Jump to content

Standard Deviation


---
 Share

Recommended Posts

Below are some results from a program that was run 6 times without moving part. If I equate these standard deviation values to a probe qualification, I understand a lower number means the deviations were less. How do you interpret what is an acceptable standard deviation? I hear some people say "It was within 3 standard deviations." I understand this phrase but is there a relationship to the standard deviations on the report?

Std Dev.JPG

Link to comment
Share on other sites

I'm not really sure what your specific question is Tom. If it is related to GR&R or Process Capability studies, these topics are very well defined in the AIAG MSA manual (Automotive Industry Action Group, Measurement Systems Analysis). I think it is up to version 4. You can probably find a pdf of it with a Google search.
Link to comment
Share on other sites

Standard deviation is a calculation of the "variability" of the data.

Here is a website that breaks down the math for you.

https://www.calculatorsoup.com/calculat ... ulator.php

The term "± 3 standard deviations" would only apply in the context of a Cpk analysis. By itself, standard deviation does not care about the USL/LSL, but Cpk does. Six Sigma or ± 3 standard deviations refers to how many times the standard deviation can fit inside of the USL/LSL - this is typically shown with a histogram (a bell-curve is the most common shown one).
Link to comment
Share on other sites

And so to answer your last question, just looking at the standard deviation without context to the specification limits is meaningless. Because a 0.01mm standard deviation might be significant on a ±0.02mm tolerance, but not so much of a ±2mm tolerance.
Link to comment
Share on other sites

Please sign in to view this quote.

I guess the question was really "What does the standard deviation do for me other than give me a warm, fuzzy feeling when it's really low?"
Link to comment
Share on other sites

Please sign in to view this quote.

By itself, STD.DEV is basically just a warm and fuzzy feeling. I'm preaching to the choir here, but it is just used to calculate other statistics. It's units aren't squared so it makes it easy to think about.

On here, your data is for sure statistically different. That is to say that your data is different by much more than 3 std.devs. On here and using the STD.Dev from the the red part, the green part is about 100 std.devs above red part
Link to comment
Share on other sites

 Share

×
×
  • Create New...