Home › Forums › Old Forums › General › Help: What is a One-Sided Normal Distribution

This topic contains 3 replies, has 4 voices, and was last updated by BTDT 11 years, 11 months ago.

Viewing 4 posts - 1 through 4 (of 4 total)

- AuthorPosts
- January 12, 2006 at 2:36 pm #98109
What is a One-Sided Normal Distribution?

What is a One-Sided Specification?January 12, 2006 at 3:37 pm #98113Hmmm I have never heard of a one sided normal distribution. There are one or two tail tests.. maybe that is what you mean?

One sided specification means that you have a failure limit on the high or low side but not on both.January 13, 2006 at 3:31 pm #98210I guess it could be for a “barrier” type of feature, such as run-out on a shaft, flatness, etc. Since 0 is the best it could be, there is only an “upper” tolerance. This might answer both the distribution and the tolerance question.

January 13, 2006 at 6:30 pm #98231A one sided normal distribution is better referred to as a half-normal

distribution and consists of only the right or left hRickManCoffee:A one sided normal distribution is better referred to as a

half-normal distribution and consists of only the right or left half of the

bell shaped curve. It is observed and used when the parameter of interest has a

natural upper or lower limit. Other cases are when only the absolute value of

the measurement is used. I have seen it used while measuring:- Hole locations deviations from center- Technical efficiency where 100% is the maximum.- Wavelength correlation for matching spectra where

maximum correlation is 1.0- Deviations from standard operating temperature where

previous data showed only positive deviations.- Contact resistance of an electrical bonding process

where the target is zero resistance. In order to calculate a process capability you can either:- Transform the data to a more normal (or at least

symmetric) distribution and use a single specification limit using the normal

distribution.- Assume a different distribution that has a natural limit

on one or both sides. For example, assume a log-normal or binomial.- Calculate the process capability assuming and using only

the half of the distribution that is applicable. I dont know of any software

that will allow this; youll have to do it by hand. Neither assumption

allows for the discussion or quantification of centering. Cheers, BTDT - AuthorPosts

Viewing 4 posts - 1 through 4 (of 4 total)

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

© Copyright iSixSigma 2000-2017. User Agreement. Any reproduction or other use of content without the express written consent of iSixSigma is prohibited. More »