process performance. Kindly share your thoughts.
0March 27, 2009 at 11:28 am #182826DPMO and sigma level are the same. If you know one, you know the
other.0March 30, 2009 at 5:27 am #182900
shaveta dattaMember@shavetadatta Include @shavetadatta in your post and this person will
be notified via email.agree on same…
both are same things, if we are 6 sigma level organization , it means we have 3.4 DPMO… so this link is established standard from past research histories..
i hope that it clarifies
0March 30, 2009 at 10:27 am #182909
So Not StanMember@SoNotStan Include @SoNotStan in your post and this person will
be notified via email.Assuming you believe in the 1.5 sigma shift and drift concept. Otherwise, 6 sigma = .002 DPMO.
0March 30, 2009 at 12:41 pm #182916Hi ,
Agred that the process of convertion and calculating DMPO is sigma level but i still think wrong intrepretetion of a process performance can be derived .0April 6, 2009 at 12:48 pm #183151
John KnaufParticipant@JohnKnauf Include @JohnKnauf in your post and this person will
be notified via email.Ram,
If I understand your question correctly, you want to know how you go back and forth between DPMO and the sigma level. As others have pointed out, they are two ways of expressing the same metric; however, actually calculating one from the other can lead you astray. Here’s why:
If a process actually ran consistently at a six sigma level, the probability of a defect falling outside the six sigma limits is one chance per BILLION per side, not 3.4 chances per million per side. That’s a total of two chances per BILLION overall. Yes, that’s billion, not million. The reason we use the 3.4 chances per million is because of the 1.5 sigma shift. If a process running at six sigma shifts one and half sigmas, it is now a 4.5 sigma process. The chances of a defect falling outside the 4.5 sigma limits is 6.8 chances per million, or 3.4 per side, or what we call a six sigma process.
Thus, if you calculate the sigma level from DPMO you need to keep in mind that the resulting sigma is presumed to represent the 1.5 sigma shift. To reiterate the example above, if you actually had a process producing 3.4 part per million per side the calculation returns a sigma level of 4.5, which we call six sigma by adding 1.5 to it.
Conversely, if you want to calculate the DPMO from the sigma value, you need to subtract 1.5 sigmas from that value before you do that. If you don’t, you’ll discover that your calculation does not agree with what the six sigma defect tables tell you it should be. An actual 4.5 sigma process, for example, needs to be calculated at +/ 3 sigma to get the DPMO.
I hope this helps to clarify it for you.
J. Knauf
0April 6, 2009 at 1:41 pm #183153John,Ever had a process so good that you needed to worry about the 3.4
per million vs the 2 per billion?0April 6, 2009 at 4:18 pm #183158
John KnaufParticipant@JohnKnauf Include @JohnKnauf in your post and this person will
be notified via email.Gary,
I’m not sure what you’re asking here. I was trying to explain why, if you calculate sigma levels versus defect rates, you’ll end up with values that don’t agree with what the DPMO tables say they should be. This process would be the same for any sigma level, not just the six sigma level. If you had a three sigma process it would be calculated at 1.5 sigma, in which case you really would need to worry about what’s coming out of it.0April 6, 2009 at 4:48 pm #183163John,I just thought it was an interesting choice of examples.Also, your example is incorrect, the 1.5 shift gets applied to the z (or
k if using Juran’s Handbook) value from a normal table.The term “sigma value” is not a recognized statistical term rather a
creation of the Six Sigma community. All “sigma value” tables I have
seen have the 1.5 baked in.0April 6, 2009 at 5:51 pm #183168
J. KnaufParticipant@J.Knauf Include @J.Knauf in your post and this person will
be notified via email.Gary,
You are correct; the 1.5 refers to the normalized (Z) value. Since I was addressing Ram’s question about DPMO versus capabililty, my operating assumption was that he would be calculating capability using normalized (Z) values, which is what I typically use to calculate DPMO as well. For the sake of the example I was using Xbar=0 and sigma=1.
Bear in mind that all the normalizing process is doing is expressing the delta in sigma units.
The Six Sigma tables that I’ve seen also have the 1.5 shift built in; my sense was that Ram might’ve been confused over why a density function for a certain sigma range (2, 3, 6, or whatever) would show a value that did not agree with what’s on the table. My point was that if were to shift the center or target by 1.5 sigmas his density calculation would then agree with the table. By doing this you also don’t need to carry tables around with you.
I’m not sure what you mean by “sigma value” not being a recognized term. Are you equating it with the 1.5 sigma shift in the Six Sigma tables?0 
AuthorPosts
The forum ‘General’ is closed to new topics and replies.