# six sigma

Six Sigma – iSixSigma › Forums › Old Forums › General › six sigma

- This topic has 40 replies, 14 voices, and was last updated 12 years, 3 months ago by sumant.

- AuthorPosts
- October 6, 2007 at 5:49 am #48341
Sigma being standard deviation ,the less it is , better the control .If it’s correct, why six sigma is better than one sigma .Please answer not with confusion but with simple mathematics

0October 6, 2007 at 6:19 am #162687

crazy guyParticipant@crazy-guy**Include @crazy-guy in your post and this person will**

be notified via email.Dont compare six sigma and one sigma in silo. Six sigma is better than one sigma is because you are able to fit in six std deviations within the specification limits, if you have only 1 sigma within the specification limits then 32% of the output would be defects.

0October 6, 2007 at 8:51 am #162690As I told , don’t please try to confuse ,unless you know the subject .With due regards to you ,question remains unanswered

0October 6, 2007 at 10:13 am #162693

fake accrington alertParticipant@fake-accrington-alert**Include @fake-accrington-alert in your post and this person will**

be notified via email.Good Question

0October 6, 2007 at 10:13 am #162694

fake accrington alertParticipant@fake-accrington-alert**Include @fake-accrington-alert in your post and this person will**

be notified via email.good answer

0October 6, 2007 at 10:32 am #162697whose answer is good ?

0October 6, 2007 at 10:45 am #162698

fake accrington alertParticipant@fake-accrington-alert**Include @fake-accrington-alert in your post and this person will**

be notified via email.Crazy Guy

0October 6, 2007 at 11:57 am #162702But what’s the answer ?

0October 6, 2007 at 1:02 pm #162706

fake accrington alertParticipant**Include @fake-accrington-alert in your post and this person will**

be notified via email.Because SS is equal to 3.4 ppm and

One Sigma is equal to 697,672 ppm.Is it clear now?

For example the average American company is at four sigma level ,which is equivalent of 0.6 % defective or 6,210 defects per million opportunities.

Sigma is a statistical term that refers to the standard deviation of a process about it’s mean.In a normally distributed process,99.73% of measurements will fall within +/- 3.0 sigma and 99.00066% will fall within +/- 4.5 sigma.In a stale attribute distributed process,99.73% of defective units,or defects,will fall within probability of 0.00135 and 0.99865.

Now if you don’t understand all that,forget the SS and sleep well.

best regards0October 6, 2007 at 4:27 pm #162708

BrandonParticipant@Brandon**Include @Brandon in your post and this person will**

be notified via email.GKP – you need to do a little more studying on your own before you ask us to give a simple answer to a complex issue.

Do your own research and if something you find in that effort is still confusing come back here and pose a question.0October 8, 2007 at 4:47 am #162744It happens to all the big bosses .When they are not knowing the answers ,they will look at his subordinate with either astonishing or humorous look ,as if ,he is telling ” Fellow ,you don’t know the answer of this simple question !!!!!!!!” And tells like you

“Now if you don’t understand all that,forget the SS and sleep well.”

Sir ,you are not talking to a subordinate and your answer is read by the forum .I would expect a little bit of decorum ,when you answer a question .No , sir , you are not even close to answer. At least ,it’s my views .The question remains open.

Regards0October 8, 2007 at 6:33 am #162746

crazy guyParticipant@crazy-guy**Include @crazy-guy in your post and this person will**

be notified via email.GKP – Brandon and FAA are big bosses indeed- They have proved their mettle in their own world to their customers and they are here to help without taking money. If they spend the time they spend here working for customers they would be millionaries- so what they do is charity..

What brandon and FAA are trying to say is please go through your statistics books first before posting your question because the question could look simple but explanation is something which cant be said in a word or two. Since this deals with probability theories its important to understand them before understanding six sigma and one sigma.

Say for example you have a mean of 50 and LSL of 45 and USL of 60. If your process is at 1 Sigma it means only 68% of the output are within 45 and 60 and rest are outside these limits(how many ever units/oppurtunities you have) and if your process is at 6 sigma it means the 99.99996% of the outputs are within 45 and 60(how many ever units/oppurtunities you have) which means deviations perse should be less such that you are abl to fit in 6 standard deviations within the spec limits.

Let me try to be more clearer. For a given set of data you find that:

Mean – 50

St dev- 2

If you are able to fit in 6*2 ( which means 99.99996% of the data)falls within your spec limits then your process is at 6 sigma but if you fit only 1*2 then your process is at 1 sigma.

If if still unclear let me back off…let me not confuse instead of creating clarity…

Sometimes i wonder whether i need to get back to my basics…

Regards,

crazy guy0October 8, 2007 at 6:50 am #162747Dear Sir ,

Thanks for your kind response .I would request you not to back off because myself is also going to basics to answer this unique but simple question ,received from my peers . I am sure to get the concept clarified by the esteemed persons of this forum but honestly ,the answer is yet to hit the bull’s eye .Till I’m myself convinced ,I can’t convince others ,with my bullshit jargons.

Warm regards0October 8, 2007 at 6:59 am #162748

fake accrington alertParticipant**Include @fake-accrington-alert in your post and this person will**

be notified via email.Crazy Guy

Very well said0October 8, 2007 at 8:55 am #162750

crazy guy aka six sigma guyParticipant@crazy-guy-aka-six-sigma-guy**Include @crazy-guy-aka-six-sigma-guy in your post and this person will**

be notified via email.thanks – FAA..

wondering how else to make it more clearer. Am i missing here something?maybe nobody asked this question to me ever :)

In practice, one often assumes that the data are from an approximately normally distributed population. This is frequently justified by the classical central limit theorem, which says that sums of many independent, identically-distributed random variables tend towards the normal distribution as a limit. If that assumption is justified, then about 68 % of the values are within 1 standard deviation of the mean to the spec limits and if your process is so then your process is at 1 sigma and so on and so forth.This is called as empirical rule and I dont question rules( maybe i need to). Maybe we can just leave it at that…

Regards,

Crazy guy aka six sigma guy0October 8, 2007 at 9:04 am #162751

Six Sigma guyMember@Six-Sigma-guy**Include @Six-Sigma-guy in your post and this person will**

be notified via email.There are two types of Six Sigma followers/experts or whatever u call.. One who take all decisions based on statistics and who know in depth of statistics and secondly those who dont rely much on statistics but use them to validate their assumptions and dont really get into the calculations( which anyways are being done by softwares). I think i belong to the second category of flock :)

I know how to use softwares,how to make interpretations,how to validate my assumptions and more importantly have business sense.So i dont care much about the formulas of the world and derivations of the world which i leave it to the statisticians and I am not one :) – Should i feel bad about this?0October 8, 2007 at 10:17 am #162752

VidyadharMember@Vidyadhar**Include @Vidyadhar in your post and this person will**

be notified via email.GKP…

I can undersatnd ur dilemma… Lets look at a simple example…

U r manufacturing globes of diameter 100m.. and your customer will only accept only if they are between 98mm to 102 mm..

98 & 102 are ur specification limits.

now after you have manufactured a million globes… they might have some variation in diameters.. say ur average dia is 100mm and ur std deviation is at 0.33mm, you can fit 6 std deviations from 100 to 102 and 6 std deviations from 98 to 100. so u will be actually operating on 6 sigma

on the other hand if ur average dia remains the same and ur std deviation is 1 mm.. U can fit only 2 std deviations between the spec limits to the central value(average).. so u will be operating at 2 sigma level…

I hope this answers ur question…

0October 8, 2007 at 11:43 am #162753Now I am really feeling embarassed to tell that concept of 6 sigma ,is yet see the bright light of the day .But I am sure someone will really clear the concept, in near future, otherwise I will withdraw my question. Anyway ,thanks for your explanation.

Regards0October 8, 2007 at 12:03 pm #162754

VidyadharMember@Vidyadhar**Include @Vidyadhar in your post and this person will**

be notified via email.GKP….

If you have any doubts… please post the same…. If you could just rephrase ur question instead using idioms and phrases… I think we could answer ur question….0October 8, 2007 at 12:13 pm #162755May be we are missing the basics .I answered all the inputs, from all of you ,I received , but my peers are giving me a blank look .I also know that I’m not really convinced myself ,with these answers .I keep the question open.

Regards0October 8, 2007 at 12:29 pm #162756It’s the effectiveness that counts. If you are excelling in your deliverables ,who bothers , it’s in which sigma ,you are operating .As Ford tells ,we are not here to make cars ,we are here to make money. And this is the ultimate today with fierce competition ,worldover.

Anyway , I will be waiting for clearing the concept on six sigma ,as I can’t satisfy my peers who feels I am confusing them with all these answers received from my friends.

Regards0October 8, 2007 at 12:40 pm #162757If it’s linked to my message , I am wondering which idiom or phrases ,I have used ! If not linked , I apolizise .My question is open and it’s very simple question .If you can answer straight ,I will be oblized.

Regards0October 8, 2007 at 1:04 pm #162760Hi GKP,

your question – “Sigma being standard deviation ,the less it is , better the control .If it’s correct, why six sigma is better than one sigma”?

I have made two Sigma’s in 2 different fonts.

Sigma as standard deviation is not same as Sigma in SIX SIGMA (Here Sigma is known as Sigma Levels), where as sigma as SD known as sigma value.

Both sigma Value and sigma Level are inversely propotional. When Sigma value is high, sigma level is low and viceversa.

So when you are saying six sigma, it means your process is operating at six sigma level which is better than one sigma level.

When Sigma level is higher, process is better, but when sigma value is lower process performance is better.

Hope I clarified your question.0October 8, 2007 at 4:14 pm #162770

BrandonParticipant@Brandon**Include @Brandon in your post and this person will**

be notified via email.The use of SS to improve business processes cannot be explained solely when one grasps the difference between one sigma and 4 or 5 or 6 sigma.

GKP – you are either toying with us or nowhere near ready to discuss SS with your superiors. Buy some books and grasp the entire concept before posing any more questions.

Other posters – I think we need to ignore this guy for a while.0October 8, 2007 at 6:26 pm #162783Not sure ‘we’ are missing the basics, but maybe putting all the responses in one post can help – if this doesn’t do it, please read the new to six sigma link and search fo rthe statistical meaning of six sigma – it will help.

Sigma (std deviation) and Sigma Level are different, different different. Don’t confuse the 2. Sigma is the greek symbol for standard deviation. Sigma Level is a unitless number of standard deviations that can fit between a mean and the nearest spec limit — in the normal distribution theory it is the Z value.

Say you have a process with a mean value, spec limits and a standard deviation. Example: Mean: 100 Std Dev.: 10 Upper Specification Customer Limit: 110. Given this, you can fit 1 std deviation between the mean and spec (100 + 10 = 110) – this system operates at a 1 sigma level.

Now, say you improve the consistency of your process by reducing the standard deviation from 10 to 2.5. If the mean remains the same, you can now fit 4 (2.5 x 4 = 10) standard deviaitons between the mean (at 100) and the upper spec limit (at 110). The Sigma Level is now 4.

In the improvement example, the consistency got better by lowering the standard deviation. You, in turn, can now fit more standard deviations between the mean and 1 spec limit (increased sigma level).0October 9, 2007 at 4:34 am #162800I am not convinced . Probably , we are not going to the route of 6 sigma concept . I repeat my question ,if sigma is standard deviation and 6 sigma is 6 x sigma ,why 6 x sigma ,which is 6 times higher than sigma ,is called a world class control with respect to one sigma or half sigma ,which is much lower.

Regards0October 9, 2007 at 4:40 am #162801At last , I feel that we are coming close to the answer and my hats off to you .But still , I need a mathematical explanation to understand the inverse proportionality ,as you rightly told. Can you please extend your expertise little further .

Best regards0October 9, 2007 at 4:55 am #162802GKP,

The “six sigma” numbers are meaningless rubbish. This paper gives an explanation of how the pseudo-maths originated:

http://qualitydigest.com/IQedit/QDarticle_text.lasso?articleid=119050October 9, 2007 at 5:45 am #162803

WaskitaParticipant@ferry-waskita**Include @ferry-waskita in your post and this person will**

be notified via email.GKP

You are a trully funny person. No offence but you’re very funny indeed. I just read this topic from the first thread and laugh all the way to the end reading your Qs. All the colleagues in this forum have tried their best to give answer from a very simple way; yet you still don’t understand …not to mention if someone provides some statistical term/formula …you’ll be even more confused.

Again no offence, but it is you who is the barrier to understand the concept since you don’t have basic statistic knowledge at all. Do a little study first on statistics or if you still prefer a shortcut as you keep on insisting, then come to my home and i’ll explain it to in person with graphics + coffee + crackers …haha. I mean it though!0October 9, 2007 at 6:01 am #162805It’s one’s perception whether it’s rubbish or religion .Sumant’s message in this concept brings the actual dimension of 6 Sigma ,which is pasted below along with my acknowledgement.

Regards

TOPIC: Re: Six Sigma

A new message by Sumant was posted in the Discussion Forum.Hi GKP,

your question – “Sigma being standard deviation ,the less it is , better the control .If it’s correct, why six sigma is better than one sigma”?

I have made two Sigma’s in 2 different fonts.

Sigma as standard deviation is not same as Sigma in SIX SIGMA (Here Sigma is known as Sigma Levels), where as sigma as SD known as sigma value.

Both sigma Value and sigma Level are inversely propotional. When Sigma value is high, sigma level is low and viceversa.

So when you are saying six sigma, it means your process is operating at six sigma level which is better than one sigma level.

When Sigma level is higher, process is better, but when sigma value is lower process performance is better.

Hope I clarified your question.

From GKP

At last , I feel that we are coming close to the answer and my hats off to you .But still , I need a mathematical explanation to understand the inverse proportionality ,as you rightly told. Can you please extend your expertise little further .

Best regards0October 9, 2007 at 6:25 am #162806Is it ? With due regards , please go through Sumant’s answer and you will find that the comments given by you on me ,is not that funny ,as you stand now , after reading ,because except Sumant ,none has come even closer to the answer .I will advise you on your own words only

“Do a little study first on statistics or if you still prefer a shortcut as you keep on insisting, then come to my home and i’ll explain it to in person with graphics + coffee + crackers …haha. I mean it though!”

Regards0October 9, 2007 at 6:43 am #162807

WaskitaParticipant@ferry-waskita**Include @ferry-waskita in your post and this person will**

be notified via email.:)

As mentioned, no offence my friend…..My apology for the words chosen in my previous post.

However; you must be joking when saying Sumant’s comment is the only one closer to the answer. The very first reply from Crazy Guy was short yet sharp, precise, and right to the points as well as others who also tried to assist.

Anyway, glad to se that you personally think you have understood the answer (at least closer as you said) …..though i still doubt it ;)0October 9, 2007 at 7:11 am #162808We must understand and maintain the decorum of the forum .My apology too ,to repeat your words .I don’t claim to be an expert but positively gone depth in the subject .Except Sumant , none ,I repeat none ,knew or forgot the basic difference between sigma level and sigma .After Sumant’s answer , I may be able to retrieve the mathematical explanation ,which I may post tomorrow .I repeat ,only Sumant’s explanation, answers the question to the point.

World is round and we may meet someday .Why only coffee & cookies ? We can have a cocktail ,shading off our little unpleasant communications on the subject. No offense taken ,even your last line being offensive , I’m not returning it back …ha..ha“”Anyway, glad to see that you personally think you have understood the answer (at least closer as you said) …..though i still doubt it ;),””

It’s part of the game ,we play, and want to win always ,even if, we know ,we will lose the game. Let’s play win – win .May be , I started the game as a fun only and got entangled by funny and crazy answers .

Regards0October 9, 2007 at 8:22 am #162809Standard deviation is the way the data points are spread across the mean using which the ‘Sigma level of a process is determined’

your statement of Sigma being the std dev is incorrect

0October 9, 2007 at 8:44 am #162810Pass .No comments

Regards0October 9, 2007 at 8:47 am #162811Pass .No comments

Regards0October 9, 2007 at 9:40 am #162813

KambhampatiParticipant@Praveen**Include @Praveen in your post and this person will**

be notified via email.GKP,

Sumant did give a good answer but other answers too were correct explanation. “Half knowledge is dangerous”..and you are trying to portray yourself as an expert based on that half knowledge. The problem lies with yu that u did not understand those answers because u lack the basic statistical knowledge. Instead of accepting your short comings you are hell bent on proving others wrong. Just be humble enough to accept your ignorance. People in this forum are experts in the field and have been practitioners. Crazy Guy gave one on the best answers which was very basic and correct. So don’t be judgemental without doing your home work properly.0October 9, 2007 at 10:32 am #162814Hi GKP,

A process with a random natural variation when there is no assignable cause involved spreads itself to three times its std. deviation on either side of the mean. Now it is an assumption that if your specification level is twice your natural variation spread when the process is in control, then you have ample space to play around (read shift of central tendency or accomodate occational outliers) and at the same time not produce any defects.

This gives a prcess capability (CPk) of 2 and Mr. Mike Harry and Mr. Bill Smith named this condition of the process as Six Sigma.

So having a smaller std. Deviation is better whereas one should aim to achieve a process capability of 2, which is six sigma level.

3.4 DPMO mathematically corresponds to 4.5 sigma, because once you reach six sigma level your process tend to shift 1.5 sigma to either side of the mean in long term.

Thanx,

SSG

0October 9, 2007 at 11:19 am #162816

fake accrington alertParticipant**Include @fake-accrington-alert in your post and this person will**

be notified via email.I believe you have to review the basic statistics to underdtand the concept,otherwise?

0October 9, 2007 at 11:36 am #162818It’s takes much less time to rectify the mistake than to justify it .Anyway ,thanks for participation and please don’t try to cover it by unpleasent words. I’m closing the topic.

Regards0October 11, 2007 at 2:05 pm #162976GKP,

Here is the relation between sigma level and sigma value.

Process Capability (Cp) = (USL-LSL)/(6*SD)

Where, USL = Upper spec limit

LSL = Lower Spec Limit

SD = Standard Deviation.

Lets assume process is centered around mean, in that case,

Sigma level = 3*Cp = 3* (USL-LSL)/(6*SD)

Keeping other things constant, Sigma level is inversely proportional to sigma value.

let me know for any further clarification .0 - AuthorPosts

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