ANOVA / X Bar R Results are different - Why is the difference is so great?

M

markzanna

Hi, I have recently completed a measurement study and analysed via XBar/R and ANOVA, the % Study Variation is 40.58 for ANOVA and 21.29 for XBar/R Method. Any idea why the difference is so great?? Thanks
 

Miner

Forum Moderator
Leader
Admin
Re: ANOVA / X Bar R Results are different - why??

This is highly unusual. I have some suspicions, but would need to see the data to be certain. There is a possibility that your gage resolution is not adequate. Chunky data could exaggerate the difference in the two methods.

You will not be able to attach the data until you have made 5 posts. Can you copy and paste the data into a post in an understandable format?
 

Marc

Fully vaccinated are you?
Leader
Re: ANOVA / X Bar R Results are different - why??

You will not be able to attach the data until you have made 5 posts.
:topic: The software is set to allow uploads from any registered visitor. There is no minimum post number set for uploading files as post attachments.
 
M

markzanna

Hi Miner,

Thanks for ther response, I have attached the original data for your review
 

Attachments

  • MSA Data.xls
    17.5 KB · Views: 414

Miner

Forum Moderator
Leader
Admin
It is as I suspected. Despite the fact that your gage can read to 0.001 decimal places, all but a few of the ranges are zero. This can result in very erratic results, particularly for the Range method which depends on the average range.

What are your tolerances, and do you use this gage for SPC or statistical studies?
 
M

markzanna

Hi Miner,

The total tolerance is 0.022mm. With regards SPC we are looking to introduce statistical control into this process but need to resolve this first! Thanks for your help.

Mark
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
For any variable SPC chart, you should have an ndc of 10 (calculated with the control limits, not the specification.) That is, you need to be able to discern 10 statistically distinct categories with the control limits (and even that is a little weak.) Otherwise, you have insignificant resolution to properly detect change.
 

Miner

Forum Moderator
Leader
Admin
Hi Miner,

The total tolerance is 0.022mm. With regards SPC we are looking to introduce statistical control into this process but need to resolve this first! Thanks for your help.

Mark
Mark,

Sorry for not getting back to you sooner.

Several observations:

  1. Your gage is acceptable for inspection %Tol = 10.58%
  2. Your gage is questionable for SPC because more than 25% of the ranges = 0.
  3. Per Wheeler's Honest Gauge Study methods, your gage should be able to measure to 0.0002 instead of the 0.001 to effectively measure this product.
 

riosimbolon

Starting to get Involved
Hi, I have recently completed a measurement study and analysed via XBar/R and ANOVA, the % Study Variation is 40.58 for ANOVA and 21.29 for XBar/R Method. Any idea why the difference is so great?? Thanks
these are the value of the range part in sequence:
Part-1 0
Part-2 0,198
Part-3 0,199
Part-4 0,173
Part-5 0,001
Part-6 0,002
Part-7 0,001
Part-8 0,172
Part-9 0,144
Part-10 0,003

range method depends that range but Anova depends on individual data measured
 
Top Bottom