Defects & Errors
This topic contains 9 replies, has 4 voices, and was last updated by Cesar Gomez 2 years, 5 months ago.
- April 17, 2013 at 2:00 am #54368
Can any one explain what is the difference between defects and errors with an example.
Srinath.CApril 17, 2013 at 4:51 am #195098
Srinath: I’m guessing that you’re interested in the more colloquial definitions, in which case, there really isn’t any. Here’s the definition from the iSS dictionary:
Any type of undesired result is a defect.
A failure to meet one of the acceptance criteria of your customers. A defective unit may have one or more defects.
A defect is a failure to conform to requirements (Crosby, Quality Is Free), whether or not those requirements have been articulated or specified.
The non-conformance to intended usage requirement.
Related Term: Defective
Which would apply to a colloquial definition of error as well. However, if you’re interested in the statistical definition of error, here it is –
Error, also called residual error, refers to variation in observations made under identical test conditions, or the amount of variation that can not be attributed to the variables included in the experiment.
(again, from the iSS Dictionary). This is not a “defect” per se, rather a difference from the expected “perfect” result. Only if it becomes too large (outside the spec limits) does it become a defect.
Hope this helps.April 17, 2013 at 4:53 am #195099
that first sentence is supposed to end “there really isn’t any difference.”
@KatieBarry – what’s up with the edit function?April 17, 2013 at 6:58 am #195102
Not all errors result in defects with some control systems like poka yokes. Defects aren’t necessarily a result of an error but potentially the result of poor process design but defects are OFTEN a result of errors.April 17, 2013 at 7:10 am #195103April 17, 2013 at 9:14 am #195105
According to Dr. Taguchi, defects usually arise as a consequence of too much, or too little work (I’ve substituted work for his ‘energy’ since work and energy share the same physical dimensions.)
If you agree with Dr. Taguchi’s statement, then all we have to ask is how statistical error, or statistical noise, influences work. If you recall, when regressing, we usually deduce a least squares fit of the set of X observations on to equivalent Y observations, and any deviation from the least squares occur as a consequence of error in Y direction. In other words, it is usual to assume each X value is fixed. Therefore, there is a potential link between statistical error and a defect, depending on the amount of process latitude available in the process. I’m not sure how this is dealt with in Six Sigma, but Dr. Taguchi handles this situation using tolerance designs – to determine the level of statistical error in a process at which 50% of the distribution become defective. (I know tolerance designs can be difficult to implement using mechanical systems, but there are many other examples where is it reasonably straight forward.
PS: I have no connection with Taguchi Methods, neither do I wish to promote his methods.April 18, 2013 at 3:15 am #195112
HI Srinath, In a simpler way, defect is an outcome of error/errors. Usually when you are talking about conformity of requirements in a product, the best way to term non conformity is as defect. In general we don’t call a product as erroneous but as defective (There can be exceptions). Errors can happen in the process of making a product. occurrence of an error may not always result in defect but defect is surely a result of error/errors.April 18, 2013 at 10:23 am #195115
The notion that a value just inside a tolerance limit is functional and a value just outside a tolerance limit is defective seems completely unrealistic to my mind. Moreover, in many manufacturing lines, in-line tolerances are not contractual, and in many cases are specified too tight to flag deteriorating equipment performance.
Now many people will not believe me, but one of the first articles about early 6 sigma described a situation where tolerances were widened to five times the measurement uncertainty because of the prevalent design engineering practice of setting unrealistic tolerances to provide an excuse for first silicon bombs.
Please don’t take my criticism personally, but I firmly believe unless professional quality engineers address these problems, we will continue to see a decline in quality, not continuous improvement. One only has to examine most products and services around us to realize most of us would be better off retrieving old products out of the garage and throwing the new shiny ones away.
Just my opinion …April 11, 2016 at 6:34 am #199599
A very simple answer from me:
error = process
defect = product
It’s important to make the distinction, but also to understand the relationship, as it helps with root cause analysis and corrective action.June 30, 2016 at 11:16 am #199840
This was a great response thank you!
You must be logged in to reply to this topic.